From e5a2a1eb16d9afd4480e422c4275de7afdfa56dc Mon Sep 17 00:00:00 2001 From: Aleksei Sviridkin Date: Thu, 2 Jul 2026 02:28:16 +0300 Subject: [PATCH 1/5] feat(device_plugin): discover vGPU VFs via vendor-specific VFIO framework Ada/Hopper+ GPUs (vGPU 17+) no longer expose vGPU instances through mdev; instead, SR-IOV Virtual Functions stay bound to the "nvidia" host driver and are configured directly on the PCI function via sysfs. The existing vGPU discovery only walks /sys/bus/mdev/devices, so it finds nothing on these GPUs and every vGPU VF ends up indistinguishable from a whole-card passthrough device, since both share the same PCI vendor:device id. Add a second discovery path that walks /sys/bus/pci/devices looking for Nvidia functions bound to the "nvidia" driver, skips Physical Functions (identified by sriov_totalvfs), and treats a Virtual Function as a configured vGPU when its current_vgpu_type is non-zero. The profile name is resolved by matching that type id against the creatable_vgpu_types catalog, merged per physical card (scoped by each VF's physfn parent) since a configured VF's own list is frequently empty or reduced to the active entry, and since vGPU type ids are only unique within a single physical card. Because these VFs are ordinary PCI/IOMMU-group devices, discovered functions are added to the same iommu/BDF maps used by GPU passthrough and dispatched through the existing GenericDevicePlugin, grouped by vGPU profile name instead of PCI device id. No changes were needed to the allocate path or device contract; the mdev-backed vGPU discovery is untouched. Document the new hardware path in the README alongside the existing mdev instructions, since it applies to a different GPU generation and neither replaces the other. Signed-off-by: Aleksei Sviridkin --- README.md | 31 +++ pkg/device_plugin/device_plugin.go | 33 ++- pkg/device_plugin/vfio_vgpu.go | 257 +++++++++++++++++ pkg/device_plugin/vfio_vgpu_test.go | 414 ++++++++++++++++++++++++++++ 4 files changed, 734 insertions(+), 1 deletion(-) create mode 100644 pkg/device_plugin/vfio_vgpu.go create mode 100644 pkg/device_plugin/vfio_vgpu_test.go diff --git a/README.md b/README.md index 68c68d58..f5f49cbf 100644 --- a/README.md +++ b/README.md @@ -171,6 +171,37 @@ total 0 lrwxrwxrwx. 1 root root 0 Nov 24 13:33 aa618089-8b16-4d01-a136-25a0f3c73123 -> ../../../devices/pci0000:00/0000:00:03.0/0000:03:00.0/0000:04:09.0/0000:06:00.0/aa618089-8b16-4d01-a136-25a0f3c73123 ``` +-------------------------------------------------------------- +### Preparing a GPU to be used in vGPU mode (Ada Lovelace / Hopper and newer) + +Starting with vGPU release 17, GPUs based on Ada Lovelace and newer architectures (for example L40S, H100, H200) no longer expose vGPU instances through mdev. `/sys/bus/mdev` does not exist on these hosts. Instead, vGPU profiles are assigned directly on each SR-IOV Virtual Function through a vendor-specific VFIO sysfs interface, and this plugin discovers vGPUs on such hosts through that interface instead of mdev. The steps below replace the mdev steps above; do not use both on the same GPU. + +##### 1. Enable SR-IOV Virtual Functions on the physical GPU. +```shell +$ /usr/lib/nvidia/sriov-manage -e 0000:41:00.0 +``` +**0000:41:00.0** -- The PCI BDF of the physical GPU (Physical Function). + +##### 2. Find the vGPU types that can be created on a Virtual Function. +```shell +$ cat /sys/bus/pci/devices/0000\:41\:00.4/nvidia/creatable_vgpu_types +1428 : NVIDIA H200X-141C +1414 : NVIDIA H200X-1-18C +``` +Each line lists a numeric type id and the corresponding profile name. + +##### 3. Create a vGPU on the Virtual Function by writing its type id. +```shell +$ echo 1414 > /sys/bus/pci/devices/0000\:41\:00.4/nvidia/current_vgpu_type +``` + +##### 4. Confirm the vGPU was created. +```shell +$ cat /sys/bus/pci/devices/0000\:41\:00.4/nvidia/current_vgpu_type +1414 +``` +A non-zero value confirms the Virtual Function now has a vGPU profile configured; this plugin advertises it as a `nvidia.com/` resource, grouped separately from Virtual Functions configured with a different profile. + ## Docs ### Deployment The Daemonset creation yaml can be used to deploy the device plugin. diff --git a/pkg/device_plugin/device_plugin.go b/pkg/device_plugin/device_plugin.go index 418fb4f4..23b1ac2f 100644 --- a/pkg/device_plugin/device_plugin.go +++ b/pkg/device_plugin/device_plugin.go @@ -89,8 +89,11 @@ var stop = make(chan struct{}) func InitiateDevicePlugin() { //Identifies GPUs and represents it in appropriate structures createIommuDeviceMap() - //Identifies vGPUs and represents it in appropriate structures + //Identifies mdev-backed vGPUs and represents it in appropriate structures createVgpuIDMap() + //Identifies vGPUs exposed through the vendor-specific VFIO framework + //(Ada/Hopper+ GPUs, which do not support mdev) + createVfioVGpuMap() //Creates and starts device plugin createDevicePlugins() } @@ -104,6 +107,7 @@ func createDevicePlugins() { log.Printf("Device Map %v", deviceMap) log.Println("vGPU Map ", vGpuMap) log.Println("GPU vGPU Map ", gpuVgpuMap) + log.Println("VFIO vGPU Map ", vfioVGpuMap) //Iterate over deivceMap to create device plugin for each type of GPU on the host for k, gpuDevices := range deviceMap { @@ -162,6 +166,33 @@ func createDevicePlugins() { vGpuDevicePlugins = append(vGpuDevicePlugins, dp) } } + //Iterate over vfioVGpuMap to create a device plugin for each vGPU profile + //exposed through the vendor-specific VFIO framework. These devices are + //grouped by profile name (unlike deviceMap, which groups by raw PCI + //device id) but otherwise use the same PCI-passthrough-style contract as + //deviceMap, since vendor-VFIO VFs are ordinary PCI/IOMMU-group devices. + for k, gpuDevices := range vfioVGpuMap { + devs = nil + for _, gpuDev := range gpuDevices { + devs = append(devs, &pluginapi.Device{ + ID: gpuDev.addr, + Health: pluginapi.Healthy, + Topology: &pluginapi.TopologyInfo{ + Nodes: []*pluginapi.NUMANode{ + {ID: gpuDev.numaNode}, + }, + }, + }) + } + log.Printf("DP Name %s", k) + dp := NewGenericDevicePlugin(k, "/dev/vfio/", devs) + err := startDevicePlugin(dp) + if err != nil { + log.Printf("Error starting %s device plugin: %v", dp.deviceName, err) + } else { + devicePlugins = append(devicePlugins, dp) + } + } <-stop log.Printf("Shutting down device plugin controller") diff --git a/pkg/device_plugin/vfio_vgpu.go b/pkg/device_plugin/vfio_vgpu.go new file mode 100644 index 00000000..6537b18e --- /dev/null +++ b/pkg/device_plugin/vfio_vgpu.go @@ -0,0 +1,257 @@ +/* + * Copyright (c) 2026, NVIDIA CORPORATION. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * * Neither the name of NVIDIA CORPORATION nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY + * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, + * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, + * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR + * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY + * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +// This file discovers Nvidia vGPUs exposed through the vendor-specific VFIO +// framework used by Ada/Hopper+ GPUs (vGPU 17+) instead of mdev. Unlike +// mdev-backed vGPUs, these are SR-IOV Virtual Functions that stay bound to +// the "nvidia" host driver and are configured directly on the PCI function +// via sysfs, rather than through /sys/bus/mdev. +package device_plugin + +import ( + "log" + "os" + "path/filepath" + "regexp" + "strings" +) + +const ( + // nvidiaDriverName is the host driver that vGPU-capable VFs stay bound to + // when using the vendor-specific VFIO framework (as opposed to vfio-pci + // used for classic GPU passthrough) + nvidiaDriverName = "nvidia" + // vfioVGpuSysfsDir is the vendor-specific sysfs directory exposed under + // each PCI function capable of vGPU (both the Physical Function and its + // Virtual Functions) + vfioVGpuSysfsDir = "nvidia" + // currentVgpuTypeFile contains the numeric vGPU type id configured on a + // VF, or "0" if no type has been configured yet + currentVgpuTypeFile = "current_vgpu_type" + // creatableVgpuTypesFile lists the vGPU types that can be created on a + // function, one per line, formatted as " : " + creatableVgpuTypesFile = "creatable_vgpu_types" + // sriovTotalVfsFile is only present on SR-IOV Physical Functions + sriovTotalVfsFile = "sriov_totalvfs" + // physfnLink is the symlink present on every Virtual Function pointing + // back at its parent Physical Function + physfnLink = "physfn" + // unconfiguredVgpuType is the current_vgpu_type value of a VF that has no + // vGPU type configured + unconfiguredVgpuType = "0" +) + +// Matches creatable_vgpu_types entries such as "1428 : NVIDIA H200X-141C" +var creatableVgpuTypeRegexp = regexp.MustCompile(`^\s*([0-9]+)\s*:\s*(.+?)\s*$`) +var whitespaceRegexp = regexp.MustCompile(`\s+`) + +// Key is the vGPU profile name (sanitized the same way as the mdev vGPU type +// name) and value is the list of Nvidia vGPU VFs of that profile, discovered +// through the vendor-specific VFIO framework +var vfioVGpuMap map[string][]NvidiaGpuDevice + +var readVfioVgpuFile = readVfioVgpuFileFunc +var isPhysicalFunction = isPhysicalFunctionFunc + +// Discovers all Nvidia vGPUs exposed through the vendor-specific VFIO +// framework and creates the corresponding maps. These vGPUs are SR-IOV VFs +// that stay bound to the "nvidia" driver and expose their vGPU state under +// /sys/bus/pci/devices/
/nvidia/, unlike mdev-backed vGPUs which are +// discovered under /sys/bus/mdev/devices by createVgpuIDMap. +// +// A VF is considered a configured vGPU when current_vgpu_type is set to a +// non-zero value. Its profile name is resolved by matching that type id +// against the creatable_vgpu_types catalog. On a configured VF this catalog +// is frequently empty or reduced to the active type only, so this function +// merges the catalog from every scanned function of the same physical card +// (the card's Physical Function and every one of its Virtual Functions, +// configured or not) before resolving names, converging as long as at least +// one function of that card still reports its full list - typically true +// because a freshly enumerated PF or an unconfigured sibling VF always does. +// The catalog is scoped per physical card (keyed by the Physical Function's +// PCI address, found via each VF's "physfn" symlink) rather than merged +// globally across the whole host, because vGPU type ids are only unique +// within one physical card - the same numeric id can mean a different +// profile on a different card, even of the same GPU model. The corner case +// of every function of a card being simultaneously configured with a +// reduced catalog that omits a given type id is not handled here; if this +// is observed in practice, a cached per-card catalog (refreshed once at +// startup) or an NVML-based lookup would be the natural fallback. +// +// Discovered VFs are added to the shared iommuMap and bdfToIommuMap used by +// GenericDevicePlugin, so Allocate and the health check for these devices +// reuse the exact same PCI/IOMMU-group contract as classic GPU passthrough - +// no vendor-VFIO-specific Allocate logic is required. +func createVfioVGpuMap() { + vfioVGpuMap = make(map[string][]NvidiaGpuDevice) + // iommuMap and bdfToIommuMap are shared with the GPU passthrough + // discovery (createIommuDeviceMap) and consumed by GenericDevicePlugin. + // Initialize them here too so this discovery also works standalone, e.g. + // in tests that call it directly. + if iommuMap == nil { + iommuMap = make(map[string][]NvidiaGpuDevice) + } + if bdfToIommuMap == nil { + bdfToIommuMap = make(map[string]string) + } + + type vfioVGpuVF struct { + device NvidiaGpuDevice + typeID string + iommuGroup string + pfAddress string + } + var vfs []vfioVGpuVF + // Keyed by the owning Physical Function's PCI address - see the name + // resolution note in the function-level comment above. + vgpuTypeNamesByPF := make(map[string]map[string]string) + + //Walk directory to discover vGPU-capable PCI functions + walkErr := filepath.Walk(basePath, func(path string, info os.FileInfo, err error) error { + if err != nil { + log.Printf("Error accessing file path %q: %v\n", path, err) + return err + } + if info.IsDir() { + return nil + } + vendorID, err := readIDFromFile(basePath, info.Name(), "vendor") + if err != nil || vendorID != nvidiaVendorID { + return nil + } + driver, err := readLink(basePath, info.Name(), "driver") + if err != nil || driver != nvidiaDriverName { + return nil + } + + isPF := isPhysicalFunction(basePath, info.Name()) + pfAddress := info.Name() + if !isPF { + physfn, err := readLink(basePath, info.Name(), physfnLink) + if err != nil { + log.Printf("Could not determine parent Physical Function for VF %s: %v", info.Name(), err) + return nil + } + pfAddress = physfn + } + + // Merge this function's creatable vGPU types into its own card's + // catalog before deciding whether it is itself a configured VF. + if data, err := readVfioVgpuFile(basePath, info.Name(), creatableVgpuTypesFile); err == nil { + if vgpuTypeNamesByPF[pfAddress] == nil { + vgpuTypeNamesByPF[pfAddress] = make(map[string]string) + } + mergeCreatableVgpuTypes(vgpuTypeNamesByPF[pfAddress], data) + } + if isPF { + return nil + } + + typeID, err := readVfioVgpuFile(basePath, info.Name(), currentVgpuTypeFile) + if err != nil { + // Not a vGPU-capable function, e.g. a GPU used directly by the + // host driver without any vGPU VFs enabled + return nil + } + if typeID == "" || typeID == unconfiguredVgpuType { + log.Printf("Skipping unconfigured vGPU VF %s", info.Name()) + return nil + } + iommuGroup, err := readLink(basePath, info.Name(), "iommu_group") + if err != nil { + log.Println("Could not get IOMMU Group for device ", info.Name()) + return nil + } + numaNode, err := readNUMANode(basePath, info.Name()) + if err != nil { + log.Printf("Could not get NUMA node for device %s: %v. Defaulting to NUMA node 0", info.Name(), err) + numaNode = 0 + } + vfs = append(vfs, vfioVGpuVF{ + device: NvidiaGpuDevice{addr: info.Name(), numaNode: numaNode}, + typeID: typeID, + iommuGroup: iommuGroup, + pfAddress: pfAddress, + }) + return nil + }) + if walkErr != nil { + log.Printf("Error discovering vendor-specific VFIO vGPU VFs: %v", walkErr) + } + + for _, vf := range vfs { + vGpuName, ok := vgpuTypeNamesByPF[vf.pfAddress][vf.typeID] + if !ok { + log.Printf("Error: could not resolve the name of vGPU type %s configured on VF %s (parent PF %s), skipping device", vf.typeID, vf.device.addr, vf.pfAddress) + continue + } + vfioVGpuMap[vGpuName] = append(vfioVGpuMap[vGpuName], vf.device) + iommuMap[vf.iommuGroup] = append(iommuMap[vf.iommuGroup], vf.device) + bdfToIommuMap[vf.device.addr] = vf.iommuGroup + } +} + +// mergeCreatableVgpuTypes parses creatable_vgpu_types content and merges its +// " : " entries into vgpuTypeNames. Lines that do not match the +// format are ignored. vGPU type names are sanitized the same way as mdev type +// names (readVgpuIDFromFileFunc) so that both vGPU frameworks expose +// identically formatted resource names for a profile of the same name. +func mergeCreatableVgpuTypes(vgpuTypeNames map[string]string, data string) { + for _, line := range strings.Split(data, "\n") { + match := creatableVgpuTypeRegexp.FindStringSubmatch(line) + if match == nil { + continue + } + typeID := match[1] + vGpuName := whitespaceRegexp.ReplaceAllString(match[2], "_") // Replace all spaces with underscore + if existing, ok := vgpuTypeNames[typeID]; ok { + if existing != vGpuName { + log.Printf("Error: conflicting names for vGPU type %s: %s and %s, keeping %s", typeID, existing, vGpuName, existing) + } + continue + } + vgpuTypeNames[typeID] = vGpuName + } +} + +// readVfioVgpuFileFunc reads a file from the vendor-specific sysfs directory +// of a vGPU-capable PCI function. +func readVfioVgpuFileFunc(basePath string, deviceAddress string, fileName string) (string, error) { + data, err := os.ReadFile(filepath.Join(basePath, deviceAddress, vfioVGpuSysfsDir, fileName)) + if err != nil { + return "", err + } + return strings.TrimSpace(string(data)), nil +} + +// isPhysicalFunctionFunc returns true if a device is an SR-IOV Physical +// Function. Only Physical Functions expose the sriov_totalvfs attribute. +func isPhysicalFunctionFunc(basePath string, deviceAddress string) bool { + _, err := os.Stat(filepath.Join(basePath, deviceAddress, sriovTotalVfsFile)) + return err == nil +} diff --git a/pkg/device_plugin/vfio_vgpu_test.go b/pkg/device_plugin/vfio_vgpu_test.go new file mode 100644 index 00000000..a51e81a6 --- /dev/null +++ b/pkg/device_plugin/vfio_vgpu_test.go @@ -0,0 +1,414 @@ +/* + * Copyright (c) 2026, NVIDIA CORPORATION. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * * Neither the name of NVIDIA CORPORATION nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY + * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, + * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, + * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR + * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY + * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +package device_plugin + +import ( + "os" + "path/filepath" + "time" + + . "github.com/onsi/ginkgo/v2" + . "github.com/onsi/gomega" +) + +var vfioPfAddress = "0000:41:00.0" +var vfioVfAddress1 = "0000:41:00.4" +var vfioVfAddress2 = "0000:41:00.5" +var vfioVfAddress3 = "0000:41:00.6" +var vfioVfAddress4 = "0000:41:00.7" +var vfioPfAddress2 = "0000:81:00.0" +var vfioVfAddress5 = "0000:81:00.4" +var vfioVgpuTypeName1 = "NVIDIA H200X-141C" +var vfioVgpuTypeName2 = "NVIDIA H200X-1-18C" +var vfioVgpuResourceName1 = "NVIDIA_H200X-141C" +var vfioVgpuResourceName2 = "NVIDIA_H200X-1-18C" + +var _ = Describe("Vendor-specific VFIO vGPU", func() { + var workDir string + var linkDir string + var err error + var origBasePath string + + type fakeDeviceOptions struct { + vendor string + driver string + iommuGroup string + numaNode string + physicalFn bool + physfn string // parent Physical Function address, for a VF + currentType string + creatableTypes string + } + + createFakeDevice := func(address string, options fakeDeviceOptions) { + deviceDir := filepath.Join(linkDir, address) + Expect(os.MkdirAll(deviceDir, 0755)).To(Succeed()) + Expect(os.WriteFile(filepath.Join(deviceDir, "vendor"), []byte(options.vendor+"\n"), 0644)).To(Succeed()) + Expect(os.Symlink(filepath.Join(linkDir, "drivers", options.driver), filepath.Join(deviceDir, "driver"))).To(Succeed()) + if options.iommuGroup != "" { + Expect(os.Symlink(filepath.Join(linkDir, "iommu_groups", options.iommuGroup), filepath.Join(deviceDir, "iommu_group"))).To(Succeed()) + } + if options.numaNode != "" { + Expect(os.WriteFile(filepath.Join(deviceDir, "numa_node"), []byte(options.numaNode+"\n"), 0644)).To(Succeed()) + } + if options.physicalFn { + Expect(os.WriteFile(filepath.Join(deviceDir, "sriov_totalvfs"), []byte("32\n"), 0644)).To(Succeed()) + } + if options.physfn != "" { + Expect(os.Symlink(filepath.Join(workDir, options.physfn), filepath.Join(deviceDir, "physfn"))).To(Succeed()) + } + if options.currentType != "" || options.creatableTypes != "" { + Expect(os.MkdirAll(filepath.Join(deviceDir, "nvidia"), 0755)).To(Succeed()) + } + if options.currentType != "" { + Expect(os.WriteFile(filepath.Join(deviceDir, "nvidia", "current_vgpu_type"), []byte(options.currentType+"\n"), 0644)).To(Succeed()) + } + if options.creatableTypes != "" { + Expect(os.WriteFile(filepath.Join(deviceDir, "nvidia", "creatable_vgpu_types"), []byte(options.creatableTypes), 0644)).To(Succeed()) + } + Expect(os.Symlink(deviceDir, filepath.Join(workDir, address))).To(Succeed()) + } + + BeforeEach(func() { + readLink = readLinkFunc + readIDFromFile = readIDFromFileFunc + readNUMANode = readNUMANodeFunc + readVfioVgpuFile = readVfioVgpuFileFunc + isPhysicalFunction = isPhysicalFunctionFunc + + linkDir, err = os.MkdirTemp("", "vfio-vgpu-link") + Expect(err).ToNot(HaveOccurred()) + workDir, err = os.MkdirTemp("", "vfio-vgpu-test") + Expect(err).ToNot(HaveOccurred()) + origBasePath = basePath + basePath = workDir + }) + + AfterEach(func() { + basePath = origBasePath + _ = os.RemoveAll(workDir) + _ = os.RemoveAll(linkDir) + }) + + Context("mergeCreatableVgpuTypes() Tests", func() { + It("Parses id and name pairs and sanitizes names", func() { + vgpuTypeNames := make(map[string]string) + mergeCreatableVgpuTypes(vgpuTypeNames, "1428 : NVIDIA H200X-141C\n1414 : NVIDIA H200X-1-18C\n") + Expect(vgpuTypeNames).To(HaveLen(2)) + Expect(vgpuTypeNames["1428"]).To(Equal(vfioVgpuResourceName1)) + Expect(vgpuTypeNames["1414"]).To(Equal(vfioVgpuResourceName2)) + }) + + It("Ignores lines that do not match the id and name format", func() { + vgpuTypeNames := make(map[string]string) + mergeCreatableVgpuTypes(vgpuTypeNames, "ID : vGPU Name\n\ngarbage\n1428 : NVIDIA H200X-141C\n") + Expect(vgpuTypeNames).To(HaveLen(1)) + Expect(vgpuTypeNames["1428"]).To(Equal(vfioVgpuResourceName1)) + }) + + It("Keeps the first name when merged entries conflict", func() { + vgpuTypeNames := make(map[string]string) + mergeCreatableVgpuTypes(vgpuTypeNames, "1428 : NVIDIA H200X-141C\n") + mergeCreatableVgpuTypes(vgpuTypeNames, "1428 : NVIDIA CONFLICTING-NAME\n") + Expect(vgpuTypeNames).To(HaveLen(1)) + Expect(vgpuTypeNames["1428"]).To(Equal(vfioVgpuResourceName1)) + }) + + It("Handles empty content", func() { + vgpuTypeNames := make(map[string]string) + mergeCreatableVgpuTypes(vgpuTypeNames, "") + Expect(vgpuTypeNames).To(BeEmpty()) + }) + }) + + Context("readVfioVgpuFileFunc() Tests", func() { + It("Reads a trimmed value without error", func() { + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + currentType: "1428", + }) + value, err := readVfioVgpuFileFunc(basePath, vfioVfAddress1, "current_vgpu_type") + Expect(err).To(BeNil()) + Expect(value).To(Equal("1428")) + }) + + It("Returns an error for a missing file", func() { + value, err := readVfioVgpuFileFunc(basePath, vfioVfAddress1, "current_vgpu_type") + Expect(err).ShouldNot(BeNil()) + Expect(value).To(Equal("")) + }) + }) + + Context("isPhysicalFunctionFunc() Tests", func() { + It("Identifies a Physical Function by sriov_totalvfs", func() { + createFakeDevice(vfioPfAddress, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physicalFn: true, + }) + Expect(isPhysicalFunctionFunc(basePath, vfioPfAddress)).To(BeTrue()) + }) + + It("Does not identify a Virtual Function as a Physical Function", func() { + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + }) + Expect(isPhysicalFunctionFunc(basePath, vfioVfAddress1)).To(BeFalse()) + }) + }) + + Context("createVfioVGpuMap() Tests", func() { + BeforeEach(func() { + iommuMap = nil + bdfToIommuMap = nil + }) + + It("Discovers configured VFs and groups them by vGPU type name", func() { + createFakeDevice(vfioPfAddress, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physicalFn: true, + creatableTypes: "1428 : " + vfioVgpuTypeName1 + "\n1414 : " + vfioVgpuTypeName2 + "\n", + }) + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + }) + createFakeDevice(vfioVfAddress2, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "72", + numaNode: "1", + currentType: "1414", + creatableTypes: "1414 : " + vfioVgpuTypeName2 + "\n", + }) + // Unconfigured VF + createFakeDevice(vfioVfAddress3, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "73", + numaNode: "0", + currentType: "0", + }) + // VF configured with a type that no creatable list resolves + createFakeDevice(vfioVfAddress4, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "74", + numaNode: "0", + currentType: "9999", + }) + // GPU passthrough device handled by the vfio-pci discovery + createFakeDevice("0000:01:00.0", fakeDeviceOptions{ + vendor: "0x10de", + driver: "vfio-pci", + iommuGroup: "10", + }) + // Device from another vendor + createFakeDevice("0000:02:00.0", fakeDeviceOptions{ + vendor: "0x8086", + driver: "nvidia", + }) + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(HaveLen(2)) + Expect(vfioVGpuMap[vfioVgpuResourceName1]).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1][0].addr).To(Equal(vfioVfAddress1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1][0].numaNode).To(Equal(int64(0))) + Expect(vfioVGpuMap[vfioVgpuResourceName2]).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName2][0].addr).To(Equal(vfioVfAddress2)) + Expect(vfioVGpuMap[vfioVgpuResourceName2][0].numaNode).To(Equal(int64(1))) + + Expect(bdfToIommuMap[vfioVfAddress1]).To(Equal("71")) + Expect(bdfToIommuMap[vfioVfAddress2]).To(Equal("72")) + Expect(iommuMap["71"][0].addr).To(Equal(vfioVfAddress1)) + Expect(iommuMap["72"][0].addr).To(Equal(vfioVfAddress2)) + + Expect(bdfToIommuMap).ToNot(HaveKey(vfioPfAddress)) + Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress3)) + Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress4)) + }) + + It("Resolves a type name from a sibling unconfigured VF", func() { + // Configured VF with an empty creatable list + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + }) + // Sibling unconfigured VF still lists the creatable types + createFakeDevice(vfioVfAddress2, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "72", + numaNode: "0", + currentType: "0", + creatableTypes: "1428 : " + vfioVgpuTypeName1 + "\n", + }) + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1][0].addr).To(Equal(vfioVfAddress1)) + }) + + It("Skips configured VFs when no creatable list resolves their type", func() { + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + }) + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(BeEmpty()) + }) + + It("Skips a VF whose parent Physical Function cannot be determined", func() { + // VF with no physfn symlink at all (readLink fails) + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + creatableTypes: "1428 : " + vfioVgpuTypeName1 + "\n", + }) + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(BeEmpty()) + Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress1)) + }) + + It("Scopes vGPU type ids per physical card so colliding ids on different cards resolve to distinct profiles", func() { + // Card 1: PF + one configured VF using type id 50 for profile 1 + createFakeDevice(vfioPfAddress, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physicalFn: true, + creatableTypes: "50 : " + vfioVgpuTypeName1 + "\n", + }) + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "50", + }) + // Card 2: a different physical card that happens to reuse type id + // 50 for a completely different profile + createFakeDevice(vfioPfAddress2, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physicalFn: true, + creatableTypes: "50 : " + vfioVgpuTypeName2 + "\n", + }) + createFakeDevice(vfioVfAddress5, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress2, + iommuGroup: "75", + numaNode: "0", + currentType: "50", + }) + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(HaveLen(2)) + Expect(vfioVGpuMap[vfioVgpuResourceName1]).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1][0].addr).To(Equal(vfioVfAddress1)) + Expect(vfioVGpuMap[vfioVgpuResourceName2]).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName2][0].addr).To(Equal(vfioVfAddress5)) + }) + }) + + Context("createDevicePlugins() vfio vGPU Tests", func() { + var startedPlugins []string + + BeforeEach(func() { + startedPlugins = nil + startDevicePlugin = func(dp *GenericDevicePlugin) error { + startedPlugins = append(startedPlugins, dp.deviceName) + return nil + } + startVgpuDevicePlugin = func(dp *GenericVGpuDevicePlugin) error { + return nil + } + + deviceMap = make(map[string][]NvidiaGpuDevice) + iommuMap = make(map[string][]NvidiaGpuDevice) + bdfToIommuMap = make(map[string]string) + gpuVgpuMap = make(map[string][]string) + vGpuMap = make(map[string][]NvidiaGpuDevice) + vfioVGpuMap = map[string][]NvidiaGpuDevice{ + vfioVgpuResourceName1: {{addr: vfioVfAddress1, numaNode: 0}}, + vfioVgpuResourceName2: {{addr: vfioVfAddress2, numaNode: 0}}, + } + }) + + AfterEach(func() { + startDevicePlugin = startDevicePluginFunc + startVgpuDevicePlugin = startVgpuDevicePluginFunc + deviceMap = nil + iommuMap = nil + bdfToIommuMap = nil + gpuVgpuMap = nil + vGpuMap = nil + vfioVGpuMap = nil + }) + + It("Starts a PCI-style device plugin per discovered vGPU profile", func() { + go createDevicePlugins() + time.Sleep(1 * time.Second) + stop <- struct{}{} + + Expect(startedPlugins).To(ConsistOf(vfioVgpuResourceName1, vfioVgpuResourceName2)) + }) + }) +}) From c13643e3e2d7d67201f15246a1d1875cec65eccf Mon Sep 17 00:00:00 2001 From: Aleksei Sviridkin Date: Thu, 2 Jul 2026 04:24:02 +0300 Subject: [PATCH 2/5] fix(device_plugin): resolve vGPU types on fully consumed cards A fully consumed card reduces the creatable_vgpu_types list of every one of its functions down to the header, so the per-card catalog cannot resolve the types of already-configured VFs and the plugin advertised nothing exactly when all capacity was allocated. Fall back to a host-wide catalog merged across all cards; numeric ids that different cards map to different names stay ambiguous and are still skipped rather than guessed. Signed-off-by: Aleksei Sviridkin --- pkg/device_plugin/vfio_vgpu.go | 29 +++++++++++- pkg/device_plugin/vfio_vgpu_test.go | 69 +++++++++++++++++++++++++++++ 2 files changed, 96 insertions(+), 2 deletions(-) diff --git a/pkg/device_plugin/vfio_vgpu.go b/pkg/device_plugin/vfio_vgpu.go index 6537b18e..489d6c5f 100644 --- a/pkg/device_plugin/vfio_vgpu.go +++ b/pkg/device_plugin/vfio_vgpu.go @@ -204,11 +204,36 @@ func createVfioVGpuMap() { log.Printf("Error discovering vendor-specific VFIO vGPU VFs: %v", walkErr) } + // Host-wide catalog fallback: a fully consumed card reduces the + // creatable_vgpu_types list of EVERY one of its functions (down to the + // header), so the card's own catalog cannot resolve the types of its + // already-configured VFs. Merge all per-card catalogs into a host-wide + // one, tracking numeric ids that different cards resolve to different + // names — those stay ambiguous and are never guessed. + globalTypeNames := make(map[string]string) + ambiguousTypeIDs := make(map[string]bool) + for _, catalog := range vgpuTypeNamesByPF { + for typeID, name := range catalog { + if existing, ok := globalTypeNames[typeID]; ok { + if existing != name { + ambiguousTypeIDs[typeID] = true + } + continue + } + globalTypeNames[typeID] = name + } + } + for _, vf := range vfs { vGpuName, ok := vgpuTypeNamesByPF[vf.pfAddress][vf.typeID] if !ok { - log.Printf("Error: could not resolve the name of vGPU type %s configured on VF %s (parent PF %s), skipping device", vf.typeID, vf.device.addr, vf.pfAddress) - continue + if name, found := globalTypeNames[vf.typeID]; found && !ambiguousTypeIDs[vf.typeID] { + log.Printf("Resolved vGPU type %s on VF %s via the host-wide catalog (the card's own catalog is reduced)", vf.typeID, vf.device.addr) + vGpuName = name + } else { + log.Printf("Error: could not resolve the name of vGPU type %s configured on VF %s (parent PF %s), skipping device", vf.typeID, vf.device.addr, vf.pfAddress) + continue + } } vfioVGpuMap[vGpuName] = append(vfioVGpuMap[vGpuName], vf.device) iommuMap[vf.iommuGroup] = append(iommuMap[vf.iommuGroup], vf.device) diff --git a/pkg/device_plugin/vfio_vgpu_test.go b/pkg/device_plugin/vfio_vgpu_test.go index a51e81a6..f086b0dc 100644 --- a/pkg/device_plugin/vfio_vgpu_test.go +++ b/pkg/device_plugin/vfio_vgpu_test.go @@ -266,6 +266,75 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress4)) }) + It("Resolves a type via the host-wide catalog when the card's own catalog is reduced", func() { + // Fully consumed card: every function's creatable list is reduced + // to the header, so the card's own catalog cannot resolve the + // configured type. + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + creatableTypes: "ID : vGPU Name\n", + }) + // A free function on ANOTHER card still lists the full catalog. + createFakeDevice(vfioVfAddress5, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress2, + iommuGroup: "81", + numaNode: "0", + currentType: "0", + creatableTypes: "1428 : " + vfioVgpuTypeName1 + "\n", + }) + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1]).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1][0].addr).To(Equal(vfioVfAddress1)) + }) + + It("Skips a VF whose type maps to conflicting names across cards", func() { + // Fully consumed card with a reduced catalog. + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + creatableTypes: "ID : vGPU Name\n", + }) + // Two other cards resolve the same numeric id to DIFFERENT names — + // the host-wide fallback must refuse to guess. + createFakeDevice(vfioVfAddress5, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress2, + iommuGroup: "81", + numaNode: "0", + currentType: "0", + creatableTypes: "1428 : " + vfioVgpuTypeName1 + "\n", + }) + createFakeDevice("0000:c1:00.4", fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: "0000:c1:00.0", + iommuGroup: "91", + numaNode: "0", + currentType: "0", + creatableTypes: "1428 : NVIDIA OTHER-NAME\n", + }) + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(BeEmpty()) + Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress1)) + }) + It("Resolves a type name from a sibling unconfigured VF", func() { // Configured VF with an empty creatable list createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ From c3abc4d40850ecadc768eb9f83afcfa0c4402cc2 Mon Sep 17 00:00:00 2001 From: Aleksei Sviridkin Date: Thu, 2 Jul 2026 04:51:36 +0300 Subject: [PATCH 3/5] feat(device_plugin): NVML fallback for vGPU type name resolution MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit On a host where every card is fully consumed, no creatable_vgpu_types catalog lists anything at all — sysfs alone cannot resolve configured type ids to profile names no matter how catalogs are merged. Resolve through NVML as the last resort: the supported-vGPU list of a card does not shrink as capacity is allocated. Scoped per physical card like the sysfs catalogs, cached per scan, and requires the host driver library to be loadable in the container. Signed-off-by: Aleksei Sviridkin --- go.mod | 1 + go.sum | 2 + pkg/device_plugin/vfio_vgpu.go | 19 + pkg/device_plugin/vfio_vgpu_nvml.go | 96 + pkg/device_plugin/vfio_vgpu_test.go | 50 + vendor/github.com/NVIDIA/go-nvml/LICENSE | 202 + vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl.go | 117 + .../NVIDIA/go-nvml/pkg/dl/dl_linux.go | 26 + .../github.com/NVIDIA/go-nvml/pkg/nvml/api.go | 56 + .../NVIDIA/go-nvml/pkg/nvml/cgo_helpers.h | 23 + .../go-nvml/pkg/nvml/cgo_helpers_static.go | 75 + .../NVIDIA/go-nvml/pkg/nvml/const.go | 2104 +++ .../NVIDIA/go-nvml/pkg/nvml/const_static.go | 27 + .../NVIDIA/go-nvml/pkg/nvml/device.go | 3592 ++++ .../github.com/NVIDIA/go-nvml/pkg/nvml/doc.go | 21 + .../go-nvml/pkg/nvml/dynamicLibrary_mock.go | 157 + .../NVIDIA/go-nvml/pkg/nvml/event_set.go | 82 + .../github.com/NVIDIA/go-nvml/pkg/nvml/gpm.go | 170 + .../NVIDIA/go-nvml/pkg/nvml/init.go | 48 + .../github.com/NVIDIA/go-nvml/pkg/nvml/lib.go | 296 + .../NVIDIA/go-nvml/pkg/nvml/nvml.go | 3749 +++++ .../github.com/NVIDIA/go-nvml/pkg/nvml/nvml.h | 13557 ++++++++++++++++ .../NVIDIA/go-nvml/pkg/nvml/refcount.go | 31 + .../NVIDIA/go-nvml/pkg/nvml/return.go | 103 + .../NVIDIA/go-nvml/pkg/nvml/system.go | 148 + .../NVIDIA/go-nvml/pkg/nvml/types_gen.go | 1462 ++ .../NVIDIA/go-nvml/pkg/nvml/unit.go | 113 + .../NVIDIA/go-nvml/pkg/nvml/vgpu.go | 511 + .../go-nvml/pkg/nvml/zz_generated.api.go | 1154 ++ vendor/modules.txt | 4 + 30 files changed, 27996 insertions(+) create mode 100644 pkg/device_plugin/vfio_vgpu_nvml.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/LICENSE create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl_linux.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/api.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers.h create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers_static.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const_static.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/device.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/doc.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/dynamicLibrary_mock.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/event_set.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/gpm.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/init.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/lib.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.h create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/refcount.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/return.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/system.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/types_gen.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/unit.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/vgpu.go create mode 100644 vendor/github.com/NVIDIA/go-nvml/pkg/nvml/zz_generated.api.go diff --git a/go.mod b/go.mod index a09b4a8b..9fb3dcea 100644 --- a/go.mod +++ b/go.mod @@ -3,6 +3,7 @@ module kubevirt-gpu-device-plugin go 1.25.0 require ( + github.com/NVIDIA/go-nvml v0.13.0-1 github.com/NVIDIA/gpu-monitoring-tools v0.0.0-20211102125545-5a2c58442e48 github.com/fsnotify/fsnotify v1.9.0 github.com/onsi/ginkgo/v2 v2.28.1 diff --git a/go.sum b/go.sum index 7db2fb6d..425a810a 100644 --- a/go.sum +++ b/go.sum @@ -1,6 +1,8 @@ github.com/Masterminds/semver v1.5.0/go.mod h1:MB6lktGJrhw8PrUyiEoblNEGEQ+RzHPF078ddwwvV3Y= github.com/Masterminds/semver/v3 v3.4.0 h1:Zog+i5UMtVoCU8oKka5P7i9q9HgrJeGzI9SA1Xbatp0= github.com/Masterminds/semver/v3 v3.4.0/go.mod h1:4V+yj/TJE1HU9XfppCwVMZq3I84lprf4nC11bSS5beM= +github.com/NVIDIA/go-nvml v0.13.0-1 h1:OLX8Jq3dONuPOQPC7rndB6+iDmDakw0XTYgzMxObkEw= +github.com/NVIDIA/go-nvml v0.13.0-1/go.mod h1:+KNA7c7gIBH7SKSJ1ntlwkfN80zdx8ovl4hrK3LmPt4= github.com/NVIDIA/gpu-monitoring-tools v0.0.0-20211102125545-5a2c58442e48 h1:JO/JF5CBte9mvATbhoh32swu9erf07ZdLgwFj8u21UQ= github.com/NVIDIA/gpu-monitoring-tools v0.0.0-20211102125545-5a2c58442e48/go.mod h1:oKPJa5eOTkWvlT4/Y4D8Nds44Fzmww5HUK+xwO+DwTA= github.com/NVIDIA/gpu-monitoring-tools/bindings/go/dcgm v0.0.0-20210325210537-29b4f1784f18/go.mod h1:8qXwltEzU3idjUcVpMOv3FNgxxbDeXZPGMLyc/khWiY= diff --git a/pkg/device_plugin/vfio_vgpu.go b/pkg/device_plugin/vfio_vgpu.go index 489d6c5f..c5a3b109 100644 --- a/pkg/device_plugin/vfio_vgpu.go +++ b/pkg/device_plugin/vfio_vgpu.go @@ -224,12 +224,31 @@ func createVfioVGpuMap() { } } + // NVML is the last-resort resolution source, scoped per card like the + // sysfs catalogs; results are cached per PF for the duration of the scan. + nvmlNamesByPF := make(map[string]map[string]string) + nvmlSupportedNames := func(pfAddress string) map[string]string { + if names, done := nvmlNamesByPF[pfAddress]; done { + return names + } + names, err := resolveVgpuTypeNamesViaNVML(pfAddress) + if err != nil { + log.Printf("Could not resolve supported vGPU types of PF %s via NVML: %v", pfAddress, err) + names = map[string]string{} + } + nvmlNamesByPF[pfAddress] = names + return names + } + for _, vf := range vfs { vGpuName, ok := vgpuTypeNamesByPF[vf.pfAddress][vf.typeID] if !ok { if name, found := globalTypeNames[vf.typeID]; found && !ambiguousTypeIDs[vf.typeID] { log.Printf("Resolved vGPU type %s on VF %s via the host-wide catalog (the card's own catalog is reduced)", vf.typeID, vf.device.addr) vGpuName = name + } else if name, found := nvmlSupportedNames(vf.pfAddress)[vf.typeID]; found { + log.Printf("Resolved vGPU type %s on VF %s via NVML (no sysfs catalog lists it)", vf.typeID, vf.device.addr) + vGpuName = name } else { log.Printf("Error: could not resolve the name of vGPU type %s configured on VF %s (parent PF %s), skipping device", vf.typeID, vf.device.addr, vf.pfAddress) continue diff --git a/pkg/device_plugin/vfio_vgpu_nvml.go b/pkg/device_plugin/vfio_vgpu_nvml.go new file mode 100644 index 00000000..f214308a --- /dev/null +++ b/pkg/device_plugin/vfio_vgpu_nvml.go @@ -0,0 +1,96 @@ +/* + * Copyright (c) 2026, NVIDIA CORPORATION. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * * Neither the name of NVIDIA CORPORATION nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY + * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, + * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, + * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR + * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY + * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +package device_plugin + +import ( + "fmt" + "reflect" + + "github.com/NVIDIA/go-nvml/pkg/nvml" +) + +// resolveVgpuTypeNamesViaNVML resolves the vGPU types SUPPORTED by the +// physical card at pfAddress into a type-id-to-sanitized-name map via NVML. +// This is the last-resort resolution source: unlike creatable_vgpu_types, +// the supported list does not shrink as capacity is allocated, so it keeps +// working on a fully consumed card where every sysfs catalog is reduced to +// the header. Requires libnvidia-ml.so.1 to be loadable in the container +// (e.g. the host driver library directory mounted and present in the +// dynamic linker search path). +var resolveVgpuTypeNamesViaNVML = func(pfAddress string) (map[string]string, error) { + return supportedVgpuTypeNames(nvml.New(), pfAddress) +} + +func supportedVgpuTypeNames(nvmllib nvml.Interface, pfAddress string) (map[string]string, error) { + ret := nvmllib.Init() + if ret != nvml.SUCCESS && ret != nvml.ERROR_ALREADY_INITIALIZED { + return nil, fmt.Errorf("error initializing NVML: %v", ret) + } + defer func() { + _ = nvmllib.Shutdown() + }() + + dev, ret := nvmllib.DeviceGetHandleByPciBusId(pfAddress) + if ret != nvml.SUCCESS { + return nil, fmt.Errorf("error getting device handle for PCI address %s: %v", pfAddress, ret) + } + + typeIDs, ret := dev.GetSupportedVgpus() + if ret != nvml.SUCCESS { + return nil, fmt.Errorf("error getting supported vGPU types for %s: %v", pfAddress, ret) + } + + names := make(map[string]string, len(typeIDs)) + for _, typeID := range typeIDs { + id, err := numericVgpuTypeID(typeID) + if err != nil { + return nil, err + } + name, ret := typeID.GetName() + if ret != nvml.SUCCESS { + return nil, fmt.Errorf("error getting the name of vGPU type %s: %v", id, ret) + } + names[id] = whitespaceRegexp.ReplaceAllString(name, "_") + } + return names, nil +} + +// numericVgpuTypeID renders an nvml.VgpuTypeId as the decimal string used by +// the sysfs vendor-specific VFIO files (current_vgpu_type, +// creatable_vgpu_types). +func numericVgpuTypeID(typeID nvml.VgpuTypeId) (string, error) { + v := reflect.ValueOf(typeID) + switch { + case v.CanUint(): + return fmt.Sprintf("%d", v.Uint()), nil + case v.CanInt(): + return fmt.Sprintf("%d", v.Int()), nil + } + return "", fmt.Errorf("unable to determine the numeric vGPU type id of %T", typeID) +} diff --git a/pkg/device_plugin/vfio_vgpu_test.go b/pkg/device_plugin/vfio_vgpu_test.go index f086b0dc..3dc4a802 100644 --- a/pkg/device_plugin/vfio_vgpu_test.go +++ b/pkg/device_plugin/vfio_vgpu_test.go @@ -29,6 +29,7 @@ package device_plugin import ( + "errors" "os" "path/filepath" "time" @@ -101,6 +102,10 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { readNUMANode = readNUMANodeFunc readVfioVgpuFile = readVfioVgpuFileFunc isPhysicalFunction = isPhysicalFunctionFunc + // NVML is unavailable in unit tests; resolve nothing by default. + resolveVgpuTypeNamesViaNVML = func(pfAddress string) (map[string]string, error) { + return map[string]string{}, nil + } linkDir, err = os.MkdirTemp("", "vfio-vgpu-link") Expect(err).ToNot(HaveOccurred()) @@ -297,6 +302,51 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { Expect(vfioVGpuMap[vfioVgpuResourceName1][0].addr).To(Equal(vfioVfAddress1)) }) + It("Resolves a type via NVML when no sysfs catalog lists it", func() { + // Every function on the host is consumed: all catalogs reduced. + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + creatableTypes: "ID : vGPU Name\n", + }) + resolvedPFs := []string{} + resolveVgpuTypeNamesViaNVML = func(pfAddress string) (map[string]string, error) { + resolvedPFs = append(resolvedPFs, pfAddress) + return map[string]string{"1428": vfioVgpuResourceName1}, nil + } + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1]).To(HaveLen(1)) + Expect(vfioVGpuMap[vfioVgpuResourceName1][0].addr).To(Equal(vfioVfAddress1)) + Expect(resolvedPFs).To(Equal([]string{vfioPfAddress})) + }) + + It("Skips the VF when NVML resolution fails too", func() { + createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ + vendor: "0x10de", + driver: "nvidia", + physfn: vfioPfAddress, + iommuGroup: "71", + numaNode: "0", + currentType: "1428", + creatableTypes: "ID : vGPU Name\n", + }) + resolveVgpuTypeNamesViaNVML = func(pfAddress string) (map[string]string, error) { + return nil, errors.New("nvml unavailable") + } + + createVfioVGpuMap() + + Expect(vfioVGpuMap).To(BeEmpty()) + Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress1)) + }) + It("Skips a VF whose type maps to conflicting names across cards", func() { // Fully consumed card with a reduced catalog. createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ diff --git a/vendor/github.com/NVIDIA/go-nvml/LICENSE b/vendor/github.com/NVIDIA/go-nvml/LICENSE new file mode 100644 index 00000000..d6456956 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/LICENSE @@ -0,0 +1,202 @@ + + Apache License + Version 2.0, January 2004 + http://www.apache.org/licenses/ + + TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION + + 1. Definitions. + + "License" shall mean the terms and conditions for use, reproduction, + and distribution as defined by Sections 1 through 9 of this document. + + "Licensor" shall mean the copyright owner or entity authorized by + the copyright owner that is granting the License. + + "Legal Entity" shall mean the union of the acting entity and all + other entities that control, are controlled by, or are under common + control with that entity. For the purposes of this definition, + "control" means (i) the power, direct or indirect, to cause the + direction or management of such entity, whether by contract or + otherwise, or (ii) ownership of fifty percent (50%) or more of the + outstanding shares, or (iii) beneficial ownership of such entity. + + "You" (or "Your") shall mean an individual or Legal Entity + exercising permissions granted by this License. + + "Source" form shall mean the preferred form for making modifications, + including but not limited to software source code, documentation + source, and configuration files. + + "Object" form shall mean any form resulting from mechanical + transformation or translation of a Source form, including but + not limited to compiled object code, generated documentation, + and conversions to other media types. + + "Work" shall mean the work of authorship, whether in Source or + Object form, made available under the License, as indicated by a + copyright notice that is included in or attached to the work + (an example is provided in the Appendix below). + + "Derivative Works" shall mean any work, whether in Source or Object + form, that is based on (or derived from) the Work and for which the + editorial revisions, annotations, elaborations, or other modifications + represent, as a whole, an original work of authorship. For the purposes + of this License, Derivative Works shall not include works that remain + separable from, or merely link (or bind by name) to the interfaces of, + the Work and Derivative Works thereof. + + "Contribution" shall mean any work of authorship, including + the original version of the Work and any modifications or additions + to that Work or Derivative Works thereof, that is intentionally + submitted to Licensor for inclusion in the Work by the copyright owner + or by an individual or Legal Entity authorized to submit on behalf of + the copyright owner. For the purposes of this definition, "submitted" + means any form of electronic, verbal, or written communication sent + to the Licensor or its representatives, including but not limited to + communication on electronic mailing lists, source code control systems, + and issue tracking systems that are managed by, or on behalf of, the + Licensor for the purpose of discussing and improving the Work, but + excluding communication that is conspicuously marked or otherwise + designated in writing by the copyright owner as "Not a Contribution." + + "Contributor" shall mean Licensor and any individual or Legal Entity + on behalf of whom a Contribution has been received by Licensor and + subsequently incorporated within the Work. + + 2. Grant of Copyright License. Subject to the terms and conditions of + this License, each Contributor hereby grants to You a perpetual, + worldwide, non-exclusive, no-charge, royalty-free, irrevocable + copyright license to reproduce, prepare Derivative Works of, + publicly display, publicly perform, sublicense, and distribute the + Work and such Derivative Works in Source or Object form. + + 3. Grant of Patent License. Subject to the terms and conditions of + this License, each Contributor hereby grants to You a perpetual, + worldwide, non-exclusive, no-charge, royalty-free, irrevocable + (except as stated in this section) patent license to make, have made, + use, offer to sell, sell, import, and otherwise transfer the Work, + where such license applies only to those patent claims licensable + by such Contributor that are necessarily infringed by their + Contribution(s) alone or by combination of their Contribution(s) + with the Work to which such Contribution(s) was submitted. If You + institute patent litigation against any entity (including a + cross-claim or counterclaim in a lawsuit) alleging that the Work + or a Contribution incorporated within the Work constitutes direct + or contributory patent infringement, then any patent licenses + granted to You under this License for that Work shall terminate + as of the date such litigation is filed. + + 4. Redistribution. You may reproduce and distribute copies of the + Work or Derivative Works thereof in any medium, with or without + modifications, and in Source or Object form, provided that You + meet the following conditions: + + (a) You must give any other recipients of the Work or + Derivative Works a copy of this License; and + + (b) You must cause any modified files to carry prominent notices + stating that You changed the files; and + + (c) You must retain, in the Source form of any Derivative Works + that You distribute, all copyright, patent, trademark, and + attribution notices from the Source form of the Work, + excluding those notices that do not pertain to any part of + the Derivative Works; and + + (d) If the Work includes a "NOTICE" text file as part of its + distribution, then any Derivative Works that You distribute must + include a readable copy of the attribution notices contained + within such NOTICE file, excluding those notices that do not + pertain to any part of the Derivative Works, in at least one + of the following places: within a NOTICE text file distributed + as part of the Derivative Works; within the Source form or + documentation, if provided along with the Derivative Works; or, + within a display generated by the Derivative Works, if and + wherever such third-party notices normally appear. The contents + of the NOTICE file are for informational purposes only and + do not modify the License. You may add Your own attribution + notices within Derivative Works that You distribute, alongside + or as an addendum to the NOTICE text from the Work, provided + that such additional attribution notices cannot be construed + as modifying the License. + + You may add Your own copyright statement to Your modifications and + may provide additional or different license terms and conditions + for use, reproduction, or distribution of Your modifications, or + for any such Derivative Works as a whole, provided Your use, + reproduction, and distribution of the Work otherwise complies with + the conditions stated in this License. + + 5. Submission of Contributions. Unless You explicitly state otherwise, + any Contribution intentionally submitted for inclusion in the Work + by You to the Licensor shall be under the terms and conditions of + this License, without any additional terms or conditions. + Notwithstanding the above, nothing herein shall supersede or modify + the terms of any separate license agreement you may have executed + with Licensor regarding such Contributions. + + 6. Trademarks. This License does not grant permission to use the trade + names, trademarks, service marks, or product names of the Licensor, + except as required for reasonable and customary use in describing the + origin of the Work and reproducing the content of the NOTICE file. + + 7. Disclaimer of Warranty. Unless required by applicable law or + agreed to in writing, Licensor provides the Work (and each + Contributor provides its Contributions) on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or + implied, including, without limitation, any warranties or conditions + of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A + PARTICULAR PURPOSE. You are solely responsible for determining the + appropriateness of using or redistributing the Work and assume any + risks associated with Your exercise of permissions under this License. + + 8. Limitation of Liability. In no event and under no legal theory, + whether in tort (including negligence), contract, or otherwise, + unless required by applicable law (such as deliberate and grossly + negligent acts) or agreed to in writing, shall any Contributor be + liable to You for damages, including any direct, indirect, special, + incidental, or consequential damages of any character arising as a + result of this License or out of the use or inability to use the + Work (including but not limited to damages for loss of goodwill, + work stoppage, computer failure or malfunction, or any and all + other commercial damages or losses), even if such Contributor + has been advised of the possibility of such damages. + + 9. Accepting Warranty or Additional Liability. While redistributing + the Work or Derivative Works thereof, You may choose to offer, + and charge a fee for, acceptance of support, warranty, indemnity, + or other liability obligations and/or rights consistent with this + License. However, in accepting such obligations, You may act only + on Your own behalf and on Your sole responsibility, not on behalf + of any other Contributor, and only if You agree to indemnify, + defend, and hold each Contributor harmless for any liability + incurred by, or claims asserted against, such Contributor by reason + of your accepting any such warranty or additional liability. + + END OF TERMS AND CONDITIONS + + APPENDIX: How to apply the Apache License to your work. + + To apply the Apache License to your work, attach the following + boilerplate notice, with the fields enclosed by brackets "[]" + replaced with your own identifying information. (Don't include + the brackets!) The text should be enclosed in the appropriate + comment syntax for the file format. We also recommend that a + file or class name and description of purpose be included on the + same "printed page" as the copyright notice for easier + identification within third-party archives. + + Copyright [yyyy] [name of copyright owner] + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + + Unless required by applicable law or agreed to in writing, software + distributed under the License is distributed on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + See the License for the specific language governing permissions and + limitations under the License. diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl.go b/vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl.go new file mode 100644 index 00000000..34948a72 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl.go @@ -0,0 +1,117 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package dl + +import ( + "errors" + "fmt" + "runtime" + "unsafe" +) + +// #cgo LDFLAGS: -ldl +// #include +// #include +import "C" + +const ( + RTLD_LAZY = C.RTLD_LAZY + RTLD_NOW = C.RTLD_NOW + RTLD_GLOBAL = C.RTLD_GLOBAL + RTLD_LOCAL = C.RTLD_LOCAL + RTLD_NODELETE = C.RTLD_NODELETE + RTLD_NOLOAD = C.RTLD_NOLOAD +) + +type DynamicLibrary struct { + Name string + Flags int + handle unsafe.Pointer +} + +func New(name string, flags int) *DynamicLibrary { + return &DynamicLibrary{ + Name: name, + Flags: flags, + handle: nil, + } +} + +func withOSLock(action func() error) error { + runtime.LockOSThread() + defer runtime.UnlockOSThread() + + return action() +} + +func dlError() error { + lastErr := C.dlerror() + if lastErr == nil { + return nil + } + return errors.New(C.GoString(lastErr)) +} + +func (dl *DynamicLibrary) Open() error { + name := C.CString(dl.Name) + defer C.free(unsafe.Pointer(name)) + + if err := withOSLock(func() error { + handle := C.dlopen(name, C.int(dl.Flags)) + if handle == nil { + return dlError() + } + dl.handle = handle + return nil + }); err != nil { + return err + } + return nil +} + +func (dl *DynamicLibrary) Close() error { + if dl.handle == nil { + return nil + } + if err := withOSLock(func() error { + if C.dlclose(dl.handle) != 0 { + return dlError() + } + dl.handle = nil + return nil + }); err != nil { + return err + } + return nil +} + +func (dl *DynamicLibrary) Lookup(symbol string) error { + sym := C.CString(symbol) + defer C.free(unsafe.Pointer(sym)) + + var pointer unsafe.Pointer + if err := withOSLock(func() error { + // Call dlError() to clear out any previous errors. + _ = dlError() + pointer = C.dlsym(dl.handle, sym) + if pointer == nil { + return fmt.Errorf("symbol %q not found: %w", symbol, dlError()) + } + return nil + }); err != nil { + return err + } + return nil +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl_linux.go b/vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl_linux.go new file mode 100644 index 00000000..ae3acd07 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/dl/dl_linux.go @@ -0,0 +1,26 @@ +/** +# Copyright (c) NVIDIA CORPORATION. All rights reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +**/ + +package dl + +// #cgo LDFLAGS: -ldl +// #include +// #include +import "C" + +const ( + RTLD_DEEPBIND = C.RTLD_DEEPBIND +) diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/api.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/api.go new file mode 100644 index 00000000..fdf27bda --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/api.go @@ -0,0 +1,56 @@ +/** +# Copyright 2023 NVIDIA CORPORATION +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +**/ + +package nvml + +// ExtendedInterface defines a set of extensions to the core NVML API. +// +// TODO: For now the list of methods in this interface need to be kept in sync +// with the list of excluded methods for the Interface type in +// gen/nvml/generateapi.go. In the future we should automate this. +// +//go:generate moq -out mock/extendedinterface.go -pkg mock . ExtendedInterface:ExtendedInterface +type ExtendedInterface interface { + LookupSymbol(string) error +} + +// libraryOptions hold the paramaters than can be set by a LibraryOption +type libraryOptions struct { + path string + flags int +} + +// LibraryOption represents a functional option to configure the underlying NVML library +type LibraryOption func(*libraryOptions) + +// WithLibraryPath provides an option to set the library name to be used by the NVML library. +func WithLibraryPath(path string) LibraryOption { + return func(o *libraryOptions) { + o.path = path + } +} + +// SetLibraryOptions applies the specified options to the NVML library. +// If this is called when a library is already loaded, an error is raised. +func SetLibraryOptions(opts ...LibraryOption) error { + libnvml.Lock() + defer libnvml.Unlock() + if libnvml.refcount != 0 { + return errLibraryAlreadyLoaded + } + libnvml.init(opts...) + return nil +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers.h b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers.h new file mode 100644 index 00000000..b25c5e5d --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers.h @@ -0,0 +1,23 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +// WARNING: THIS FILE WAS AUTOMATICALLY GENERATED. +// Code generated by https://git.io/c-for-go. DO NOT EDIT. + +#include "nvml.h" +#include +#pragma once + +#define __CGOGEN 1 + diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers_static.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers_static.go new file mode 100644 index 00000000..1f30eaae --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/cgo_helpers_static.go @@ -0,0 +1,75 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +import ( + "unsafe" +) + +import "C" + +var cgoAllocsUnknown = new(struct{}) + +type stringHeader struct { + Data unsafe.Pointer + Len int +} + +func clen(n []byte) int { + for i := 0; i < len(n); i++ { + if n[i] == 0 { + return i + } + } + return len(n) +} + +func uint32SliceToIntSlice(s []uint32) []int { + ret := make([]int, len(s)) + for i := range s { + ret[i] = int(s[i]) + } + return ret +} + +func convertSlice[T any, I any](input []T) []I { + output := make([]I, len(input)) + for i, obj := range input { + switch v := any(obj).(type) { + case I: + output[i] = v + } + } + return output +} + +// packPCharString creates a Go string backed by *C.char and avoids copying. +func packPCharString(p *C.char) (raw string) { + if p != nil && *p != 0 { + h := (*stringHeader)(unsafe.Pointer(&raw)) + h.Data = unsafe.Pointer(p) + for *p != 0 { + p = (*C.char)(unsafe.Pointer(uintptr(unsafe.Pointer(p)) + 1)) // p++ + } + h.Len = int(uintptr(unsafe.Pointer(p)) - uintptr(h.Data)) + } + return +} + +// unpackPCharString represents the data from Go string as *C.char and avoids copying. +func unpackPCharString(str string) (*C.char, *struct{}) { + h := (*stringHeader)(unsafe.Pointer(&str)) + return (*C.char)(h.Data), cgoAllocsUnknown +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const.go new file mode 100644 index 00000000..8a6a93c2 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const.go @@ -0,0 +1,2104 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +// WARNING: THIS FILE WAS AUTOMATICALLY GENERATED. +// Code generated by https://git.io/c-for-go. DO NOT EDIT. + +package nvml + +/* +#cgo linux LDFLAGS: -Wl,--export-dynamic -Wl,--unresolved-symbols=ignore-in-object-files +#cgo darwin LDFLAGS: -Wl,-undefined,dynamic_lookup +#cgo CFLAGS: -DNVML_NO_UNVERSIONED_FUNC_DEFS=1 +#include "nvml.h" +#include +#include "cgo_helpers.h" +*/ +import "C" + +const ( + // NO_UNVERSIONED_FUNC_DEFS as defined in go-nvml/:24 + NO_UNVERSIONED_FUNC_DEFS = 1 + // API_VERSION as defined in nvml/nvml.h + API_VERSION = 13 + // API_VERSION_STR as defined in nvml/nvml.h + API_VERSION_STR = "13" + // VALUE_NOT_AVAILABLE as defined in nvml/nvml.h + VALUE_NOT_AVAILABLE = -1 + // DEVICE_PCI_BUS_ID_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_PCI_BUS_ID_BUFFER_SIZE = 32 + // DEVICE_PCI_BUS_ID_BUFFER_V2_SIZE as defined in nvml/nvml.h + DEVICE_PCI_BUS_ID_BUFFER_V2_SIZE = 16 + // DEVICE_PCI_BUS_ID_LEGACY_FMT as defined in nvml/nvml.h + DEVICE_PCI_BUS_ID_LEGACY_FMT = "%04X:%02X:%02X.0" + // DEVICE_PCI_BUS_ID_FMT as defined in nvml/nvml.h + DEVICE_PCI_BUS_ID_FMT = "%08X:%02X:%02X.0" + // NVLINK_MAX_LINKS as defined in nvml/nvml.h + NVLINK_MAX_LINKS = 18 + // TOPOLOGY_CPU as defined in nvml/nvml.h + TOPOLOGY_CPU = 0 + // MAX_PHYSICAL_BRIDGE as defined in nvml/nvml.h + MAX_PHYSICAL_BRIDGE = 128 + // MAX_THERMAL_SENSORS_PER_GPU as defined in nvml/nvml.h + MAX_THERMAL_SENSORS_PER_GPU = 3 + // DEVICE_UUID_ASCII_LEN as defined in nvml/nvml.h + DEVICE_UUID_ASCII_LEN = 41 + // DEVICE_UUID_BINARY_LEN as defined in nvml/nvml.h + DEVICE_UUID_BINARY_LEN = 16 + // FlagDefault as defined in nvml/nvml.h + FlagDefault = 0 + // FlagForce as defined in nvml/nvml.h + FlagForce = 1 + // SINGLE_BIT_ECC as defined in nvml/nvml.h + SINGLE_BIT_ECC = 0 + // DOUBLE_BIT_ECC as defined in nvml/nvml.h + DOUBLE_BIT_ECC = 0 + // MAX_GPU_PERF_PSTATES as defined in nvml/nvml.h + MAX_GPU_PERF_PSTATES = 16 + // PERF_MODES_BUFFER_SIZE as defined in nvml/nvml.h + PERF_MODES_BUFFER_SIZE = 2048 + // POWER_MIZER_MODE_ADAPTIVE as defined in nvml/nvml.h + POWER_MIZER_MODE_ADAPTIVE = 0 + // POWER_MIZER_MODE_PREFER_MAXIMUM_PERFORMANCE as defined in nvml/nvml.h + POWER_MIZER_MODE_PREFER_MAXIMUM_PERFORMANCE = 1 + // POWER_MIZER_MODE_AUTO as defined in nvml/nvml.h + POWER_MIZER_MODE_AUTO = 2 + // POWER_MIZER_MODE_PREFER_CONSISTENT_PERFORMANCE as defined in nvml/nvml.h + POWER_MIZER_MODE_PREFER_CONSISTENT_PERFORMANCE = 3 + // GSP_FIRMWARE_VERSION_BUF_SIZE as defined in nvml/nvml.h + GSP_FIRMWARE_VERSION_BUF_SIZE = 64 + // DEVICE_ARCH_KEPLER as defined in nvml/nvml.h + DEVICE_ARCH_KEPLER = 2 + // DEVICE_ARCH_MAXWELL as defined in nvml/nvml.h + DEVICE_ARCH_MAXWELL = 3 + // DEVICE_ARCH_PASCAL as defined in nvml/nvml.h + DEVICE_ARCH_PASCAL = 4 + // DEVICE_ARCH_VOLTA as defined in nvml/nvml.h + DEVICE_ARCH_VOLTA = 5 + // DEVICE_ARCH_TURING as defined in nvml/nvml.h + DEVICE_ARCH_TURING = 6 + // DEVICE_ARCH_AMPERE as defined in nvml/nvml.h + DEVICE_ARCH_AMPERE = 7 + // DEVICE_ARCH_ADA as defined in nvml/nvml.h + DEVICE_ARCH_ADA = 8 + // DEVICE_ARCH_HOPPER as defined in nvml/nvml.h + DEVICE_ARCH_HOPPER = 9 + // DEVICE_ARCH_BLACKWELL as defined in nvml/nvml.h + DEVICE_ARCH_BLACKWELL = 10 + // DEVICE_ARCH_UNKNOWN as defined in nvml/nvml.h + DEVICE_ARCH_UNKNOWN = 4294967295 + // BUS_TYPE_UNKNOWN as defined in nvml/nvml.h + BUS_TYPE_UNKNOWN = 0 + // BUS_TYPE_PCI as defined in nvml/nvml.h + BUS_TYPE_PCI = 1 + // BUS_TYPE_PCIE as defined in nvml/nvml.h + BUS_TYPE_PCIE = 2 + // BUS_TYPE_FPCI as defined in nvml/nvml.h + BUS_TYPE_FPCI = 3 + // BUS_TYPE_AGP as defined in nvml/nvml.h + BUS_TYPE_AGP = 4 + // FAN_POLICY_TEMPERATURE_CONTINOUS_SW as defined in nvml/nvml.h + FAN_POLICY_TEMPERATURE_CONTINOUS_SW = 0 + // FAN_POLICY_MANUAL as defined in nvml/nvml.h + FAN_POLICY_MANUAL = 1 + // POWER_SOURCE_AC as defined in nvml/nvml.h + POWER_SOURCE_AC = 0 + // POWER_SOURCE_BATTERY as defined in nvml/nvml.h + POWER_SOURCE_BATTERY = 1 + // POWER_SOURCE_UNDERSIZED as defined in nvml/nvml.h + POWER_SOURCE_UNDERSIZED = 2 + // PCIE_LINK_MAX_SPEED_INVALID as defined in nvml/nvml.h + PCIE_LINK_MAX_SPEED_INVALID = 0 + // PCIE_LINK_MAX_SPEED_2500MBPS as defined in nvml/nvml.h + PCIE_LINK_MAX_SPEED_2500MBPS = 1 + // PCIE_LINK_MAX_SPEED_5000MBPS as defined in nvml/nvml.h + PCIE_LINK_MAX_SPEED_5000MBPS = 2 + // PCIE_LINK_MAX_SPEED_8000MBPS as defined in nvml/nvml.h + PCIE_LINK_MAX_SPEED_8000MBPS = 3 + // PCIE_LINK_MAX_SPEED_16000MBPS as defined in nvml/nvml.h + PCIE_LINK_MAX_SPEED_16000MBPS = 4 + // PCIE_LINK_MAX_SPEED_32000MBPS as defined in nvml/nvml.h + PCIE_LINK_MAX_SPEED_32000MBPS = 5 + // PCIE_LINK_MAX_SPEED_64000MBPS as defined in nvml/nvml.h + PCIE_LINK_MAX_SPEED_64000MBPS = 6 + // ADAPTIVE_CLOCKING_INFO_STATUS_DISABLED as defined in nvml/nvml.h + ADAPTIVE_CLOCKING_INFO_STATUS_DISABLED = 0 + // ADAPTIVE_CLOCKING_INFO_STATUS_ENABLED as defined in nvml/nvml.h + ADAPTIVE_CLOCKING_INFO_STATUS_ENABLED = 1 + // MAX_GPU_UTILIZATIONS as defined in nvml/nvml.h + MAX_GPU_UTILIZATIONS = 8 + // PCIE_ATOMICS_CAP_FETCHADD32 as defined in nvml/nvml.h + PCIE_ATOMICS_CAP_FETCHADD32 = 1 + // PCIE_ATOMICS_CAP_FETCHADD64 as defined in nvml/nvml.h + PCIE_ATOMICS_CAP_FETCHADD64 = 2 + // PCIE_ATOMICS_CAP_SWAP32 as defined in nvml/nvml.h + PCIE_ATOMICS_CAP_SWAP32 = 4 + // PCIE_ATOMICS_CAP_SWAP64 as defined in nvml/nvml.h + PCIE_ATOMICS_CAP_SWAP64 = 8 + // PCIE_ATOMICS_CAP_CAS32 as defined in nvml/nvml.h + PCIE_ATOMICS_CAP_CAS32 = 16 + // PCIE_ATOMICS_CAP_CAS64 as defined in nvml/nvml.h + PCIE_ATOMICS_CAP_CAS64 = 32 + // PCIE_ATOMICS_CAP_CAS128 as defined in nvml/nvml.h + PCIE_ATOMICS_CAP_CAS128 = 64 + // PCIE_ATOMICS_OPS_MAX as defined in nvml/nvml.h + PCIE_ATOMICS_OPS_MAX = 7 + // POWER_SCOPE_GPU as defined in nvml/nvml.h + POWER_SCOPE_GPU = 0 + // POWER_SCOPE_MODULE as defined in nvml/nvml.h + POWER_SCOPE_MODULE = 1 + // POWER_SCOPE_MEMORY as defined in nvml/nvml.h + POWER_SCOPE_MEMORY = 2 + // GRID_LICENSE_EXPIRY_NOT_AVAILABLE as defined in nvml/nvml.h + GRID_LICENSE_EXPIRY_NOT_AVAILABLE = 0 + // GRID_LICENSE_EXPIRY_INVALID as defined in nvml/nvml.h + GRID_LICENSE_EXPIRY_INVALID = 1 + // GRID_LICENSE_EXPIRY_VALID as defined in nvml/nvml.h + GRID_LICENSE_EXPIRY_VALID = 2 + // GRID_LICENSE_EXPIRY_NOT_APPLICABLE as defined in nvml/nvml.h + GRID_LICENSE_EXPIRY_NOT_APPLICABLE = 3 + // GRID_LICENSE_EXPIRY_PERMANENT as defined in nvml/nvml.h + GRID_LICENSE_EXPIRY_PERMANENT = 4 + // GRID_LICENSE_BUFFER_SIZE as defined in nvml/nvml.h + GRID_LICENSE_BUFFER_SIZE = 128 + // VGPU_NAME_BUFFER_SIZE as defined in nvml/nvml.h + VGPU_NAME_BUFFER_SIZE = 64 + // GRID_LICENSE_FEATURE_MAX_COUNT as defined in nvml/nvml.h + GRID_LICENSE_FEATURE_MAX_COUNT = 3 + // INVALID_VGPU_PLACEMENT_ID as defined in nvml/nvml.h + INVALID_VGPU_PLACEMENT_ID = 65535 + // VGPU_PGPU_HETEROGENEOUS_MODE as defined in nvml/nvml.h + VGPU_PGPU_HETEROGENEOUS_MODE = 0 + // VGPU_PGPU_HOMOGENEOUS_MODE as defined in nvml/nvml.h + VGPU_PGPU_HOMOGENEOUS_MODE = 1 + // VGPU_SCHEDULER_POLICY_UNKNOWN as defined in nvml/nvml.h + VGPU_SCHEDULER_POLICY_UNKNOWN = 0 + // VGPU_SCHEDULER_POLICY_BEST_EFFORT as defined in nvml/nvml.h + VGPU_SCHEDULER_POLICY_BEST_EFFORT = 1 + // VGPU_SCHEDULER_POLICY_EQUAL_SHARE as defined in nvml/nvml.h + VGPU_SCHEDULER_POLICY_EQUAL_SHARE = 2 + // VGPU_SCHEDULER_POLICY_FIXED_SHARE as defined in nvml/nvml.h + VGPU_SCHEDULER_POLICY_FIXED_SHARE = 3 + // SUPPORTED_VGPU_SCHEDULER_POLICY_COUNT as defined in nvml/nvml.h + SUPPORTED_VGPU_SCHEDULER_POLICY_COUNT = 3 + // SCHEDULER_SW_MAX_LOG_ENTRIES as defined in nvml/nvml.h + SCHEDULER_SW_MAX_LOG_ENTRIES = 200 + // VGPU_SCHEDULER_ARR_DEFAULT as defined in nvml/nvml.h + VGPU_SCHEDULER_ARR_DEFAULT = 0 + // VGPU_SCHEDULER_ARR_DISABLE as defined in nvml/nvml.h + VGPU_SCHEDULER_ARR_DISABLE = 1 + // VGPU_SCHEDULER_ARR_ENABLE as defined in nvml/nvml.h + VGPU_SCHEDULER_ARR_ENABLE = 2 + // VGPU_SCHEDULER_ENGINE_TYPE_GRAPHICS as defined in nvml/nvml.h + VGPU_SCHEDULER_ENGINE_TYPE_GRAPHICS = 1 + // GRID_LICENSE_STATE_UNKNOWN as defined in nvml/nvml.h + GRID_LICENSE_STATE_UNKNOWN = 0 + // GRID_LICENSE_STATE_UNINITIALIZED as defined in nvml/nvml.h + GRID_LICENSE_STATE_UNINITIALIZED = 1 + // GRID_LICENSE_STATE_UNLICENSED_UNRESTRICTED as defined in nvml/nvml.h + GRID_LICENSE_STATE_UNLICENSED_UNRESTRICTED = 2 + // GRID_LICENSE_STATE_UNLICENSED_RESTRICTED as defined in nvml/nvml.h + GRID_LICENSE_STATE_UNLICENSED_RESTRICTED = 3 + // GRID_LICENSE_STATE_UNLICENSED as defined in nvml/nvml.h + GRID_LICENSE_STATE_UNLICENSED = 4 + // GRID_LICENSE_STATE_LICENSED as defined in nvml/nvml.h + GRID_LICENSE_STATE_LICENSED = 5 + // FI_DEV_ECC_CURRENT as defined in nvml/nvml.h + FI_DEV_ECC_CURRENT = 1 + // FI_DEV_ECC_PENDING as defined in nvml/nvml.h + FI_DEV_ECC_PENDING = 2 + // FI_DEV_ECC_SBE_VOL_TOTAL as defined in nvml/nvml.h + FI_DEV_ECC_SBE_VOL_TOTAL = 3 + // FI_DEV_ECC_DBE_VOL_TOTAL as defined in nvml/nvml.h + FI_DEV_ECC_DBE_VOL_TOTAL = 4 + // FI_DEV_ECC_SBE_AGG_TOTAL as defined in nvml/nvml.h + FI_DEV_ECC_SBE_AGG_TOTAL = 5 + // FI_DEV_ECC_DBE_AGG_TOTAL as defined in nvml/nvml.h + FI_DEV_ECC_DBE_AGG_TOTAL = 6 + // FI_DEV_ECC_SBE_VOL_L1 as defined in nvml/nvml.h + FI_DEV_ECC_SBE_VOL_L1 = 7 + // FI_DEV_ECC_DBE_VOL_L1 as defined in nvml/nvml.h + FI_DEV_ECC_DBE_VOL_L1 = 8 + // FI_DEV_ECC_SBE_VOL_L2 as defined in nvml/nvml.h + FI_DEV_ECC_SBE_VOL_L2 = 9 + // FI_DEV_ECC_DBE_VOL_L2 as defined in nvml/nvml.h + FI_DEV_ECC_DBE_VOL_L2 = 10 + // FI_DEV_ECC_SBE_VOL_DEV as defined in nvml/nvml.h + FI_DEV_ECC_SBE_VOL_DEV = 11 + // FI_DEV_ECC_DBE_VOL_DEV as defined in nvml/nvml.h + FI_DEV_ECC_DBE_VOL_DEV = 12 + // FI_DEV_ECC_SBE_VOL_REG as defined in nvml/nvml.h + FI_DEV_ECC_SBE_VOL_REG = 13 + // FI_DEV_ECC_DBE_VOL_REG as defined in nvml/nvml.h + FI_DEV_ECC_DBE_VOL_REG = 14 + // FI_DEV_ECC_SBE_VOL_TEX as defined in nvml/nvml.h + FI_DEV_ECC_SBE_VOL_TEX = 15 + // FI_DEV_ECC_DBE_VOL_TEX as defined in nvml/nvml.h + FI_DEV_ECC_DBE_VOL_TEX = 16 + // FI_DEV_ECC_DBE_VOL_CBU as defined in nvml/nvml.h + FI_DEV_ECC_DBE_VOL_CBU = 17 + // FI_DEV_ECC_SBE_AGG_L1 as defined in nvml/nvml.h + FI_DEV_ECC_SBE_AGG_L1 = 18 + // FI_DEV_ECC_DBE_AGG_L1 as defined in nvml/nvml.h + FI_DEV_ECC_DBE_AGG_L1 = 19 + // FI_DEV_ECC_SBE_AGG_L2 as defined in nvml/nvml.h + FI_DEV_ECC_SBE_AGG_L2 = 20 + // FI_DEV_ECC_DBE_AGG_L2 as defined in nvml/nvml.h + FI_DEV_ECC_DBE_AGG_L2 = 21 + // FI_DEV_ECC_SBE_AGG_DEV as defined in nvml/nvml.h + FI_DEV_ECC_SBE_AGG_DEV = 22 + // FI_DEV_ECC_DBE_AGG_DEV as defined in nvml/nvml.h + FI_DEV_ECC_DBE_AGG_DEV = 23 + // FI_DEV_ECC_SBE_AGG_REG as defined in nvml/nvml.h + FI_DEV_ECC_SBE_AGG_REG = 24 + // FI_DEV_ECC_DBE_AGG_REG as defined in nvml/nvml.h + FI_DEV_ECC_DBE_AGG_REG = 25 + // FI_DEV_ECC_SBE_AGG_TEX as defined in nvml/nvml.h + FI_DEV_ECC_SBE_AGG_TEX = 26 + // FI_DEV_ECC_DBE_AGG_TEX as defined in nvml/nvml.h + FI_DEV_ECC_DBE_AGG_TEX = 27 + // FI_DEV_ECC_DBE_AGG_CBU as defined in nvml/nvml.h + FI_DEV_ECC_DBE_AGG_CBU = 28 + // FI_DEV_RETIRED_SBE as defined in nvml/nvml.h + FI_DEV_RETIRED_SBE = 29 + // FI_DEV_RETIRED_DBE as defined in nvml/nvml.h + FI_DEV_RETIRED_DBE = 30 + // FI_DEV_RETIRED_PENDING as defined in nvml/nvml.h + FI_DEV_RETIRED_PENDING = 31 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L0 = 32 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L1 = 33 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L2 = 34 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L3 = 35 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L4 = 36 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L5 = 37 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_TOTAL as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_TOTAL = 38 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L0 = 39 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L1 = 40 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L2 = 41 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L3 = 42 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L4 = 43 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L5 = 44 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_TOTAL as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_TOTAL = 45 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L0 = 46 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L1 = 47 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L2 = 48 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L3 = 49 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L4 = 50 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L5 = 51 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_TOTAL as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_TOTAL = 52 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L0 = 53 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L1 = 54 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L2 = 55 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L3 = 56 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L4 = 57 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L5 = 58 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_TOTAL as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_TOTAL = 59 + // FI_DEV_NVLINK_BANDWIDTH_C0_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L0 = 60 + // FI_DEV_NVLINK_BANDWIDTH_C0_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L1 = 61 + // FI_DEV_NVLINK_BANDWIDTH_C0_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L2 = 62 + // FI_DEV_NVLINK_BANDWIDTH_C0_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L3 = 63 + // FI_DEV_NVLINK_BANDWIDTH_C0_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L4 = 64 + // FI_DEV_NVLINK_BANDWIDTH_C0_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L5 = 65 + // FI_DEV_NVLINK_BANDWIDTH_C0_TOTAL as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_TOTAL = 66 + // FI_DEV_NVLINK_BANDWIDTH_C1_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L0 = 67 + // FI_DEV_NVLINK_BANDWIDTH_C1_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L1 = 68 + // FI_DEV_NVLINK_BANDWIDTH_C1_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L2 = 69 + // FI_DEV_NVLINK_BANDWIDTH_C1_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L3 = 70 + // FI_DEV_NVLINK_BANDWIDTH_C1_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L4 = 71 + // FI_DEV_NVLINK_BANDWIDTH_C1_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L5 = 72 + // FI_DEV_NVLINK_BANDWIDTH_C1_TOTAL as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_TOTAL = 73 + // FI_DEV_PERF_POLICY_POWER as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_POWER = 74 + // FI_DEV_PERF_POLICY_THERMAL as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_THERMAL = 75 + // FI_DEV_PERF_POLICY_SYNC_BOOST as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_SYNC_BOOST = 76 + // FI_DEV_PERF_POLICY_BOARD_LIMIT as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_BOARD_LIMIT = 77 + // FI_DEV_PERF_POLICY_LOW_UTILIZATION as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_LOW_UTILIZATION = 78 + // FI_DEV_PERF_POLICY_RELIABILITY as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_RELIABILITY = 79 + // FI_DEV_PERF_POLICY_TOTAL_APP_CLOCKS as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_TOTAL_APP_CLOCKS = 80 + // FI_DEV_PERF_POLICY_TOTAL_BASE_CLOCKS as defined in nvml/nvml.h + FI_DEV_PERF_POLICY_TOTAL_BASE_CLOCKS = 81 + // FI_DEV_MEMORY_TEMP as defined in nvml/nvml.h + FI_DEV_MEMORY_TEMP = 82 + // FI_DEV_TOTAL_ENERGY_CONSUMPTION as defined in nvml/nvml.h + FI_DEV_TOTAL_ENERGY_CONSUMPTION = 83 + // FI_DEV_NVLINK_SPEED_MBPS_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L0 = 84 + // FI_DEV_NVLINK_SPEED_MBPS_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L1 = 85 + // FI_DEV_NVLINK_SPEED_MBPS_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L2 = 86 + // FI_DEV_NVLINK_SPEED_MBPS_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L3 = 87 + // FI_DEV_NVLINK_SPEED_MBPS_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L4 = 88 + // FI_DEV_NVLINK_SPEED_MBPS_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L5 = 89 + // FI_DEV_NVLINK_SPEED_MBPS_COMMON as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_COMMON = 90 + // FI_DEV_NVLINK_LINK_COUNT as defined in nvml/nvml.h + FI_DEV_NVLINK_LINK_COUNT = 91 + // FI_DEV_RETIRED_PENDING_SBE as defined in nvml/nvml.h + FI_DEV_RETIRED_PENDING_SBE = 92 + // FI_DEV_RETIRED_PENDING_DBE as defined in nvml/nvml.h + FI_DEV_RETIRED_PENDING_DBE = 93 + // FI_DEV_PCIE_REPLAY_COUNTER as defined in nvml/nvml.h + FI_DEV_PCIE_REPLAY_COUNTER = 94 + // FI_DEV_PCIE_REPLAY_ROLLOVER_COUNTER as defined in nvml/nvml.h + FI_DEV_PCIE_REPLAY_ROLLOVER_COUNTER = 95 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L6 = 96 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L7 = 97 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L8 = 98 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L9 = 99 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L10 = 100 + // FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L11 = 101 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L6 = 102 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L7 = 103 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L8 = 104 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L9 = 105 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L10 = 106 + // FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L11 = 107 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L6 = 108 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L7 = 109 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L8 = 110 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L9 = 111 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L10 = 112 + // FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L11 = 113 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L6 = 114 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L7 = 115 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L8 = 116 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L9 = 117 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L10 = 118 + // FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L11 = 119 + // FI_DEV_NVLINK_BANDWIDTH_C0_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L6 = 120 + // FI_DEV_NVLINK_BANDWIDTH_C0_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L7 = 121 + // FI_DEV_NVLINK_BANDWIDTH_C0_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L8 = 122 + // FI_DEV_NVLINK_BANDWIDTH_C0_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L9 = 123 + // FI_DEV_NVLINK_BANDWIDTH_C0_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L10 = 124 + // FI_DEV_NVLINK_BANDWIDTH_C0_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C0_L11 = 125 + // FI_DEV_NVLINK_BANDWIDTH_C1_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L6 = 126 + // FI_DEV_NVLINK_BANDWIDTH_C1_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L7 = 127 + // FI_DEV_NVLINK_BANDWIDTH_C1_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L8 = 128 + // FI_DEV_NVLINK_BANDWIDTH_C1_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L9 = 129 + // FI_DEV_NVLINK_BANDWIDTH_C1_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L10 = 130 + // FI_DEV_NVLINK_BANDWIDTH_C1_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_BANDWIDTH_C1_L11 = 131 + // FI_DEV_NVLINK_SPEED_MBPS_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L6 = 132 + // FI_DEV_NVLINK_SPEED_MBPS_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L7 = 133 + // FI_DEV_NVLINK_SPEED_MBPS_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L8 = 134 + // FI_DEV_NVLINK_SPEED_MBPS_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L9 = 135 + // FI_DEV_NVLINK_SPEED_MBPS_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L10 = 136 + // FI_DEV_NVLINK_SPEED_MBPS_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_SPEED_MBPS_L11 = 137 + // FI_DEV_NVLINK_THROUGHPUT_DATA_TX as defined in nvml/nvml.h + FI_DEV_NVLINK_THROUGHPUT_DATA_TX = 138 + // FI_DEV_NVLINK_THROUGHPUT_DATA_RX as defined in nvml/nvml.h + FI_DEV_NVLINK_THROUGHPUT_DATA_RX = 139 + // FI_DEV_NVLINK_THROUGHPUT_RAW_TX as defined in nvml/nvml.h + FI_DEV_NVLINK_THROUGHPUT_RAW_TX = 140 + // FI_DEV_NVLINK_THROUGHPUT_RAW_RX as defined in nvml/nvml.h + FI_DEV_NVLINK_THROUGHPUT_RAW_RX = 141 + // FI_DEV_REMAPPED_COR as defined in nvml/nvml.h + FI_DEV_REMAPPED_COR = 142 + // FI_DEV_REMAPPED_UNC as defined in nvml/nvml.h + FI_DEV_REMAPPED_UNC = 143 + // FI_DEV_REMAPPED_PENDING as defined in nvml/nvml.h + FI_DEV_REMAPPED_PENDING = 144 + // FI_DEV_REMAPPED_FAILURE as defined in nvml/nvml.h + FI_DEV_REMAPPED_FAILURE = 145 + // FI_DEV_NVLINK_REMOTE_NVLINK_ID as defined in nvml/nvml.h + FI_DEV_NVLINK_REMOTE_NVLINK_ID = 146 + // FI_DEV_NVSWITCH_CONNECTED_LINK_COUNT as defined in nvml/nvml.h + FI_DEV_NVSWITCH_CONNECTED_LINK_COUNT = 147 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L0 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L0 = 148 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L1 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L1 = 149 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L2 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L2 = 150 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L3 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L3 = 151 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L4 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L4 = 152 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L5 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L5 = 153 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L6 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L6 = 154 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L7 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L7 = 155 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L8 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L8 = 156 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L9 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L9 = 157 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L10 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L10 = 158 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L11 as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L11 = 159 + // FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_TOTAL as defined in nvml/nvml.h + FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_TOTAL = 160 + // FI_DEV_NVLINK_ERROR_DL_REPLAY as defined in nvml/nvml.h + FI_DEV_NVLINK_ERROR_DL_REPLAY = 161 + // FI_DEV_NVLINK_ERROR_DL_RECOVERY as defined in nvml/nvml.h + FI_DEV_NVLINK_ERROR_DL_RECOVERY = 162 + // FI_DEV_NVLINK_ERROR_DL_CRC as defined in nvml/nvml.h + FI_DEV_NVLINK_ERROR_DL_CRC = 163 + // FI_DEV_NVLINK_GET_SPEED as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_SPEED = 164 + // FI_DEV_NVLINK_GET_STATE as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_STATE = 165 + // FI_DEV_NVLINK_GET_VERSION as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_VERSION = 166 + // FI_DEV_NVLINK_GET_POWER_STATE as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_POWER_STATE = 167 + // FI_DEV_NVLINK_GET_POWER_THRESHOLD as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_POWER_THRESHOLD = 168 + // FI_DEV_PCIE_L0_TO_RECOVERY_COUNTER as defined in nvml/nvml.h + FI_DEV_PCIE_L0_TO_RECOVERY_COUNTER = 169 + // FI_DEV_C2C_LINK_COUNT as defined in nvml/nvml.h + FI_DEV_C2C_LINK_COUNT = 170 + // FI_DEV_C2C_LINK_GET_STATUS as defined in nvml/nvml.h + FI_DEV_C2C_LINK_GET_STATUS = 171 + // FI_DEV_C2C_LINK_GET_MAX_BW as defined in nvml/nvml.h + FI_DEV_C2C_LINK_GET_MAX_BW = 172 + // FI_DEV_PCIE_COUNT_CORRECTABLE_ERRORS as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_CORRECTABLE_ERRORS = 173 + // FI_DEV_PCIE_COUNT_NAKS_RECEIVED as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_NAKS_RECEIVED = 174 + // FI_DEV_PCIE_COUNT_RECEIVER_ERROR as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_RECEIVER_ERROR = 175 + // FI_DEV_PCIE_COUNT_BAD_TLP as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_BAD_TLP = 176 + // FI_DEV_PCIE_COUNT_NAKS_SENT as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_NAKS_SENT = 177 + // FI_DEV_PCIE_COUNT_BAD_DLLP as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_BAD_DLLP = 178 + // FI_DEV_PCIE_COUNT_NON_FATAL_ERROR as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_NON_FATAL_ERROR = 179 + // FI_DEV_PCIE_COUNT_FATAL_ERROR as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_FATAL_ERROR = 180 + // FI_DEV_PCIE_COUNT_UNSUPPORTED_REQ as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_UNSUPPORTED_REQ = 181 + // FI_DEV_PCIE_COUNT_LCRC_ERROR as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_LCRC_ERROR = 182 + // FI_DEV_PCIE_COUNT_LANE_ERROR as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_LANE_ERROR = 183 + // FI_DEV_IS_RESETLESS_MIG_SUPPORTED as defined in nvml/nvml.h + FI_DEV_IS_RESETLESS_MIG_SUPPORTED = 184 + // FI_DEV_POWER_AVERAGE as defined in nvml/nvml.h + FI_DEV_POWER_AVERAGE = 185 + // FI_DEV_POWER_INSTANT as defined in nvml/nvml.h + FI_DEV_POWER_INSTANT = 186 + // FI_DEV_POWER_MIN_LIMIT as defined in nvml/nvml.h + FI_DEV_POWER_MIN_LIMIT = 187 + // FI_DEV_POWER_MAX_LIMIT as defined in nvml/nvml.h + FI_DEV_POWER_MAX_LIMIT = 188 + // FI_DEV_POWER_DEFAULT_LIMIT as defined in nvml/nvml.h + FI_DEV_POWER_DEFAULT_LIMIT = 189 + // FI_DEV_POWER_CURRENT_LIMIT as defined in nvml/nvml.h + FI_DEV_POWER_CURRENT_LIMIT = 190 + // FI_DEV_ENERGY as defined in nvml/nvml.h + FI_DEV_ENERGY = 191 + // FI_DEV_POWER_REQUESTED_LIMIT as defined in nvml/nvml.h + FI_DEV_POWER_REQUESTED_LIMIT = 192 + // FI_DEV_TEMPERATURE_SHUTDOWN_TLIMIT as defined in nvml/nvml.h + FI_DEV_TEMPERATURE_SHUTDOWN_TLIMIT = 193 + // FI_DEV_TEMPERATURE_SLOWDOWN_TLIMIT as defined in nvml/nvml.h + FI_DEV_TEMPERATURE_SLOWDOWN_TLIMIT = 194 + // FI_DEV_TEMPERATURE_MEM_MAX_TLIMIT as defined in nvml/nvml.h + FI_DEV_TEMPERATURE_MEM_MAX_TLIMIT = 195 + // FI_DEV_TEMPERATURE_GPU_MAX_TLIMIT as defined in nvml/nvml.h + FI_DEV_TEMPERATURE_GPU_MAX_TLIMIT = 196 + // FI_DEV_PCIE_COUNT_TX_BYTES as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_TX_BYTES = 197 + // FI_DEV_PCIE_COUNT_RX_BYTES as defined in nvml/nvml.h + FI_DEV_PCIE_COUNT_RX_BYTES = 198 + // FI_DEV_IS_MIG_MODE_INDEPENDENT_MIG_QUERY_CAPABLE as defined in nvml/nvml.h + FI_DEV_IS_MIG_MODE_INDEPENDENT_MIG_QUERY_CAPABLE = 199 + // FI_DEV_NVLINK_GET_POWER_THRESHOLD_MAX as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_POWER_THRESHOLD_MAX = 200 + // FI_DEV_NVLINK_COUNT_XMIT_PACKETS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_XMIT_PACKETS = 201 + // FI_DEV_NVLINK_COUNT_XMIT_BYTES as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_XMIT_BYTES = 202 + // FI_DEV_NVLINK_COUNT_RCV_PACKETS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RCV_PACKETS = 203 + // FI_DEV_NVLINK_COUNT_RCV_BYTES as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RCV_BYTES = 204 + // FI_DEV_NVLINK_COUNT_VL15_DROPPED as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_VL15_DROPPED = 205 + // FI_DEV_NVLINK_COUNT_MALFORMED_PACKET_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_MALFORMED_PACKET_ERRORS = 206 + // FI_DEV_NVLINK_COUNT_BUFFER_OVERRUN_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_BUFFER_OVERRUN_ERRORS = 207 + // FI_DEV_NVLINK_COUNT_RCV_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RCV_ERRORS = 208 + // FI_DEV_NVLINK_COUNT_RCV_REMOTE_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RCV_REMOTE_ERRORS = 209 + // FI_DEV_NVLINK_COUNT_RCV_GENERAL_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RCV_GENERAL_ERRORS = 210 + // FI_DEV_NVLINK_COUNT_LOCAL_LINK_INTEGRITY_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_LOCAL_LINK_INTEGRITY_ERRORS = 211 + // FI_DEV_NVLINK_COUNT_XMIT_DISCARDS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_XMIT_DISCARDS = 212 + // FI_DEV_NVLINK_COUNT_LINK_RECOVERY_SUCCESSFUL_EVENTS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_LINK_RECOVERY_SUCCESSFUL_EVENTS = 213 + // FI_DEV_NVLINK_COUNT_LINK_RECOVERY_FAILED_EVENTS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_LINK_RECOVERY_FAILED_EVENTS = 214 + // FI_DEV_NVLINK_COUNT_LINK_RECOVERY_EVENTS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_LINK_RECOVERY_EVENTS = 215 + // FI_DEV_NVLINK_COUNT_RAW_BER_LANE0 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RAW_BER_LANE0 = 216 + // FI_DEV_NVLINK_COUNT_RAW_BER_LANE1 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RAW_BER_LANE1 = 217 + // FI_DEV_NVLINK_COUNT_RAW_BER as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_RAW_BER = 218 + // FI_DEV_NVLINK_COUNT_EFFECTIVE_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_EFFECTIVE_ERRORS = 219 + // FI_DEV_NVLINK_COUNT_EFFECTIVE_BER as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_EFFECTIVE_BER = 220 + // FI_DEV_NVLINK_COUNT_SYMBOL_ERRORS as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_SYMBOL_ERRORS = 221 + // FI_DEV_NVLINK_COUNT_SYMBOL_BER as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_SYMBOL_BER = 222 + // FI_DEV_NVLINK_GET_POWER_THRESHOLD_MIN as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_POWER_THRESHOLD_MIN = 223 + // FI_DEV_NVLINK_GET_POWER_THRESHOLD_UNITS as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_POWER_THRESHOLD_UNITS = 224 + // FI_DEV_NVLINK_GET_POWER_THRESHOLD_SUPPORTED as defined in nvml/nvml.h + FI_DEV_NVLINK_GET_POWER_THRESHOLD_SUPPORTED = 225 + // FI_DEV_RESET_STATUS as defined in nvml/nvml.h + FI_DEV_RESET_STATUS = 226 + // FI_DEV_DRAIN_AND_RESET_STATUS as defined in nvml/nvml.h + FI_DEV_DRAIN_AND_RESET_STATUS = 227 + // FI_DEV_PCIE_OUTBOUND_ATOMICS_MASK as defined in nvml/nvml.h + FI_DEV_PCIE_OUTBOUND_ATOMICS_MASK = 228 + // FI_DEV_PCIE_INBOUND_ATOMICS_MASK as defined in nvml/nvml.h + FI_DEV_PCIE_INBOUND_ATOMICS_MASK = 229 + // FI_DEV_GET_GPU_RECOVERY_ACTION as defined in nvml/nvml.h + FI_DEV_GET_GPU_RECOVERY_ACTION = 230 + // FI_DEV_C2C_LINK_ERROR_INTR as defined in nvml/nvml.h + FI_DEV_C2C_LINK_ERROR_INTR = 231 + // FI_DEV_C2C_LINK_ERROR_REPLAY as defined in nvml/nvml.h + FI_DEV_C2C_LINK_ERROR_REPLAY = 232 + // FI_DEV_C2C_LINK_ERROR_REPLAY_B2B as defined in nvml/nvml.h + FI_DEV_C2C_LINK_ERROR_REPLAY_B2B = 233 + // FI_DEV_C2C_LINK_POWER_STATE as defined in nvml/nvml.h + FI_DEV_C2C_LINK_POWER_STATE = 234 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_0 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_0 = 235 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_1 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_1 = 236 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_2 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_2 = 237 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_3 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_3 = 238 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_4 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_4 = 239 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_5 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_5 = 240 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_6 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_6 = 241 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_7 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_7 = 242 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_8 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_8 = 243 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_9 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_9 = 244 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_10 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_10 = 245 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_11 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_11 = 246 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_12 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_12 = 247 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_13 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_13 = 248 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_14 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_14 = 249 + // FI_DEV_NVLINK_COUNT_FEC_HISTORY_15 as defined in nvml/nvml.h + FI_DEV_NVLINK_COUNT_FEC_HISTORY_15 = 250 + // FI_DEV_CLOCKS_EVENT_REASON_SW_POWER_CAP as defined in nvml/nvml.h + FI_DEV_CLOCKS_EVENT_REASON_SW_POWER_CAP = 74 + // FI_DEV_CLOCKS_EVENT_REASON_SYNC_BOOST as defined in nvml/nvml.h + FI_DEV_CLOCKS_EVENT_REASON_SYNC_BOOST = 76 + // FI_DEV_CLOCKS_EVENT_REASON_SW_THERM_SLOWDOWN as defined in nvml/nvml.h + FI_DEV_CLOCKS_EVENT_REASON_SW_THERM_SLOWDOWN = 251 + // FI_DEV_CLOCKS_EVENT_REASON_HW_THERM_SLOWDOWN as defined in nvml/nvml.h + FI_DEV_CLOCKS_EVENT_REASON_HW_THERM_SLOWDOWN = 252 + // FI_DEV_CLOCKS_EVENT_REASON_HW_POWER_BRAKE_SLOWDOWN as defined in nvml/nvml.h + FI_DEV_CLOCKS_EVENT_REASON_HW_POWER_BRAKE_SLOWDOWN = 253 + // FI_DEV_POWER_SYNC_BALANCING_FREQ as defined in nvml/nvml.h + FI_DEV_POWER_SYNC_BALANCING_FREQ = 254 + // FI_DEV_POWER_SYNC_BALANCING_AF as defined in nvml/nvml.h + FI_DEV_POWER_SYNC_BALANCING_AF = 255 + // FI_PWR_SMOOTHING_ENABLED as defined in nvml/nvml.h + FI_PWR_SMOOTHING_ENABLED = 256 + // FI_PWR_SMOOTHING_PRIV_LVL as defined in nvml/nvml.h + FI_PWR_SMOOTHING_PRIV_LVL = 257 + // FI_PWR_SMOOTHING_IMM_RAMP_DOWN_ENABLED as defined in nvml/nvml.h + FI_PWR_SMOOTHING_IMM_RAMP_DOWN_ENABLED = 258 + // FI_PWR_SMOOTHING_APPLIED_TMP_CEIL as defined in nvml/nvml.h + FI_PWR_SMOOTHING_APPLIED_TMP_CEIL = 259 + // FI_PWR_SMOOTHING_APPLIED_TMP_FLOOR as defined in nvml/nvml.h + FI_PWR_SMOOTHING_APPLIED_TMP_FLOOR = 260 + // FI_PWR_SMOOTHING_MAX_PERCENT_TMP_FLOOR_SETTING as defined in nvml/nvml.h + FI_PWR_SMOOTHING_MAX_PERCENT_TMP_FLOOR_SETTING = 261 + // FI_PWR_SMOOTHING_MIN_PERCENT_TMP_FLOOR_SETTING as defined in nvml/nvml.h + FI_PWR_SMOOTHING_MIN_PERCENT_TMP_FLOOR_SETTING = 262 + // FI_PWR_SMOOTHING_HW_CIRCUITRY_PERCENT_LIFETIME_REMAINING as defined in nvml/nvml.h + FI_PWR_SMOOTHING_HW_CIRCUITRY_PERCENT_LIFETIME_REMAINING = 263 + // FI_PWR_SMOOTHING_MAX_NUM_PRESET_PROFILES as defined in nvml/nvml.h + FI_PWR_SMOOTHING_MAX_NUM_PRESET_PROFILES = 264 + // FI_PWR_SMOOTHING_PROFILE_PERCENT_TMP_FLOOR as defined in nvml/nvml.h + FI_PWR_SMOOTHING_PROFILE_PERCENT_TMP_FLOOR = 265 + // FI_PWR_SMOOTHING_PROFILE_RAMP_UP_RATE as defined in nvml/nvml.h + FI_PWR_SMOOTHING_PROFILE_RAMP_UP_RATE = 266 + // FI_PWR_SMOOTHING_PROFILE_RAMP_DOWN_RATE as defined in nvml/nvml.h + FI_PWR_SMOOTHING_PROFILE_RAMP_DOWN_RATE = 267 + // FI_PWR_SMOOTHING_PROFILE_RAMP_DOWN_HYST_VAL as defined in nvml/nvml.h + FI_PWR_SMOOTHING_PROFILE_RAMP_DOWN_HYST_VAL = 268 + // FI_PWR_SMOOTHING_ACTIVE_PRESET_PROFILE as defined in nvml/nvml.h + FI_PWR_SMOOTHING_ACTIVE_PRESET_PROFILE = 269 + // FI_PWR_SMOOTHING_ADMIN_OVERRIDE_PERCENT_TMP_FLOOR as defined in nvml/nvml.h + FI_PWR_SMOOTHING_ADMIN_OVERRIDE_PERCENT_TMP_FLOOR = 270 + // FI_PWR_SMOOTHING_ADMIN_OVERRIDE_RAMP_UP_RATE as defined in nvml/nvml.h + FI_PWR_SMOOTHING_ADMIN_OVERRIDE_RAMP_UP_RATE = 271 + // FI_PWR_SMOOTHING_ADMIN_OVERRIDE_RAMP_DOWN_RATE as defined in nvml/nvml.h + FI_PWR_SMOOTHING_ADMIN_OVERRIDE_RAMP_DOWN_RATE = 272 + // FI_PWR_SMOOTHING_ADMIN_OVERRIDE_RAMP_DOWN_HYST_VAL as defined in nvml/nvml.h + FI_PWR_SMOOTHING_ADMIN_OVERRIDE_RAMP_DOWN_HYST_VAL = 273 + // FI_MAX as defined in nvml/nvml.h + FI_MAX = 274 + // NVLINK_LOW_POWER_THRESHOLD_UNIT_100US as defined in nvml/nvml.h + NVLINK_LOW_POWER_THRESHOLD_UNIT_100US = 0 + // NVLINK_LOW_POWER_THRESHOLD_UNIT_50US as defined in nvml/nvml.h + NVLINK_LOW_POWER_THRESHOLD_UNIT_50US = 1 + // NVLINK_POWER_STATE_HIGH_SPEED as defined in nvml/nvml.h + NVLINK_POWER_STATE_HIGH_SPEED = 0 + // NVLINK_POWER_STATE_LOW as defined in nvml/nvml.h + NVLINK_POWER_STATE_LOW = 1 + // NVLINK_LOW_POWER_THRESHOLD_MIN as defined in nvml/nvml.h + NVLINK_LOW_POWER_THRESHOLD_MIN = 1 + // NVLINK_LOW_POWER_THRESHOLD_MAX as defined in nvml/nvml.h + NVLINK_LOW_POWER_THRESHOLD_MAX = 8191 + // NVLINK_LOW_POWER_THRESHOLD_RESET as defined in nvml/nvml.h + NVLINK_LOW_POWER_THRESHOLD_RESET = 4294967295 + // NVLINK_LOW_POWER_THRESHOLD_DEFAULT as defined in nvml/nvml.h + NVLINK_LOW_POWER_THRESHOLD_DEFAULT = 4294967295 + // C2C_POWER_STATE_FULL_POWER as defined in nvml/nvml.h + C2C_POWER_STATE_FULL_POWER = 0 + // C2C_POWER_STATE_LOW_POWER as defined in nvml/nvml.h + C2C_POWER_STATE_LOW_POWER = 1 + // EventTypeNone as defined in nvml/nvml.h + EventTypeNone = 0 + // EventTypeSingleBitEccError as defined in nvml/nvml.h + EventTypeSingleBitEccError = 1 + // EventTypeDoubleBitEccError as defined in nvml/nvml.h + EventTypeDoubleBitEccError = 2 + // EventTypePState as defined in nvml/nvml.h + EventTypePState = 4 + // EventTypeXidCriticalError as defined in nvml/nvml.h + EventTypeXidCriticalError = 8 + // EventTypeClock as defined in nvml/nvml.h + EventTypeClock = 16 + // EventTypePowerSourceChange as defined in nvml/nvml.h + EventTypePowerSourceChange = 128 + // EventMigConfigChange as defined in nvml/nvml.h + EventMigConfigChange = 256 + // EventTypeSingleBitEccErrorStorm as defined in nvml/nvml.h + EventTypeSingleBitEccErrorStorm = 512 + // EventTypeDramRetirementEvent as defined in nvml/nvml.h + EventTypeDramRetirementEvent = 1024 + // EventTypeDramRetirementFailure as defined in nvml/nvml.h + EventTypeDramRetirementFailure = 2048 + // EventTypeNonFatalPoisonError as defined in nvml/nvml.h + EventTypeNonFatalPoisonError = 4096 + // EventTypeFatalPoisonError as defined in nvml/nvml.h + EventTypeFatalPoisonError = 8192 + // EventTypeGpuUnavailableError as defined in nvml/nvml.h + EventTypeGpuUnavailableError = 16384 + // EventTypeGpuRecoveryAction as defined in nvml/nvml.h + EventTypeGpuRecoveryAction = 32768 + // EventTypeAll as defined in nvml/nvml.h + EventTypeAll = 65439 + // SystemEventTypeGpuDriverUnbind as defined in nvml/nvml.h + SystemEventTypeGpuDriverUnbind = 1 + // SystemEventTypeGpuDriverBind as defined in nvml/nvml.h + SystemEventTypeGpuDriverBind = 2 + // SystemEventTypeCount as defined in nvml/nvml.h + SystemEventTypeCount = 2 + // ClocksEventReasonGpuIdle as defined in nvml/nvml.h + ClocksEventReasonGpuIdle = 1 + // ClocksEventReasonApplicationsClocksSetting as defined in nvml/nvml.h + ClocksEventReasonApplicationsClocksSetting = 2 + // ClocksThrottleReasonUserDefinedClocks as defined in nvml/nvml.h + ClocksThrottleReasonUserDefinedClocks = 2 + // ClocksEventReasonSwPowerCap as defined in nvml/nvml.h + ClocksEventReasonSwPowerCap = 4 + // ClocksThrottleReasonHwSlowdown as defined in nvml/nvml.h + ClocksThrottleReasonHwSlowdown = 8 + // ClocksEventReasonSyncBoost as defined in nvml/nvml.h + ClocksEventReasonSyncBoost = 16 + // ClocksEventReasonSwThermalSlowdown as defined in nvml/nvml.h + ClocksEventReasonSwThermalSlowdown = 32 + // ClocksThrottleReasonHwThermalSlowdown as defined in nvml/nvml.h + ClocksThrottleReasonHwThermalSlowdown = 64 + // ClocksThrottleReasonHwPowerBrakeSlowdown as defined in nvml/nvml.h + ClocksThrottleReasonHwPowerBrakeSlowdown = 128 + // ClocksEventReasonDisplayClockSetting as defined in nvml/nvml.h + ClocksEventReasonDisplayClockSetting = 256 + // ClocksEventReasonNone as defined in nvml/nvml.h + ClocksEventReasonNone = 0 + // ClocksEventReasonAll as defined in nvml/nvml.h + ClocksEventReasonAll = 511 + // ClocksThrottleReasonGpuIdle as defined in nvml/nvml.h + ClocksThrottleReasonGpuIdle = 1 + // ClocksThrottleReasonApplicationsClocksSetting as defined in nvml/nvml.h + ClocksThrottleReasonApplicationsClocksSetting = 2 + // ClocksThrottleReasonSyncBoost as defined in nvml/nvml.h + ClocksThrottleReasonSyncBoost = 16 + // ClocksThrottleReasonSwPowerCap as defined in nvml/nvml.h + ClocksThrottleReasonSwPowerCap = 4 + // ClocksThrottleReasonSwThermalSlowdown as defined in nvml/nvml.h + ClocksThrottleReasonSwThermalSlowdown = 32 + // ClocksThrottleReasonDisplayClockSetting as defined in nvml/nvml.h + ClocksThrottleReasonDisplayClockSetting = 256 + // ClocksThrottleReasonNone as defined in nvml/nvml.h + ClocksThrottleReasonNone = 0 + // ClocksThrottleReasonAll as defined in nvml/nvml.h + ClocksThrottleReasonAll = 511 + // NVFBC_SESSION_FLAG_DIFFMAP_ENABLED as defined in nvml/nvml.h + NVFBC_SESSION_FLAG_DIFFMAP_ENABLED = 1 + // NVFBC_SESSION_FLAG_CLASSIFICATIONMAP_ENABLED as defined in nvml/nvml.h + NVFBC_SESSION_FLAG_CLASSIFICATIONMAP_ENABLED = 2 + // NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_NO_WAIT as defined in nvml/nvml.h + NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_NO_WAIT = 4 + // NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_INFINITE as defined in nvml/nvml.h + NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_INFINITE = 8 + // NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_TIMEOUT as defined in nvml/nvml.h + NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_TIMEOUT = 16 + // CC_SYSTEM_CPU_CAPS_NONE as defined in nvml/nvml.h + CC_SYSTEM_CPU_CAPS_NONE = 0 + // CC_SYSTEM_CPU_CAPS_AMD_SEV as defined in nvml/nvml.h + CC_SYSTEM_CPU_CAPS_AMD_SEV = 1 + // CC_SYSTEM_CPU_CAPS_INTEL_TDX as defined in nvml/nvml.h + CC_SYSTEM_CPU_CAPS_INTEL_TDX = 2 + // CC_SYSTEM_CPU_CAPS_AMD_SEV_SNP as defined in nvml/nvml.h + CC_SYSTEM_CPU_CAPS_AMD_SEV_SNP = 3 + // CC_SYSTEM_CPU_CAPS_AMD_SNP_VTOM as defined in nvml/nvml.h + CC_SYSTEM_CPU_CAPS_AMD_SNP_VTOM = 4 + // CC_SYSTEM_GPUS_CC_NOT_CAPABLE as defined in nvml/nvml.h + CC_SYSTEM_GPUS_CC_NOT_CAPABLE = 0 + // CC_SYSTEM_GPUS_CC_CAPABLE as defined in nvml/nvml.h + CC_SYSTEM_GPUS_CC_CAPABLE = 1 + // CC_SYSTEM_DEVTOOLS_MODE_OFF as defined in nvml/nvml.h + CC_SYSTEM_DEVTOOLS_MODE_OFF = 0 + // CC_SYSTEM_DEVTOOLS_MODE_ON as defined in nvml/nvml.h + CC_SYSTEM_DEVTOOLS_MODE_ON = 1 + // CC_SYSTEM_ENVIRONMENT_UNAVAILABLE as defined in nvml/nvml.h + CC_SYSTEM_ENVIRONMENT_UNAVAILABLE = 0 + // CC_SYSTEM_ENVIRONMENT_SIM as defined in nvml/nvml.h + CC_SYSTEM_ENVIRONMENT_SIM = 1 + // CC_SYSTEM_ENVIRONMENT_PROD as defined in nvml/nvml.h + CC_SYSTEM_ENVIRONMENT_PROD = 2 + // CC_SYSTEM_FEATURE_DISABLED as defined in nvml/nvml.h + CC_SYSTEM_FEATURE_DISABLED = 0 + // CC_SYSTEM_FEATURE_ENABLED as defined in nvml/nvml.h + CC_SYSTEM_FEATURE_ENABLED = 1 + // CC_SYSTEM_MULTIGPU_NONE as defined in nvml/nvml.h + CC_SYSTEM_MULTIGPU_NONE = 0 + // CC_SYSTEM_MULTIGPU_PROTECTED_PCIE as defined in nvml/nvml.h + CC_SYSTEM_MULTIGPU_PROTECTED_PCIE = 1 + // CC_SYSTEM_MULTIGPU_NVLE as defined in nvml/nvml.h + CC_SYSTEM_MULTIGPU_NVLE = 2 + // CC_ACCEPTING_CLIENT_REQUESTS_FALSE as defined in nvml/nvml.h + CC_ACCEPTING_CLIENT_REQUESTS_FALSE = 0 + // CC_ACCEPTING_CLIENT_REQUESTS_TRUE as defined in nvml/nvml.h + CC_ACCEPTING_CLIENT_REQUESTS_TRUE = 1 + // GPU_CERT_CHAIN_SIZE as defined in nvml/nvml.h + GPU_CERT_CHAIN_SIZE = 4096 + // GPU_ATTESTATION_CERT_CHAIN_SIZE as defined in nvml/nvml.h + GPU_ATTESTATION_CERT_CHAIN_SIZE = 5120 + // CC_GPU_CEC_NONCE_SIZE as defined in nvml/nvml.h + CC_GPU_CEC_NONCE_SIZE = 32 + // CC_GPU_ATTESTATION_REPORT_SIZE as defined in nvml/nvml.h + CC_GPU_ATTESTATION_REPORT_SIZE = 8192 + // CC_GPU_CEC_ATTESTATION_REPORT_SIZE as defined in nvml/nvml.h + CC_GPU_CEC_ATTESTATION_REPORT_SIZE = 4096 + // CC_CEC_ATTESTATION_REPORT_NOT_PRESENT as defined in nvml/nvml.h + CC_CEC_ATTESTATION_REPORT_NOT_PRESENT = 0 + // CC_CEC_ATTESTATION_REPORT_PRESENT as defined in nvml/nvml.h + CC_CEC_ATTESTATION_REPORT_PRESENT = 1 + // CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MIN as defined in nvml/nvml.h + CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MIN = 50 + // CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MAX as defined in nvml/nvml.h + CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MAX = 65 + // GPU_FABRIC_UUID_LEN as defined in nvml/nvml.h + GPU_FABRIC_UUID_LEN = 16 + // GPU_FABRIC_STATE_NOT_SUPPORTED as defined in nvml/nvml.h + GPU_FABRIC_STATE_NOT_SUPPORTED = 0 + // GPU_FABRIC_STATE_NOT_STARTED as defined in nvml/nvml.h + GPU_FABRIC_STATE_NOT_STARTED = 1 + // GPU_FABRIC_STATE_IN_PROGRESS as defined in nvml/nvml.h + GPU_FABRIC_STATE_IN_PROGRESS = 2 + // GPU_FABRIC_STATE_COMPLETED as defined in nvml/nvml.h + GPU_FABRIC_STATE_COMPLETED = 3 + // GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_NOT_SUPPORTED as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_NOT_SUPPORTED = 0 + // GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_TRUE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_TRUE = 1 + // GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_FALSE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_FALSE = 2 + // GPU_FABRIC_HEALTH_MASK_SHIFT_DEGRADED_BW as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_SHIFT_DEGRADED_BW = 0 + // GPU_FABRIC_HEALTH_MASK_WIDTH_DEGRADED_BW as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_WIDTH_DEGRADED_BW = 3 + // GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_NOT_SUPPORTED as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_NOT_SUPPORTED = 0 + // GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_TRUE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_TRUE = 1 + // GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_FALSE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_FALSE = 2 + // GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_RECOVERY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_RECOVERY = 2 + // GPU_FABRIC_HEALTH_MASK_WIDTH_ROUTE_RECOVERY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_WIDTH_ROUTE_RECOVERY = 3 + // GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_NOT_SUPPORTED as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_NOT_SUPPORTED = 0 + // GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_TRUE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_TRUE = 1 + // GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_FALSE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_FALSE = 2 + // GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_UNHEALTHY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_UNHEALTHY = 4 + // GPU_FABRIC_HEALTH_MASK_WIDTH_ROUTE_UNHEALTHY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_WIDTH_ROUTE_UNHEALTHY = 3 + // GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_NOT_SUPPORTED as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_NOT_SUPPORTED = 0 + // GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_TRUE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_TRUE = 1 + // GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_FALSE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_FALSE = 2 + // GPU_FABRIC_HEALTH_MASK_SHIFT_ACCESS_TIMEOUT_RECOVERY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_SHIFT_ACCESS_TIMEOUT_RECOVERY = 6 + // GPU_FABRIC_HEALTH_MASK_WIDTH_ACCESS_TIMEOUT_RECOVERY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_WIDTH_ACCESS_TIMEOUT_RECOVERY = 3 + // GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NOT_SUPPORTED as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NOT_SUPPORTED = 0 + // GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NONE as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NONE = 1 + // GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INCORRECT_SYSGUID as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INCORRECT_SYSGUID = 2 + // GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INCORRECT_CHASSIS_SN as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INCORRECT_CHASSIS_SN = 3 + // GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NO_PARTITION as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NO_PARTITION = 4 + // GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INSUFFICIENT_NVLINKS as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INSUFFICIENT_NVLINKS = 5 + // GPU_FABRIC_HEALTH_MASK_SHIFT_INCORRECT_CONFIGURATION as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_SHIFT_INCORRECT_CONFIGURATION = 8 + // GPU_FABRIC_HEALTH_MASK_WIDTH_INCORRECT_CONFIGURATION as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_MASK_WIDTH_INCORRECT_CONFIGURATION = 15 + // GPU_FABRIC_HEALTH_SUMMARY_NOT_SUPPORTED as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_SUMMARY_NOT_SUPPORTED = 0 + // GPU_FABRIC_HEALTH_SUMMARY_HEALTHY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_SUMMARY_HEALTHY = 1 + // GPU_FABRIC_HEALTH_SUMMARY_UNHEALTHY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_SUMMARY_UNHEALTHY = 2 + // GPU_FABRIC_HEALTH_SUMMARY_LIMITED_CAPACITY as defined in nvml/nvml.h + GPU_FABRIC_HEALTH_SUMMARY_LIMITED_CAPACITY = 3 + // INIT_FLAG_NO_GPUS as defined in nvml/nvml.h + INIT_FLAG_NO_GPUS = 1 + // INIT_FLAG_NO_ATTACH as defined in nvml/nvml.h + INIT_FLAG_NO_ATTACH = 2 + // DEVICE_INFOROM_VERSION_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_INFOROM_VERSION_BUFFER_SIZE = 16 + // DEVICE_UUID_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_UUID_BUFFER_SIZE = 80 + // DEVICE_UUID_V2_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_UUID_V2_BUFFER_SIZE = 96 + // DEVICE_PART_NUMBER_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_PART_NUMBER_BUFFER_SIZE = 80 + // SYSTEM_DRIVER_VERSION_BUFFER_SIZE as defined in nvml/nvml.h + SYSTEM_DRIVER_VERSION_BUFFER_SIZE = 80 + // SYSTEM_NVML_VERSION_BUFFER_SIZE as defined in nvml/nvml.h + SYSTEM_NVML_VERSION_BUFFER_SIZE = 80 + // DEVICE_NAME_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_NAME_BUFFER_SIZE = 64 + // DEVICE_NAME_V2_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_NAME_V2_BUFFER_SIZE = 96 + // DEVICE_SERIAL_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_SERIAL_BUFFER_SIZE = 30 + // DEVICE_VBIOS_VERSION_BUFFER_SIZE as defined in nvml/nvml.h + DEVICE_VBIOS_VERSION_BUFFER_SIZE = 32 + // AFFINITY_SCOPE_NODE as defined in nvml/nvml.h + AFFINITY_SCOPE_NODE = 0 + // AFFINITY_SCOPE_SOCKET as defined in nvml/nvml.h + AFFINITY_SCOPE_SOCKET = 1 + // NVLINK_BER_MANTISSA_SHIFT as defined in nvml/nvml.h + NVLINK_BER_MANTISSA_SHIFT = 8 + // NVLINK_BER_MANTISSA_WIDTH as defined in nvml/nvml.h + NVLINK_BER_MANTISSA_WIDTH = 15 + // NVLINK_BER_EXP_SHIFT as defined in nvml/nvml.h + NVLINK_BER_EXP_SHIFT = 0 + // NVLINK_BER_EXP_WIDTH as defined in nvml/nvml.h + NVLINK_BER_EXP_WIDTH = 255 + // NVLINK_STATE_INACTIVE as defined in nvml/nvml.h + NVLINK_STATE_INACTIVE = 0 + // NVLINK_STATE_ACTIVE as defined in nvml/nvml.h + NVLINK_STATE_ACTIVE = 1 + // NVLINK_STATE_SLEEP as defined in nvml/nvml.h + NVLINK_STATE_SLEEP = 2 + // NVLINK_TOTAL_SUPPORTED_BW_MODES as defined in nvml/nvml.h + NVLINK_TOTAL_SUPPORTED_BW_MODES = 23 + // NVLINK_FIRMWARE_UCODE_TYPE_MSE as defined in nvml/nvml.h + NVLINK_FIRMWARE_UCODE_TYPE_MSE = 1 + // NVLINK_FIRMWARE_UCODE_TYPE_NETIR as defined in nvml/nvml.h + NVLINK_FIRMWARE_UCODE_TYPE_NETIR = 2 + // NVLINK_FIRMWARE_UCODE_TYPE_NETIR_UPHY as defined in nvml/nvml.h + NVLINK_FIRMWARE_UCODE_TYPE_NETIR_UPHY = 3 + // NVLINK_FIRMWARE_UCODE_TYPE_NETIR_CLN as defined in nvml/nvml.h + NVLINK_FIRMWARE_UCODE_TYPE_NETIR_CLN = 4 + // NVLINK_FIRMWARE_UCODE_TYPE_NETIR_DLN as defined in nvml/nvml.h + NVLINK_FIRMWARE_UCODE_TYPE_NETIR_DLN = 5 + // NVLINK_FIRMWARE_VERSION_LENGTH as defined in nvml/nvml.h + NVLINK_FIRMWARE_VERSION_LENGTH = 100 + // PRM_DATA_MAX_SIZE as defined in nvml/nvml.h + PRM_DATA_MAX_SIZE = 496 + // DEVICE_MIG_DISABLE as defined in nvml/nvml.h + DEVICE_MIG_DISABLE = 0 + // DEVICE_MIG_ENABLE as defined in nvml/nvml.h + DEVICE_MIG_ENABLE = 1 + // GPU_INSTANCE_PROFILE_1_SLICE as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_1_SLICE = 0 + // GPU_INSTANCE_PROFILE_2_SLICE as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_2_SLICE = 1 + // GPU_INSTANCE_PROFILE_3_SLICE as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_3_SLICE = 2 + // GPU_INSTANCE_PROFILE_4_SLICE as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_4_SLICE = 3 + // GPU_INSTANCE_PROFILE_7_SLICE as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_7_SLICE = 4 + // GPU_INSTANCE_PROFILE_8_SLICE as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_8_SLICE = 5 + // GPU_INSTANCE_PROFILE_6_SLICE as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_6_SLICE = 6 + // GPU_INSTANCE_PROFILE_1_SLICE_REV1 as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_1_SLICE_REV1 = 7 + // GPU_INSTANCE_PROFILE_2_SLICE_REV1 as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_2_SLICE_REV1 = 8 + // GPU_INSTANCE_PROFILE_1_SLICE_REV2 as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_1_SLICE_REV2 = 9 + // GPU_INSTANCE_PROFILE_1_SLICE_GFX as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_1_SLICE_GFX = 10 + // GPU_INSTANCE_PROFILE_2_SLICE_GFX as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_2_SLICE_GFX = 11 + // GPU_INSTANCE_PROFILE_4_SLICE_GFX as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_4_SLICE_GFX = 12 + // GPU_INSTANCE_PROFILE_1_SLICE_NO_ME as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_1_SLICE_NO_ME = 13 + // GPU_INSTANCE_PROFILE_2_SLICE_NO_ME as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_2_SLICE_NO_ME = 14 + // GPU_INSTANCE_PROFILE_1_SLICE_ALL_ME as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_1_SLICE_ALL_ME = 15 + // GPU_INSTANCE_PROFILE_2_SLICE_ALL_ME as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_2_SLICE_ALL_ME = 16 + // GPU_INSTANCE_PROFILE_COUNT as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_COUNT = 17 + // GPU_INSTANCE_PROFILE_CAPS_P2P as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_CAPS_P2P = 1 + // GPU_INTSTANCE_PROFILE_CAPS_P2P as defined in nvml/nvml.h + GPU_INTSTANCE_PROFILE_CAPS_P2P = 1 + // GPU_INSTANCE_PROFILE_CAPS_GFX as defined in nvml/nvml.h + GPU_INSTANCE_PROFILE_CAPS_GFX = 2 + // COMPUTE_INSTANCE_PROFILE_CAPS_GFX as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_CAPS_GFX = 1 + // COMPUTE_INSTANCE_PROFILE_1_SLICE as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_1_SLICE = 0 + // COMPUTE_INSTANCE_PROFILE_2_SLICE as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_2_SLICE = 1 + // COMPUTE_INSTANCE_PROFILE_3_SLICE as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_3_SLICE = 2 + // COMPUTE_INSTANCE_PROFILE_4_SLICE as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_4_SLICE = 3 + // COMPUTE_INSTANCE_PROFILE_7_SLICE as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_7_SLICE = 4 + // COMPUTE_INSTANCE_PROFILE_8_SLICE as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_8_SLICE = 5 + // COMPUTE_INSTANCE_PROFILE_6_SLICE as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_6_SLICE = 6 + // COMPUTE_INSTANCE_PROFILE_1_SLICE_REV1 as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_1_SLICE_REV1 = 7 + // COMPUTE_INSTANCE_PROFILE_COUNT as defined in nvml/nvml.h + COMPUTE_INSTANCE_PROFILE_COUNT = 8 + // COMPUTE_INSTANCE_ENGINE_PROFILE_SHARED as defined in nvml/nvml.h + COMPUTE_INSTANCE_ENGINE_PROFILE_SHARED = 0 + // COMPUTE_INSTANCE_ENGINE_PROFILE_COUNT as defined in nvml/nvml.h + COMPUTE_INSTANCE_ENGINE_PROFILE_COUNT = 1 + // GPM_METRICS_GET_VERSION as defined in nvml/nvml.h + GPM_METRICS_GET_VERSION = 1 + // GPM_SUPPORT_VERSION as defined in nvml/nvml.h + GPM_SUPPORT_VERSION = 1 + // DEV_CAP_EGM as defined in nvml/nvml.h + DEV_CAP_EGM = 1 + // WORKLOAD_POWER_MAX_PROFILES as defined in nvml/nvml.h + WORKLOAD_POWER_MAX_PROFILES = 255 + // POWER_SMOOTHING_MAX_NUM_PROFILES as defined in nvml/nvml.h + POWER_SMOOTHING_MAX_NUM_PROFILES = 5 + // POWER_SMOOTHING_NUM_PROFILE_PARAMS as defined in nvml/nvml.h + POWER_SMOOTHING_NUM_PROFILE_PARAMS = 4 + // POWER_SMOOTHING_ADMIN_OVERRIDE_NOT_SET as defined in nvml/nvml.h + POWER_SMOOTHING_ADMIN_OVERRIDE_NOT_SET = 4294967295 + // POWER_SMOOTHING_PROFILE_PARAM_PERCENT_TMP_FLOOR as defined in nvml/nvml.h + POWER_SMOOTHING_PROFILE_PARAM_PERCENT_TMP_FLOOR = 0 + // POWER_SMOOTHING_PROFILE_PARAM_RAMP_UP_RATE as defined in nvml/nvml.h + POWER_SMOOTHING_PROFILE_PARAM_RAMP_UP_RATE = 1 + // POWER_SMOOTHING_PROFILE_PARAM_RAMP_DOWN_RATE as defined in nvml/nvml.h + POWER_SMOOTHING_PROFILE_PARAM_RAMP_DOWN_RATE = 2 + // POWER_SMOOTHING_PROFILE_PARAM_RAMP_DOWN_HYSTERESIS as defined in nvml/nvml.h + POWER_SMOOTHING_PROFILE_PARAM_RAMP_DOWN_HYSTERESIS = 3 +) + +// BridgeChipType as declared in nvml/nvml.h +type BridgeChipType int32 + +// BridgeChipType enumeration from nvml/nvml.h +const ( + BRIDGE_CHIP_PLX BridgeChipType = iota + BRIDGE_CHIP_BRO4 BridgeChipType = 1 +) + +// NvLinkUtilizationCountUnits as declared in nvml/nvml.h +type NvLinkUtilizationCountUnits int32 + +// NvLinkUtilizationCountUnits enumeration from nvml/nvml.h +const ( + NVLINK_COUNTER_UNIT_CYCLES NvLinkUtilizationCountUnits = iota + NVLINK_COUNTER_UNIT_PACKETS NvLinkUtilizationCountUnits = 1 + NVLINK_COUNTER_UNIT_BYTES NvLinkUtilizationCountUnits = 2 + NVLINK_COUNTER_UNIT_RESERVED NvLinkUtilizationCountUnits = 3 + NVLINK_COUNTER_UNIT_COUNT NvLinkUtilizationCountUnits = 4 +) + +// NvLinkUtilizationCountPktTypes as declared in nvml/nvml.h +type NvLinkUtilizationCountPktTypes int32 + +// NvLinkUtilizationCountPktTypes enumeration from nvml/nvml.h +const ( + NVLINK_COUNTER_PKTFILTER_NOP NvLinkUtilizationCountPktTypes = 1 + NVLINK_COUNTER_PKTFILTER_READ NvLinkUtilizationCountPktTypes = 2 + NVLINK_COUNTER_PKTFILTER_WRITE NvLinkUtilizationCountPktTypes = 4 + NVLINK_COUNTER_PKTFILTER_RATOM NvLinkUtilizationCountPktTypes = 8 + NVLINK_COUNTER_PKTFILTER_NRATOM NvLinkUtilizationCountPktTypes = 16 + NVLINK_COUNTER_PKTFILTER_FLUSH NvLinkUtilizationCountPktTypes = 32 + NVLINK_COUNTER_PKTFILTER_RESPDATA NvLinkUtilizationCountPktTypes = 64 + NVLINK_COUNTER_PKTFILTER_RESPNODATA NvLinkUtilizationCountPktTypes = 128 + NVLINK_COUNTER_PKTFILTER_ALL NvLinkUtilizationCountPktTypes = 255 +) + +// NvLinkCapability as declared in nvml/nvml.h +type NvLinkCapability int32 + +// NvLinkCapability enumeration from nvml/nvml.h +const ( + NVLINK_CAP_P2P_SUPPORTED NvLinkCapability = iota + NVLINK_CAP_SYSMEM_ACCESS NvLinkCapability = 1 + NVLINK_CAP_P2P_ATOMICS NvLinkCapability = 2 + NVLINK_CAP_SYSMEM_ATOMICS NvLinkCapability = 3 + NVLINK_CAP_SLI_BRIDGE NvLinkCapability = 4 + NVLINK_CAP_VALID NvLinkCapability = 5 + NVLINK_CAP_COUNT NvLinkCapability = 6 +) + +// NvLinkErrorCounter as declared in nvml/nvml.h +type NvLinkErrorCounter int32 + +// NvLinkErrorCounter enumeration from nvml/nvml.h +const ( + NVLINK_ERROR_DL_REPLAY NvLinkErrorCounter = iota + NVLINK_ERROR_DL_RECOVERY NvLinkErrorCounter = 1 + NVLINK_ERROR_DL_CRC_FLIT NvLinkErrorCounter = 2 + NVLINK_ERROR_DL_CRC_DATA NvLinkErrorCounter = 3 + NVLINK_ERROR_DL_ECC_DATA NvLinkErrorCounter = 4 + NVLINK_ERROR_COUNT NvLinkErrorCounter = 5 +) + +// IntNvLinkDeviceType as declared in nvml/nvml.h +type IntNvLinkDeviceType int32 + +// IntNvLinkDeviceType enumeration from nvml/nvml.h +const ( + NVLINK_DEVICE_TYPE_GPU IntNvLinkDeviceType = iota + NVLINK_DEVICE_TYPE_IBMNPU IntNvLinkDeviceType = 1 + NVLINK_DEVICE_TYPE_SWITCH IntNvLinkDeviceType = 2 + NVLINK_DEVICE_TYPE_UNKNOWN IntNvLinkDeviceType = 255 +) + +// GpuTopologyLevel as declared in nvml/nvml.h +type GpuTopologyLevel int32 + +// GpuTopologyLevel enumeration from nvml/nvml.h +const ( + TOPOLOGY_INTERNAL GpuTopologyLevel = iota + TOPOLOGY_SINGLE GpuTopologyLevel = 10 + TOPOLOGY_MULTIPLE GpuTopologyLevel = 20 + TOPOLOGY_HOSTBRIDGE GpuTopologyLevel = 30 + TOPOLOGY_NODE GpuTopologyLevel = 40 + TOPOLOGY_SYSTEM GpuTopologyLevel = 50 +) + +// GpuP2PStatus as declared in nvml/nvml.h +type GpuP2PStatus int32 + +// GpuP2PStatus enumeration from nvml/nvml.h +const ( + P2P_STATUS_OK GpuP2PStatus = iota + P2P_STATUS_CHIPSET_NOT_SUPPORED GpuP2PStatus = 1 + P2P_STATUS_CHIPSET_NOT_SUPPORTED GpuP2PStatus = 1 + P2P_STATUS_GPU_NOT_SUPPORTED GpuP2PStatus = 2 + P2P_STATUS_IOH_TOPOLOGY_NOT_SUPPORTED GpuP2PStatus = 3 + P2P_STATUS_DISABLED_BY_REGKEY GpuP2PStatus = 4 + P2P_STATUS_NOT_SUPPORTED GpuP2PStatus = 5 + P2P_STATUS_UNKNOWN GpuP2PStatus = 6 +) + +// GpuP2PCapsIndex as declared in nvml/nvml.h +type GpuP2PCapsIndex int32 + +// GpuP2PCapsIndex enumeration from nvml/nvml.h +const ( + P2P_CAPS_INDEX_READ GpuP2PCapsIndex = iota + P2P_CAPS_INDEX_WRITE GpuP2PCapsIndex = 1 + P2P_CAPS_INDEX_NVLINK GpuP2PCapsIndex = 2 + P2P_CAPS_INDEX_ATOMICS GpuP2PCapsIndex = 3 + P2P_CAPS_INDEX_PCI GpuP2PCapsIndex = 4 + P2P_CAPS_INDEX_PROP GpuP2PCapsIndex = 4 + P2P_CAPS_INDEX_UNKNOWN GpuP2PCapsIndex = 5 +) + +// SamplingType as declared in nvml/nvml.h +type SamplingType int32 + +// SamplingType enumeration from nvml/nvml.h +const ( + TOTAL_POWER_SAMPLES SamplingType = iota + GPU_UTILIZATION_SAMPLES SamplingType = 1 + MEMORY_UTILIZATION_SAMPLES SamplingType = 2 + ENC_UTILIZATION_SAMPLES SamplingType = 3 + DEC_UTILIZATION_SAMPLES SamplingType = 4 + PROCESSOR_CLK_SAMPLES SamplingType = 5 + MEMORY_CLK_SAMPLES SamplingType = 6 + MODULE_POWER_SAMPLES SamplingType = 7 + JPG_UTILIZATION_SAMPLES SamplingType = 8 + OFA_UTILIZATION_SAMPLES SamplingType = 9 + SAMPLINGTYPE_COUNT SamplingType = 10 +) + +// PcieUtilCounter as declared in nvml/nvml.h +type PcieUtilCounter int32 + +// PcieUtilCounter enumeration from nvml/nvml.h +const ( + PCIE_UTIL_TX_BYTES PcieUtilCounter = iota + PCIE_UTIL_RX_BYTES PcieUtilCounter = 1 + PCIE_UTIL_COUNT PcieUtilCounter = 2 +) + +// ValueType as declared in nvml/nvml.h +type ValueType int32 + +// ValueType enumeration from nvml/nvml.h +const ( + VALUE_TYPE_DOUBLE ValueType = iota + VALUE_TYPE_UNSIGNED_INT ValueType = 1 + VALUE_TYPE_UNSIGNED_LONG ValueType = 2 + VALUE_TYPE_UNSIGNED_LONG_LONG ValueType = 3 + VALUE_TYPE_SIGNED_LONG_LONG ValueType = 4 + VALUE_TYPE_SIGNED_INT ValueType = 5 + VALUE_TYPE_UNSIGNED_SHORT ValueType = 6 + VALUE_TYPE_COUNT ValueType = 7 +) + +// PerfPolicyType as declared in nvml/nvml.h +type PerfPolicyType int32 + +// PerfPolicyType enumeration from nvml/nvml.h +const ( + PERF_POLICY_POWER PerfPolicyType = iota + PERF_POLICY_THERMAL PerfPolicyType = 1 + PERF_POLICY_SYNC_BOOST PerfPolicyType = 2 + PERF_POLICY_BOARD_LIMIT PerfPolicyType = 3 + PERF_POLICY_LOW_UTILIZATION PerfPolicyType = 4 + PERF_POLICY_RELIABILITY PerfPolicyType = 5 + PERF_POLICY_TOTAL_APP_CLOCKS PerfPolicyType = 10 + PERF_POLICY_TOTAL_BASE_CLOCKS PerfPolicyType = 11 + PERF_POLICY_COUNT PerfPolicyType = 12 +) + +// CoolerControl as declared in nvml/nvml.h +type CoolerControl int32 + +// CoolerControl enumeration from nvml/nvml.h +const ( + THERMAL_COOLER_SIGNAL_NONE CoolerControl = iota + THERMAL_COOLER_SIGNAL_TOGGLE CoolerControl = 1 + THERMAL_COOLER_SIGNAL_VARIABLE CoolerControl = 2 + THERMAL_COOLER_SIGNAL_COUNT CoolerControl = 3 +) + +// CoolerTarget as declared in nvml/nvml.h +type CoolerTarget int32 + +// CoolerTarget enumeration from nvml/nvml.h +const ( + THERMAL_COOLER_TARGET_NONE CoolerTarget = 1 + THERMAL_COOLER_TARGET_GPU CoolerTarget = 2 + THERMAL_COOLER_TARGET_MEMORY CoolerTarget = 4 + THERMAL_COOLER_TARGET_POWER_SUPPLY CoolerTarget = 8 + THERMAL_COOLER_TARGET_GPU_RELATED CoolerTarget = 14 +) + +// EnableState as declared in nvml/nvml.h +type EnableState int32 + +// EnableState enumeration from nvml/nvml.h +const ( + FEATURE_DISABLED EnableState = iota + FEATURE_ENABLED EnableState = 1 +) + +// BrandType as declared in nvml/nvml.h +type BrandType int32 + +// BrandType enumeration from nvml/nvml.h +const ( + BRAND_UNKNOWN BrandType = iota + BRAND_QUADRO BrandType = 1 + BRAND_TESLA BrandType = 2 + BRAND_NVS BrandType = 3 + BRAND_GRID BrandType = 4 + BRAND_GEFORCE BrandType = 5 + BRAND_TITAN BrandType = 6 + BRAND_NVIDIA_VAPPS BrandType = 7 + BRAND_NVIDIA_VPC BrandType = 8 + BRAND_NVIDIA_VCS BrandType = 9 + BRAND_NVIDIA_VWS BrandType = 10 + BRAND_NVIDIA_CLOUD_GAMING BrandType = 11 + BRAND_NVIDIA_VGAMING BrandType = 11 + BRAND_QUADRO_RTX BrandType = 12 + BRAND_NVIDIA_RTX BrandType = 13 + BRAND_NVIDIA BrandType = 14 + BRAND_GEFORCE_RTX BrandType = 15 + BRAND_TITAN_RTX BrandType = 16 + BRAND_COUNT BrandType = 18 +) + +// TemperatureThresholds as declared in nvml/nvml.h +type TemperatureThresholds int32 + +// TemperatureThresholds enumeration from nvml/nvml.h +const ( + TEMPERATURE_THRESHOLD_SHUTDOWN TemperatureThresholds = iota + TEMPERATURE_THRESHOLD_SLOWDOWN TemperatureThresholds = 1 + TEMPERATURE_THRESHOLD_MEM_MAX TemperatureThresholds = 2 + TEMPERATURE_THRESHOLD_GPU_MAX TemperatureThresholds = 3 + TEMPERATURE_THRESHOLD_ACOUSTIC_MIN TemperatureThresholds = 4 + TEMPERATURE_THRESHOLD_ACOUSTIC_CURR TemperatureThresholds = 5 + TEMPERATURE_THRESHOLD_ACOUSTIC_MAX TemperatureThresholds = 6 + TEMPERATURE_THRESHOLD_GPS_CURR TemperatureThresholds = 7 + TEMPERATURE_THRESHOLD_COUNT TemperatureThresholds = 8 +) + +// TemperatureSensors as declared in nvml/nvml.h +type TemperatureSensors int32 + +// TemperatureSensors enumeration from nvml/nvml.h +const ( + TEMPERATURE_GPU TemperatureSensors = iota + TEMPERATURE_COUNT TemperatureSensors = 1 +) + +// ComputeMode as declared in nvml/nvml.h +type ComputeMode int32 + +// ComputeMode enumeration from nvml/nvml.h +const ( + COMPUTEMODE_DEFAULT ComputeMode = iota + COMPUTEMODE_EXCLUSIVE_THREAD ComputeMode = 1 + COMPUTEMODE_PROHIBITED ComputeMode = 2 + COMPUTEMODE_EXCLUSIVE_PROCESS ComputeMode = 3 + COMPUTEMODE_COUNT ComputeMode = 4 +) + +// MemoryErrorType as declared in nvml/nvml.h +type MemoryErrorType int32 + +// MemoryErrorType enumeration from nvml/nvml.h +const ( + MEMORY_ERROR_TYPE_CORRECTED MemoryErrorType = iota + MEMORY_ERROR_TYPE_UNCORRECTED MemoryErrorType = 1 + MEMORY_ERROR_TYPE_COUNT MemoryErrorType = 2 +) + +// NvlinkVersion as declared in nvml/nvml.h +type NvlinkVersion int32 + +// NvlinkVersion enumeration from nvml/nvml.h +const ( + NVLINK_VERSION_INVALID NvlinkVersion = iota + NVLINK_VERSION_1_0 NvlinkVersion = 1 + NVLINK_VERSION_2_0 NvlinkVersion = 2 + NVLINK_VERSION_2_2 NvlinkVersion = 3 + NVLINK_VERSION_3_0 NvlinkVersion = 4 + NVLINK_VERSION_3_1 NvlinkVersion = 5 + NVLINK_VERSION_4_0 NvlinkVersion = 6 + NVLINK_VERSION_5_0 NvlinkVersion = 7 +) + +// EccCounterType as declared in nvml/nvml.h +type EccCounterType int32 + +// EccCounterType enumeration from nvml/nvml.h +const ( + VOLATILE_ECC EccCounterType = iota + AGGREGATE_ECC EccCounterType = 1 + ECC_COUNTER_TYPE_COUNT EccCounterType = 2 +) + +// ClockType as declared in nvml/nvml.h +type ClockType int32 + +// ClockType enumeration from nvml/nvml.h +const ( + CLOCK_GRAPHICS ClockType = iota + CLOCK_SM ClockType = 1 + CLOCK_MEM ClockType = 2 + CLOCK_VIDEO ClockType = 3 + CLOCK_COUNT ClockType = 4 +) + +// ClockId as declared in nvml/nvml.h +type ClockId int32 + +// ClockId enumeration from nvml/nvml.h +const ( + CLOCK_ID_CURRENT ClockId = iota + CLOCK_ID_APP_CLOCK_TARGET ClockId = 1 + CLOCK_ID_APP_CLOCK_DEFAULT ClockId = 2 + CLOCK_ID_CUSTOMER_BOOST_MAX ClockId = 3 + CLOCK_ID_COUNT ClockId = 4 +) + +// DriverModel as declared in nvml/nvml.h +type DriverModel int32 + +// DriverModel enumeration from nvml/nvml.h +const ( + DRIVER_WDDM DriverModel = iota + DRIVER_WDM DriverModel = 1 + DRIVER_MCDM DriverModel = 2 +) + +// Pstates as declared in nvml/nvml.h +type Pstates int32 + +// Pstates enumeration from nvml/nvml.h +const ( + PSTATE_0 Pstates = iota + PSTATE_1 Pstates = 1 + PSTATE_2 Pstates = 2 + PSTATE_3 Pstates = 3 + PSTATE_4 Pstates = 4 + PSTATE_5 Pstates = 5 + PSTATE_6 Pstates = 6 + PSTATE_7 Pstates = 7 + PSTATE_8 Pstates = 8 + PSTATE_9 Pstates = 9 + PSTATE_10 Pstates = 10 + PSTATE_11 Pstates = 11 + PSTATE_12 Pstates = 12 + PSTATE_13 Pstates = 13 + PSTATE_14 Pstates = 14 + PSTATE_15 Pstates = 15 + PSTATE_UNKNOWN Pstates = 32 +) + +// GpuOperationMode as declared in nvml/nvml.h +type GpuOperationMode int32 + +// GpuOperationMode enumeration from nvml/nvml.h +const ( + GOM_ALL_ON GpuOperationMode = iota + GOM_COMPUTE GpuOperationMode = 1 + GOM_LOW_DP GpuOperationMode = 2 +) + +// InforomObject as declared in nvml/nvml.h +type InforomObject int32 + +// InforomObject enumeration from nvml/nvml.h +const ( + INFOROM_OEM InforomObject = iota + INFOROM_ECC InforomObject = 1 + INFOROM_POWER InforomObject = 2 + INFOROM_DEN InforomObject = 3 + INFOROM_COUNT InforomObject = 4 +) + +// Return as declared in nvml/nvml.h +type Return int32 + +// Return enumeration from nvml/nvml.h +const ( + SUCCESS Return = iota + ERROR_UNINITIALIZED Return = 1 + ERROR_INVALID_ARGUMENT Return = 2 + ERROR_NOT_SUPPORTED Return = 3 + ERROR_NO_PERMISSION Return = 4 + ERROR_ALREADY_INITIALIZED Return = 5 + ERROR_NOT_FOUND Return = 6 + ERROR_INSUFFICIENT_SIZE Return = 7 + ERROR_INSUFFICIENT_POWER Return = 8 + ERROR_DRIVER_NOT_LOADED Return = 9 + ERROR_TIMEOUT Return = 10 + ERROR_IRQ_ISSUE Return = 11 + ERROR_LIBRARY_NOT_FOUND Return = 12 + ERROR_FUNCTION_NOT_FOUND Return = 13 + ERROR_CORRUPTED_INFOROM Return = 14 + ERROR_GPU_IS_LOST Return = 15 + ERROR_RESET_REQUIRED Return = 16 + ERROR_OPERATING_SYSTEM Return = 17 + ERROR_LIB_RM_VERSION_MISMATCH Return = 18 + ERROR_IN_USE Return = 19 + ERROR_MEMORY Return = 20 + ERROR_NO_DATA Return = 21 + ERROR_VGPU_ECC_NOT_SUPPORTED Return = 22 + ERROR_INSUFFICIENT_RESOURCES Return = 23 + ERROR_FREQ_NOT_SUPPORTED Return = 24 + ERROR_ARGUMENT_VERSION_MISMATCH Return = 25 + ERROR_DEPRECATED Return = 26 + ERROR_NOT_READY Return = 27 + ERROR_GPU_NOT_FOUND Return = 28 + ERROR_INVALID_STATE Return = 29 + ERROR_RESET_TYPE_NOT_SUPPORTED Return = 30 + ERROR_UNKNOWN Return = 999 +) + +// MemoryLocation as declared in nvml/nvml.h +type MemoryLocation int32 + +// MemoryLocation enumeration from nvml/nvml.h +const ( + MEMORY_LOCATION_L1_CACHE MemoryLocation = iota + MEMORY_LOCATION_L2_CACHE MemoryLocation = 1 + MEMORY_LOCATION_DRAM MemoryLocation = 2 + MEMORY_LOCATION_DEVICE_MEMORY MemoryLocation = 2 + MEMORY_LOCATION_REGISTER_FILE MemoryLocation = 3 + MEMORY_LOCATION_TEXTURE_MEMORY MemoryLocation = 4 + MEMORY_LOCATION_TEXTURE_SHM MemoryLocation = 5 + MEMORY_LOCATION_CBU MemoryLocation = 6 + MEMORY_LOCATION_SRAM MemoryLocation = 7 + MEMORY_LOCATION_COUNT MemoryLocation = 8 +) + +// PageRetirementCause as declared in nvml/nvml.h +type PageRetirementCause int32 + +// PageRetirementCause enumeration from nvml/nvml.h +const ( + PAGE_RETIREMENT_CAUSE_MULTIPLE_SINGLE_BIT_ECC_ERRORS PageRetirementCause = iota + PAGE_RETIREMENT_CAUSE_DOUBLE_BIT_ECC_ERROR PageRetirementCause = 1 + PAGE_RETIREMENT_CAUSE_COUNT PageRetirementCause = 2 +) + +// RestrictedAPI as declared in nvml/nvml.h +type RestrictedAPI int32 + +// RestrictedAPI enumeration from nvml/nvml.h +const ( + RESTRICTED_API_SET_APPLICATION_CLOCKS RestrictedAPI = iota + RESTRICTED_API_SET_AUTO_BOOSTED_CLOCKS RestrictedAPI = 1 + RESTRICTED_API_COUNT RestrictedAPI = 2 +) + +// GpuUtilizationDomainId as declared in nvml/nvml.h +type GpuUtilizationDomainId int32 + +// GpuUtilizationDomainId enumeration from nvml/nvml.h +const ( + GPU_UTILIZATION_DOMAIN_GPU GpuUtilizationDomainId = iota + GPU_UTILIZATION_DOMAIN_FB GpuUtilizationDomainId = 1 + GPU_UTILIZATION_DOMAIN_VID GpuUtilizationDomainId = 2 + GPU_UTILIZATION_DOMAIN_BUS GpuUtilizationDomainId = 3 +) + +// GpuVirtualizationMode as declared in nvml/nvml.h +type GpuVirtualizationMode int32 + +// GpuVirtualizationMode enumeration from nvml/nvml.h +const ( + GPU_VIRTUALIZATION_MODE_NONE GpuVirtualizationMode = iota + GPU_VIRTUALIZATION_MODE_PASSTHROUGH GpuVirtualizationMode = 1 + GPU_VIRTUALIZATION_MODE_VGPU GpuVirtualizationMode = 2 + GPU_VIRTUALIZATION_MODE_HOST_VGPU GpuVirtualizationMode = 3 + GPU_VIRTUALIZATION_MODE_HOST_VSGA GpuVirtualizationMode = 4 +) + +// HostVgpuMode as declared in nvml/nvml.h +type HostVgpuMode int32 + +// HostVgpuMode enumeration from nvml/nvml.h +const ( + HOST_VGPU_MODE_NON_SRIOV HostVgpuMode = iota + HOST_VGPU_MODE_SRIOV HostVgpuMode = 1 +) + +// VgpuVmIdType as declared in nvml/nvml.h +type VgpuVmIdType int32 + +// VgpuVmIdType enumeration from nvml/nvml.h +const ( + VGPU_VM_ID_DOMAIN_ID VgpuVmIdType = iota + VGPU_VM_ID_UUID VgpuVmIdType = 1 +) + +// VgpuGuestInfoState as declared in nvml/nvml.h +type VgpuGuestInfoState int32 + +// VgpuGuestInfoState enumeration from nvml/nvml.h +const ( + VGPU_INSTANCE_GUEST_INFO_STATE_UNINITIALIZED VgpuGuestInfoState = iota + VGPU_INSTANCE_GUEST_INFO_STATE_INITIALIZED VgpuGuestInfoState = 1 +) + +// VgpuCapability as declared in nvml/nvml.h +type VgpuCapability int32 + +// VgpuCapability enumeration from nvml/nvml.h +const ( + VGPU_CAP_NVLINK_P2P VgpuCapability = iota + VGPU_CAP_GPUDIRECT VgpuCapability = 1 + VGPU_CAP_MULTI_VGPU_EXCLUSIVE VgpuCapability = 2 + VGPU_CAP_EXCLUSIVE_TYPE VgpuCapability = 3 + VGPU_CAP_EXCLUSIVE_SIZE VgpuCapability = 4 + VGPU_CAP_COUNT VgpuCapability = 5 +) + +// VgpuDriverCapability as declared in nvml/nvml.h +type VgpuDriverCapability int32 + +// VgpuDriverCapability enumeration from nvml/nvml.h +const ( + VGPU_DRIVER_CAP_HETEROGENEOUS_MULTI_VGPU VgpuDriverCapability = iota + VGPU_DRIVER_CAP_WARM_UPDATE VgpuDriverCapability = 1 + VGPU_DRIVER_CAP_COUNT VgpuDriverCapability = 2 +) + +// DeviceVgpuCapability as declared in nvml/nvml.h +type DeviceVgpuCapability int32 + +// DeviceVgpuCapability enumeration from nvml/nvml.h +const ( + DEVICE_VGPU_CAP_FRACTIONAL_MULTI_VGPU DeviceVgpuCapability = iota + DEVICE_VGPU_CAP_HETEROGENEOUS_TIMESLICE_PROFILES DeviceVgpuCapability = 1 + DEVICE_VGPU_CAP_HETEROGENEOUS_TIMESLICE_SIZES DeviceVgpuCapability = 2 + DEVICE_VGPU_CAP_READ_DEVICE_BUFFER_BW DeviceVgpuCapability = 3 + DEVICE_VGPU_CAP_WRITE_DEVICE_BUFFER_BW DeviceVgpuCapability = 4 + DEVICE_VGPU_CAP_DEVICE_STREAMING DeviceVgpuCapability = 5 + DEVICE_VGPU_CAP_MINI_QUARTER_GPU DeviceVgpuCapability = 6 + DEVICE_VGPU_CAP_COMPUTE_MEDIA_ENGINE_GPU DeviceVgpuCapability = 7 + DEVICE_VGPU_CAP_WARM_UPDATE DeviceVgpuCapability = 8 + DEVICE_VGPU_CAP_HOMOGENEOUS_PLACEMENTS DeviceVgpuCapability = 9 + DEVICE_VGPU_CAP_MIG_TIMESLICING_SUPPORTED DeviceVgpuCapability = 10 + DEVICE_VGPU_CAP_MIG_TIMESLICING_ENABLED DeviceVgpuCapability = 11 + DEVICE_VGPU_CAP_COUNT DeviceVgpuCapability = 12 +) + +// DeviceGpuRecoveryAction as declared in nvml/nvml.h +type DeviceGpuRecoveryAction int32 + +// DeviceGpuRecoveryAction enumeration from nvml/nvml.h +const ( + GPU_RECOVERY_ACTION_NONE DeviceGpuRecoveryAction = iota + GPU_RECOVERY_ACTION_GPU_RESET DeviceGpuRecoveryAction = 1 + GPU_RECOVERY_ACTION_NODE_REBOOT DeviceGpuRecoveryAction = 2 + GPU_RECOVERY_ACTION_DRAIN_P2P DeviceGpuRecoveryAction = 3 + GPU_RECOVERY_ACTION_DRAIN_AND_RESET DeviceGpuRecoveryAction = 4 +) + +// FanState as declared in nvml/nvml.h +type FanState int32 + +// FanState enumeration from nvml/nvml.h +const ( + FAN_NORMAL FanState = iota + FAN_FAILED FanState = 1 +) + +// LedColor as declared in nvml/nvml.h +type LedColor int32 + +// LedColor enumeration from nvml/nvml.h +const ( + LED_COLOR_GREEN LedColor = iota + LED_COLOR_AMBER LedColor = 1 +) + +// EncoderType as declared in nvml/nvml.h +type EncoderType int32 + +// EncoderType enumeration from nvml/nvml.h +const ( + ENCODER_QUERY_H264 EncoderType = iota + ENCODER_QUERY_HEVC EncoderType = 1 + ENCODER_QUERY_AV1 EncoderType = 2 + ENCODER_QUERY_UNKNOWN EncoderType = 255 +) + +// FBCSessionType as declared in nvml/nvml.h +type FBCSessionType int32 + +// FBCSessionType enumeration from nvml/nvml.h +const ( + FBC_SESSION_TYPE_UNKNOWN FBCSessionType = iota + FBC_SESSION_TYPE_TOSYS FBCSessionType = 1 + FBC_SESSION_TYPE_CUDA FBCSessionType = 2 + FBC_SESSION_TYPE_VID FBCSessionType = 3 + FBC_SESSION_TYPE_HWENC FBCSessionType = 4 +) + +// DetachGpuState as declared in nvml/nvml.h +type DetachGpuState int32 + +// DetachGpuState enumeration from nvml/nvml.h +const ( + DETACH_GPU_KEEP DetachGpuState = iota + DETACH_GPU_REMOVE DetachGpuState = 1 +) + +// PcieLinkState as declared in nvml/nvml.h +type PcieLinkState int32 + +// PcieLinkState enumeration from nvml/nvml.h +const ( + PCIE_LINK_KEEP PcieLinkState = iota + PCIE_LINK_SHUT_DOWN PcieLinkState = 1 +) + +// ClockLimitId as declared in nvml/nvml.h +type ClockLimitId int32 + +// ClockLimitId enumeration from nvml/nvml.h +const ( + CLOCK_LIMIT_ID_RANGE_START ClockLimitId = -256 + CLOCK_LIMIT_ID_TDP ClockLimitId = -255 + CLOCK_LIMIT_ID_UNLIMITED ClockLimitId = -254 +) + +// VgpuVmCompatibility as declared in nvml/nvml.h +type VgpuVmCompatibility int32 + +// VgpuVmCompatibility enumeration from nvml/nvml.h +const ( + VGPU_VM_COMPATIBILITY_NONE VgpuVmCompatibility = iota + VGPU_VM_COMPATIBILITY_COLD VgpuVmCompatibility = 1 + VGPU_VM_COMPATIBILITY_HIBERNATE VgpuVmCompatibility = 2 + VGPU_VM_COMPATIBILITY_SLEEP VgpuVmCompatibility = 4 + VGPU_VM_COMPATIBILITY_LIVE VgpuVmCompatibility = 8 +) + +// VgpuPgpuCompatibilityLimitCode as declared in nvml/nvml.h +type VgpuPgpuCompatibilityLimitCode int32 + +// VgpuPgpuCompatibilityLimitCode enumeration from nvml/nvml.h +const ( + VGPU_COMPATIBILITY_LIMIT_NONE VgpuPgpuCompatibilityLimitCode = iota + VGPU_COMPATIBILITY_LIMIT_HOST_DRIVER VgpuPgpuCompatibilityLimitCode = 1 + VGPU_COMPATIBILITY_LIMIT_GUEST_DRIVER VgpuPgpuCompatibilityLimitCode = 2 + VGPU_COMPATIBILITY_LIMIT_GPU VgpuPgpuCompatibilityLimitCode = 4 + VGPU_COMPATIBILITY_LIMIT_OTHER VgpuPgpuCompatibilityLimitCode = -2147483648 +) + +// DeviceAddressingModeType as declared in nvml/nvml.h +type DeviceAddressingModeType int32 + +// DeviceAddressingModeType enumeration from nvml/nvml.h +const ( + DEVICE_ADDRESSING_MODE_NONE DeviceAddressingModeType = iota + DEVICE_ADDRESSING_MODE_HMM DeviceAddressingModeType = 1 + DEVICE_ADDRESSING_MODE_ATS DeviceAddressingModeType = 2 +) + +// ThermalTarget as declared in nvml/nvml.h +type ThermalTarget int32 + +// ThermalTarget enumeration from nvml/nvml.h +const ( + THERMAL_TARGET_NONE ThermalTarget = iota + THERMAL_TARGET_GPU ThermalTarget = 1 + THERMAL_TARGET_MEMORY ThermalTarget = 2 + THERMAL_TARGET_POWER_SUPPLY ThermalTarget = 4 + THERMAL_TARGET_BOARD ThermalTarget = 8 + THERMAL_TARGET_VCD_BOARD ThermalTarget = 9 + THERMAL_TARGET_VCD_INLET ThermalTarget = 10 + THERMAL_TARGET_VCD_OUTLET ThermalTarget = 11 + THERMAL_TARGET_ALL ThermalTarget = 15 + THERMAL_TARGET_UNKNOWN ThermalTarget = -1 +) + +// ThermalController as declared in nvml/nvml.h +type ThermalController int32 + +// ThermalController enumeration from nvml/nvml.h +const ( + THERMAL_CONTROLLER_NONE ThermalController = iota + THERMAL_CONTROLLER_GPU_INTERNAL ThermalController = 1 + THERMAL_CONTROLLER_ADM1032 ThermalController = 2 + THERMAL_CONTROLLER_ADT7461 ThermalController = 3 + THERMAL_CONTROLLER_MAX6649 ThermalController = 4 + THERMAL_CONTROLLER_MAX1617 ThermalController = 5 + THERMAL_CONTROLLER_LM99 ThermalController = 6 + THERMAL_CONTROLLER_LM89 ThermalController = 7 + THERMAL_CONTROLLER_LM64 ThermalController = 8 + THERMAL_CONTROLLER_G781 ThermalController = 9 + THERMAL_CONTROLLER_ADT7473 ThermalController = 10 + THERMAL_CONTROLLER_SBMAX6649 ThermalController = 11 + THERMAL_CONTROLLER_VBIOSEVT ThermalController = 12 + THERMAL_CONTROLLER_OS ThermalController = 13 + THERMAL_CONTROLLER_NVSYSCON_CANOAS ThermalController = 14 + THERMAL_CONTROLLER_NVSYSCON_E551 ThermalController = 15 + THERMAL_CONTROLLER_MAX6649R ThermalController = 16 + THERMAL_CONTROLLER_ADT7473S ThermalController = 17 + THERMAL_CONTROLLER_UNKNOWN ThermalController = -1 +) + +// UUIDType as declared in nvml/nvml.h +type UUIDType int32 + +// UUIDType enumeration from nvml/nvml.h +const ( + UUID_TYPE_NONE UUIDType = iota + UUID_TYPE_ASCII UUIDType = 1 + UUID_TYPE_BINARY UUIDType = 2 +) + +// GridLicenseFeatureCode as declared in nvml/nvml.h +type GridLicenseFeatureCode int32 + +// GridLicenseFeatureCode enumeration from nvml/nvml.h +const ( + GRID_LICENSE_FEATURE_CODE_UNKNOWN GridLicenseFeatureCode = iota + GRID_LICENSE_FEATURE_CODE_VGPU GridLicenseFeatureCode = 1 + GRID_LICENSE_FEATURE_CODE_NVIDIA_RTX GridLicenseFeatureCode = 2 + GRID_LICENSE_FEATURE_CODE_VWORKSTATION GridLicenseFeatureCode = 2 + GRID_LICENSE_FEATURE_CODE_GAMING GridLicenseFeatureCode = 3 + GRID_LICENSE_FEATURE_CODE_COMPUTE GridLicenseFeatureCode = 4 +) + +// GpmMetricId as declared in nvml/nvml.h +type GpmMetricId int32 + +// GpmMetricId enumeration from nvml/nvml.h +const ( + GPM_METRIC_GRAPHICS_UTIL GpmMetricId = 1 + GPM_METRIC_SM_UTIL GpmMetricId = 2 + GPM_METRIC_SM_OCCUPANCY GpmMetricId = 3 + GPM_METRIC_INTEGER_UTIL GpmMetricId = 4 + GPM_METRIC_ANY_TENSOR_UTIL GpmMetricId = 5 + GPM_METRIC_DFMA_TENSOR_UTIL GpmMetricId = 6 + GPM_METRIC_HMMA_TENSOR_UTIL GpmMetricId = 7 + GPM_METRIC_IMMA_TENSOR_UTIL GpmMetricId = 9 + GPM_METRIC_DRAM_BW_UTIL GpmMetricId = 10 + GPM_METRIC_FP64_UTIL GpmMetricId = 11 + GPM_METRIC_FP32_UTIL GpmMetricId = 12 + GPM_METRIC_FP16_UTIL GpmMetricId = 13 + GPM_METRIC_PCIE_TX_PER_SEC GpmMetricId = 20 + GPM_METRIC_PCIE_RX_PER_SEC GpmMetricId = 21 + GPM_METRIC_NVDEC_0_UTIL GpmMetricId = 30 + GPM_METRIC_NVDEC_1_UTIL GpmMetricId = 31 + GPM_METRIC_NVDEC_2_UTIL GpmMetricId = 32 + GPM_METRIC_NVDEC_3_UTIL GpmMetricId = 33 + GPM_METRIC_NVDEC_4_UTIL GpmMetricId = 34 + GPM_METRIC_NVDEC_5_UTIL GpmMetricId = 35 + GPM_METRIC_NVDEC_6_UTIL GpmMetricId = 36 + GPM_METRIC_NVDEC_7_UTIL GpmMetricId = 37 + GPM_METRIC_NVJPG_0_UTIL GpmMetricId = 40 + GPM_METRIC_NVJPG_1_UTIL GpmMetricId = 41 + GPM_METRIC_NVJPG_2_UTIL GpmMetricId = 42 + GPM_METRIC_NVJPG_3_UTIL GpmMetricId = 43 + GPM_METRIC_NVJPG_4_UTIL GpmMetricId = 44 + GPM_METRIC_NVJPG_5_UTIL GpmMetricId = 45 + GPM_METRIC_NVJPG_6_UTIL GpmMetricId = 46 + GPM_METRIC_NVJPG_7_UTIL GpmMetricId = 47 + GPM_METRIC_NVOFA_0_UTIL GpmMetricId = 50 + GPM_METRIC_NVOFA_1_UTIL GpmMetricId = 51 + GPM_METRIC_NVLINK_TOTAL_RX_PER_SEC GpmMetricId = 60 + GPM_METRIC_NVLINK_TOTAL_TX_PER_SEC GpmMetricId = 61 + GPM_METRIC_NVLINK_L0_RX_PER_SEC GpmMetricId = 62 + GPM_METRIC_NVLINK_L0_TX_PER_SEC GpmMetricId = 63 + GPM_METRIC_NVLINK_L1_RX_PER_SEC GpmMetricId = 64 + GPM_METRIC_NVLINK_L1_TX_PER_SEC GpmMetricId = 65 + GPM_METRIC_NVLINK_L2_RX_PER_SEC GpmMetricId = 66 + GPM_METRIC_NVLINK_L2_TX_PER_SEC GpmMetricId = 67 + GPM_METRIC_NVLINK_L3_RX_PER_SEC GpmMetricId = 68 + GPM_METRIC_NVLINK_L3_TX_PER_SEC GpmMetricId = 69 + GPM_METRIC_NVLINK_L4_RX_PER_SEC GpmMetricId = 70 + GPM_METRIC_NVLINK_L4_TX_PER_SEC GpmMetricId = 71 + GPM_METRIC_NVLINK_L5_RX_PER_SEC GpmMetricId = 72 + GPM_METRIC_NVLINK_L5_TX_PER_SEC GpmMetricId = 73 + GPM_METRIC_NVLINK_L6_RX_PER_SEC GpmMetricId = 74 + GPM_METRIC_NVLINK_L6_TX_PER_SEC GpmMetricId = 75 + GPM_METRIC_NVLINK_L7_RX_PER_SEC GpmMetricId = 76 + GPM_METRIC_NVLINK_L7_TX_PER_SEC GpmMetricId = 77 + GPM_METRIC_NVLINK_L8_RX_PER_SEC GpmMetricId = 78 + GPM_METRIC_NVLINK_L8_TX_PER_SEC GpmMetricId = 79 + GPM_METRIC_NVLINK_L9_RX_PER_SEC GpmMetricId = 80 + GPM_METRIC_NVLINK_L9_TX_PER_SEC GpmMetricId = 81 + GPM_METRIC_NVLINK_L10_RX_PER_SEC GpmMetricId = 82 + GPM_METRIC_NVLINK_L10_TX_PER_SEC GpmMetricId = 83 + GPM_METRIC_NVLINK_L11_RX_PER_SEC GpmMetricId = 84 + GPM_METRIC_NVLINK_L11_TX_PER_SEC GpmMetricId = 85 + GPM_METRIC_NVLINK_L12_RX_PER_SEC GpmMetricId = 86 + GPM_METRIC_NVLINK_L12_TX_PER_SEC GpmMetricId = 87 + GPM_METRIC_NVLINK_L13_RX_PER_SEC GpmMetricId = 88 + GPM_METRIC_NVLINK_L13_TX_PER_SEC GpmMetricId = 89 + GPM_METRIC_NVLINK_L14_RX_PER_SEC GpmMetricId = 90 + GPM_METRIC_NVLINK_L14_TX_PER_SEC GpmMetricId = 91 + GPM_METRIC_NVLINK_L15_RX_PER_SEC GpmMetricId = 92 + GPM_METRIC_NVLINK_L15_TX_PER_SEC GpmMetricId = 93 + GPM_METRIC_NVLINK_L16_RX_PER_SEC GpmMetricId = 94 + GPM_METRIC_NVLINK_L16_TX_PER_SEC GpmMetricId = 95 + GPM_METRIC_NVLINK_L17_RX_PER_SEC GpmMetricId = 96 + GPM_METRIC_NVLINK_L17_TX_PER_SEC GpmMetricId = 97 + GPM_METRIC_C2C_TOTAL_TX_PER_SEC GpmMetricId = 100 + GPM_METRIC_C2C_TOTAL_RX_PER_SEC GpmMetricId = 101 + GPM_METRIC_C2C_DATA_TX_PER_SEC GpmMetricId = 102 + GPM_METRIC_C2C_DATA_RX_PER_SEC GpmMetricId = 103 + GPM_METRIC_C2C_LINK0_TOTAL_TX_PER_SEC GpmMetricId = 104 + GPM_METRIC_C2C_LINK0_TOTAL_RX_PER_SEC GpmMetricId = 105 + GPM_METRIC_C2C_LINK0_DATA_TX_PER_SEC GpmMetricId = 106 + GPM_METRIC_C2C_LINK0_DATA_RX_PER_SEC GpmMetricId = 107 + GPM_METRIC_C2C_LINK1_TOTAL_TX_PER_SEC GpmMetricId = 108 + GPM_METRIC_C2C_LINK1_TOTAL_RX_PER_SEC GpmMetricId = 109 + GPM_METRIC_C2C_LINK1_DATA_TX_PER_SEC GpmMetricId = 110 + GPM_METRIC_C2C_LINK1_DATA_RX_PER_SEC GpmMetricId = 111 + GPM_METRIC_C2C_LINK2_TOTAL_TX_PER_SEC GpmMetricId = 112 + GPM_METRIC_C2C_LINK2_TOTAL_RX_PER_SEC GpmMetricId = 113 + GPM_METRIC_C2C_LINK2_DATA_TX_PER_SEC GpmMetricId = 114 + GPM_METRIC_C2C_LINK2_DATA_RX_PER_SEC GpmMetricId = 115 + GPM_METRIC_C2C_LINK3_TOTAL_TX_PER_SEC GpmMetricId = 116 + GPM_METRIC_C2C_LINK3_TOTAL_RX_PER_SEC GpmMetricId = 117 + GPM_METRIC_C2C_LINK3_DATA_TX_PER_SEC GpmMetricId = 118 + GPM_METRIC_C2C_LINK3_DATA_RX_PER_SEC GpmMetricId = 119 + GPM_METRIC_C2C_LINK4_TOTAL_TX_PER_SEC GpmMetricId = 120 + GPM_METRIC_C2C_LINK4_TOTAL_RX_PER_SEC GpmMetricId = 121 + GPM_METRIC_C2C_LINK4_DATA_TX_PER_SEC GpmMetricId = 122 + GPM_METRIC_C2C_LINK4_DATA_RX_PER_SEC GpmMetricId = 123 + GPM_METRIC_C2C_LINK5_TOTAL_TX_PER_SEC GpmMetricId = 124 + GPM_METRIC_C2C_LINK5_TOTAL_RX_PER_SEC GpmMetricId = 125 + GPM_METRIC_C2C_LINK5_DATA_TX_PER_SEC GpmMetricId = 126 + GPM_METRIC_C2C_LINK5_DATA_RX_PER_SEC GpmMetricId = 127 + GPM_METRIC_C2C_LINK6_TOTAL_TX_PER_SEC GpmMetricId = 128 + GPM_METRIC_C2C_LINK6_TOTAL_RX_PER_SEC GpmMetricId = 129 + GPM_METRIC_C2C_LINK6_DATA_TX_PER_SEC GpmMetricId = 130 + GPM_METRIC_C2C_LINK6_DATA_RX_PER_SEC GpmMetricId = 131 + GPM_METRIC_C2C_LINK7_TOTAL_TX_PER_SEC GpmMetricId = 132 + GPM_METRIC_C2C_LINK7_TOTAL_RX_PER_SEC GpmMetricId = 133 + GPM_METRIC_C2C_LINK7_DATA_TX_PER_SEC GpmMetricId = 134 + GPM_METRIC_C2C_LINK7_DATA_RX_PER_SEC GpmMetricId = 135 + GPM_METRIC_C2C_LINK8_TOTAL_TX_PER_SEC GpmMetricId = 136 + GPM_METRIC_C2C_LINK8_TOTAL_RX_PER_SEC GpmMetricId = 137 + GPM_METRIC_C2C_LINK8_DATA_TX_PER_SEC GpmMetricId = 138 + GPM_METRIC_C2C_LINK8_DATA_RX_PER_SEC GpmMetricId = 139 + GPM_METRIC_C2C_LINK9_TOTAL_TX_PER_SEC GpmMetricId = 140 + GPM_METRIC_C2C_LINK9_TOTAL_RX_PER_SEC GpmMetricId = 141 + GPM_METRIC_C2C_LINK9_DATA_TX_PER_SEC GpmMetricId = 142 + GPM_METRIC_C2C_LINK9_DATA_RX_PER_SEC GpmMetricId = 143 + GPM_METRIC_C2C_LINK10_TOTAL_TX_PER_SEC GpmMetricId = 144 + GPM_METRIC_C2C_LINK10_TOTAL_RX_PER_SEC GpmMetricId = 145 + GPM_METRIC_C2C_LINK10_DATA_TX_PER_SEC GpmMetricId = 146 + GPM_METRIC_C2C_LINK10_DATA_RX_PER_SEC GpmMetricId = 147 + GPM_METRIC_C2C_LINK11_TOTAL_TX_PER_SEC GpmMetricId = 148 + GPM_METRIC_C2C_LINK11_TOTAL_RX_PER_SEC GpmMetricId = 149 + GPM_METRIC_C2C_LINK11_DATA_TX_PER_SEC GpmMetricId = 150 + GPM_METRIC_C2C_LINK11_DATA_RX_PER_SEC GpmMetricId = 151 + GPM_METRIC_C2C_LINK12_TOTAL_TX_PER_SEC GpmMetricId = 152 + GPM_METRIC_C2C_LINK12_TOTAL_RX_PER_SEC GpmMetricId = 153 + GPM_METRIC_C2C_LINK12_DATA_TX_PER_SEC GpmMetricId = 154 + GPM_METRIC_C2C_LINK12_DATA_RX_PER_SEC GpmMetricId = 155 + GPM_METRIC_C2C_LINK13_TOTAL_TX_PER_SEC GpmMetricId = 156 + GPM_METRIC_C2C_LINK13_TOTAL_RX_PER_SEC GpmMetricId = 157 + GPM_METRIC_C2C_LINK13_DATA_TX_PER_SEC GpmMetricId = 158 + GPM_METRIC_C2C_LINK13_DATA_RX_PER_SEC GpmMetricId = 159 + GPM_METRIC_HOSTMEM_CACHE_HIT GpmMetricId = 160 + GPM_METRIC_HOSTMEM_CACHE_MISS GpmMetricId = 161 + GPM_METRIC_PEERMEM_CACHE_HIT GpmMetricId = 162 + GPM_METRIC_PEERMEM_CACHE_MISS GpmMetricId = 163 + GPM_METRIC_DRAM_CACHE_HIT GpmMetricId = 164 + GPM_METRIC_DRAM_CACHE_MISS GpmMetricId = 165 + GPM_METRIC_NVENC_0_UTIL GpmMetricId = 166 + GPM_METRIC_NVENC_1_UTIL GpmMetricId = 167 + GPM_METRIC_NVENC_2_UTIL GpmMetricId = 168 + GPM_METRIC_NVENC_3_UTIL GpmMetricId = 169 + GPM_METRIC_GR0_CTXSW_CYCLES_ELAPSED GpmMetricId = 170 + GPM_METRIC_GR0_CTXSW_CYCLES_ACTIVE GpmMetricId = 171 + GPM_METRIC_GR0_CTXSW_REQUESTS GpmMetricId = 172 + GPM_METRIC_GR0_CTXSW_CYCLES_PER_REQ GpmMetricId = 173 + GPM_METRIC_GR0_CTXSW_ACTIVE_PCT GpmMetricId = 174 + GPM_METRIC_GR1_CTXSW_CYCLES_ELAPSED GpmMetricId = 175 + GPM_METRIC_GR1_CTXSW_CYCLES_ACTIVE GpmMetricId = 176 + GPM_METRIC_GR1_CTXSW_REQUESTS GpmMetricId = 177 + GPM_METRIC_GR1_CTXSW_CYCLES_PER_REQ GpmMetricId = 178 + GPM_METRIC_GR1_CTXSW_ACTIVE_PCT GpmMetricId = 179 + GPM_METRIC_GR2_CTXSW_CYCLES_ELAPSED GpmMetricId = 180 + GPM_METRIC_GR2_CTXSW_CYCLES_ACTIVE GpmMetricId = 181 + GPM_METRIC_GR2_CTXSW_REQUESTS GpmMetricId = 182 + GPM_METRIC_GR2_CTXSW_CYCLES_PER_REQ GpmMetricId = 183 + GPM_METRIC_GR2_CTXSW_ACTIVE_PCT GpmMetricId = 184 + GPM_METRIC_GR3_CTXSW_CYCLES_ELAPSED GpmMetricId = 185 + GPM_METRIC_GR3_CTXSW_CYCLES_ACTIVE GpmMetricId = 186 + GPM_METRIC_GR3_CTXSW_REQUESTS GpmMetricId = 187 + GPM_METRIC_GR3_CTXSW_CYCLES_PER_REQ GpmMetricId = 188 + GPM_METRIC_GR3_CTXSW_ACTIVE_PCT GpmMetricId = 189 + GPM_METRIC_GR4_CTXSW_CYCLES_ELAPSED GpmMetricId = 190 + GPM_METRIC_GR4_CTXSW_CYCLES_ACTIVE GpmMetricId = 191 + GPM_METRIC_GR4_CTXSW_REQUESTS GpmMetricId = 192 + GPM_METRIC_GR4_CTXSW_CYCLES_PER_REQ GpmMetricId = 193 + GPM_METRIC_GR4_CTXSW_ACTIVE_PCT GpmMetricId = 194 + GPM_METRIC_GR5_CTXSW_CYCLES_ELAPSED GpmMetricId = 195 + GPM_METRIC_GR5_CTXSW_CYCLES_ACTIVE GpmMetricId = 196 + GPM_METRIC_GR5_CTXSW_REQUESTS GpmMetricId = 197 + GPM_METRIC_GR5_CTXSW_CYCLES_PER_REQ GpmMetricId = 198 + GPM_METRIC_GR5_CTXSW_ACTIVE_PCT GpmMetricId = 199 + GPM_METRIC_GR6_CTXSW_CYCLES_ELAPSED GpmMetricId = 200 + GPM_METRIC_GR6_CTXSW_CYCLES_ACTIVE GpmMetricId = 201 + GPM_METRIC_GR6_CTXSW_REQUESTS GpmMetricId = 202 + GPM_METRIC_GR6_CTXSW_CYCLES_PER_REQ GpmMetricId = 203 + GPM_METRIC_GR6_CTXSW_ACTIVE_PCT GpmMetricId = 204 + GPM_METRIC_GR7_CTXSW_CYCLES_ELAPSED GpmMetricId = 205 + GPM_METRIC_GR7_CTXSW_CYCLES_ACTIVE GpmMetricId = 206 + GPM_METRIC_GR7_CTXSW_REQUESTS GpmMetricId = 207 + GPM_METRIC_GR7_CTXSW_CYCLES_PER_REQ GpmMetricId = 208 + GPM_METRIC_GR7_CTXSW_ACTIVE_PCT GpmMetricId = 209 + GPM_METRIC_MAX GpmMetricId = 210 +) + +// PowerProfileType as declared in nvml/nvml.h +type PowerProfileType int32 + +// PowerProfileType enumeration from nvml/nvml.h +const ( + POWER_PROFILE_MAX_P PowerProfileType = iota + POWER_PROFILE_MAX_Q PowerProfileType = 1 + POWER_PROFILE_COMPUTE PowerProfileType = 2 + POWER_PROFILE_MEMORY_BOUND PowerProfileType = 3 + POWER_PROFILE_NETWORK PowerProfileType = 4 + POWER_PROFILE_BALANCED PowerProfileType = 5 + POWER_PROFILE_LLM_INFERENCE PowerProfileType = 6 + POWER_PROFILE_LLM_TRAINING PowerProfileType = 7 + POWER_PROFILE_RBM PowerProfileType = 8 + POWER_PROFILE_DCPCIE PowerProfileType = 9 + POWER_PROFILE_HMMA_SPARSE PowerProfileType = 10 + POWER_PROFILE_HMMA_DENSE PowerProfileType = 11 + POWER_PROFILE_SYNC_BALANCED PowerProfileType = 12 + POWER_PROFILE_HPC PowerProfileType = 13 + POWER_PROFILE_MIG PowerProfileType = 14 + POWER_PROFILE_MAX PowerProfileType = 15 +) diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const_static.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const_static.go new file mode 100644 index 00000000..9038b312 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/const_static.go @@ -0,0 +1,27 @@ +// Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +import ( + "reflect" +) + +const ( + SYSTEM_PROCESS_NAME_BUFFER_SIZE = 256 +) + +func STRUCT_VERSION(data interface{}, version uint32) uint32 { + return uint32(uint32(reflect.Indirect(reflect.ValueOf(data)).Type().Size()) | (version << uint32(24))) +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/device.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/device.go new file mode 100644 index 00000000..d341e157 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/device.go @@ -0,0 +1,3592 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +import ( + "fmt" + "reflect" + "unsafe" +) + +// nvmlDeviceHandle attempts to convert a device d to an nvmlDevice. +// This is required for functions such as GetTopologyCommonAncestor which +// accept Device arguments that need to be passed to internal nvml* functions +// as nvmlDevice parameters. +func nvmlDeviceHandle(d Device) nvmlDevice { + var helper func(val reflect.Value) nvmlDevice + helper = func(val reflect.Value) nvmlDevice { + if val.Kind() == reflect.Interface { + val = val.Elem() + } + + if val.Kind() == reflect.Ptr { + val = val.Elem() + } + + if val.Type() == reflect.TypeOf(nvmlDevice{}) { + return val.Interface().(nvmlDevice) + } + + if val.Kind() != reflect.Struct { + panic(fmt.Errorf("unable to convert non-struct type %v to nvmlDevice", val.Kind())) + } + + for i := 0; i < val.Type().NumField(); i++ { + if !val.Type().Field(i).Anonymous { + continue + } + if !val.Field(i).Type().Implements(reflect.TypeOf((*Device)(nil)).Elem()) { + continue + } + return helper(val.Field(i)) + } + panic(fmt.Errorf("unable to convert %T to nvmlDevice", d)) + } + return helper(reflect.ValueOf(d)) +} + +// EccBitType +type EccBitType = MemoryErrorType + +// GpuInstanceInfo includes an interface type for Device instead of nvmlDevice +type GpuInstanceInfo struct { + Device Device + Id uint32 + ProfileId uint32 + Placement GpuInstancePlacement +} + +func (g nvmlGpuInstanceInfo) convert() GpuInstanceInfo { + out := GpuInstanceInfo{ + Device: g.Device, + Id: g.Id, + ProfileId: g.ProfileId, + Placement: g.Placement, + } + return out +} + +// ComputeInstanceInfo includes an interface type for Device instead of nvmlDevice +type ComputeInstanceInfo struct { + Device Device + GpuInstance GpuInstance + Id uint32 + ProfileId uint32 + Placement ComputeInstancePlacement +} + +func (c nvmlComputeInstanceInfo) convert() ComputeInstanceInfo { + out := ComputeInstanceInfo{ + Device: c.Device, + GpuInstance: c.GpuInstance, + Id: c.Id, + ProfileId: c.ProfileId, + Placement: c.Placement, + } + return out +} + +// nvml.DeviceGetCount() +func (l *library) DeviceGetCount() (int, Return) { + var deviceCount uint32 + ret := nvmlDeviceGetCount(&deviceCount) + return int(deviceCount), ret +} + +// nvml.DeviceGetHandleByIndex() +func (l *library) DeviceGetHandleByIndex(index int) (Device, Return) { + var device nvmlDevice + ret := nvmlDeviceGetHandleByIndex(uint32(index), &device) + return device, ret +} + +// nvml.DeviceGetHandleBySerial() +// +// Deprecated: Use DeviceGetHandleByUUID instead. +func (l *library) DeviceGetHandleBySerial(serial string) (Device, Return) { + var device nvmlDevice + ret := nvmlDeviceGetHandleBySerial(serial+string(rune(0)), &device) + return device, ret +} + +// nvml.DeviceGetHandleByUUID() +func (l *library) DeviceGetHandleByUUID(uuid string) (Device, Return) { + var device nvmlDevice + ret := nvmlDeviceGetHandleByUUID(uuid+string(rune(0)), &device) + return device, ret +} + +// nvml.DeviceGetHandleByUUIDV() +func (l *library) DeviceGetHandleByUUIDV(uuid *UUID) (Device, Return) { + var device nvmlDevice + ret := nvmlDeviceGetHandleByUUIDV(uuid, &device) + return device, ret +} + +// nvml.DeviceGetHandleByPciBusId() +func (l *library) DeviceGetHandleByPciBusId(pciBusId string) (Device, Return) { + var device nvmlDevice + ret := nvmlDeviceGetHandleByPciBusId(pciBusId+string(rune(0)), &device) + return device, ret +} + +// nvml.DeviceGetName() +func (l *library) DeviceGetName(device Device) (string, Return) { + return device.GetName() +} + +func (device nvmlDevice) GetName() (string, Return) { + name := make([]byte, DEVICE_NAME_V2_BUFFER_SIZE) + ret := nvmlDeviceGetName(device, &name[0], DEVICE_NAME_V2_BUFFER_SIZE) + return string(name[:clen(name)]), ret +} + +// nvml.DeviceGetBrand() +func (l *library) DeviceGetBrand(device Device) (BrandType, Return) { + return device.GetBrand() +} + +func (device nvmlDevice) GetBrand() (BrandType, Return) { + var brandType BrandType + ret := nvmlDeviceGetBrand(device, &brandType) + return brandType, ret +} + +// nvml.DeviceGetIndex() +func (l *library) DeviceGetIndex(device Device) (int, Return) { + return device.GetIndex() +} + +func (device nvmlDevice) GetIndex() (int, Return) { + var index uint32 + ret := nvmlDeviceGetIndex(device, &index) + return int(index), ret +} + +// nvml.DeviceGetSerial() +func (l *library) DeviceGetSerial(device Device) (string, Return) { + return device.GetSerial() +} + +func (device nvmlDevice) GetSerial() (string, Return) { + serial := make([]byte, DEVICE_SERIAL_BUFFER_SIZE) + ret := nvmlDeviceGetSerial(device, &serial[0], DEVICE_SERIAL_BUFFER_SIZE) + return string(serial[:clen(serial)]), ret +} + +// nvml.DeviceGetCpuAffinity() +func (l *library) DeviceGetCpuAffinity(device Device, numCPUs int) ([]uint, Return) { + return device.GetCpuAffinity(numCPUs) +} + +func (device nvmlDevice) GetCpuAffinity(numCPUs int) ([]uint, Return) { + cpuSetSize := uint32((numCPUs-1)/int(unsafe.Sizeof(uint(0))) + 1) + cpuSet := make([]uint, cpuSetSize) + ret := nvmlDeviceGetCpuAffinity(device, cpuSetSize, &cpuSet[0]) + return cpuSet, ret +} + +// nvml.DeviceSetCpuAffinity() +func (l *library) DeviceSetCpuAffinity(device Device) Return { + return device.SetCpuAffinity() +} + +func (device nvmlDevice) SetCpuAffinity() Return { + return nvmlDeviceSetCpuAffinity(device) +} + +// nvml.DeviceClearCpuAffinity() +func (l *library) DeviceClearCpuAffinity(device Device) Return { + return device.ClearCpuAffinity() +} + +func (device nvmlDevice) ClearCpuAffinity() Return { + return nvmlDeviceClearCpuAffinity(device) +} + +// nvml.DeviceGetMemoryAffinity() +func (l *library) DeviceGetMemoryAffinity(device Device, numNodes int, scope AffinityScope) ([]uint, Return) { + return device.GetMemoryAffinity(numNodes, scope) +} + +func (device nvmlDevice) GetMemoryAffinity(numNodes int, scope AffinityScope) ([]uint, Return) { + nodeSetSize := uint32((numNodes-1)/int(unsafe.Sizeof(uint(0))) + 1) + nodeSet := make([]uint, nodeSetSize) + ret := nvmlDeviceGetMemoryAffinity(device, nodeSetSize, &nodeSet[0], scope) + return nodeSet, ret +} + +// nvml.DeviceGetCpuAffinityWithinScope() +func (l *library) DeviceGetCpuAffinityWithinScope(device Device, numCPUs int, scope AffinityScope) ([]uint, Return) { + return device.GetCpuAffinityWithinScope(numCPUs, scope) +} + +func (device nvmlDevice) GetCpuAffinityWithinScope(numCPUs int, scope AffinityScope) ([]uint, Return) { + cpuSetSize := uint32((numCPUs-1)/int(unsafe.Sizeof(uint(0))) + 1) + cpuSet := make([]uint, cpuSetSize) + ret := nvmlDeviceGetCpuAffinityWithinScope(device, cpuSetSize, &cpuSet[0], scope) + return cpuSet, ret +} + +// nvml.DeviceGetTopologyCommonAncestor() +func (l *library) DeviceGetTopologyCommonAncestor(device1 Device, device2 Device) (GpuTopologyLevel, Return) { + return device1.GetTopologyCommonAncestor(device2) +} + +func (device1 nvmlDevice) GetTopologyCommonAncestor(device2 Device) (GpuTopologyLevel, Return) { + var pathInfo GpuTopologyLevel + ret := nvmlDeviceGetTopologyCommonAncestorStub(device1, nvmlDeviceHandle(device2), &pathInfo) + return pathInfo, ret +} + +// nvmlDeviceGetTopologyCommonAncestorStub allows us to override this for testing. +var nvmlDeviceGetTopologyCommonAncestorStub = nvmlDeviceGetTopologyCommonAncestor + +// nvml.DeviceGetTopologyNearestGpus() +func (l *library) DeviceGetTopologyNearestGpus(device Device, level GpuTopologyLevel) ([]Device, Return) { + return device.GetTopologyNearestGpus(level) +} + +func (device nvmlDevice) GetTopologyNearestGpus(level GpuTopologyLevel) ([]Device, Return) { + var count uint32 + ret := nvmlDeviceGetTopologyNearestGpus(device, level, &count, nil) + if ret != SUCCESS { + return nil, ret + } + if count == 0 { + return []Device{}, ret + } + deviceArray := make([]nvmlDevice, count) + ret = nvmlDeviceGetTopologyNearestGpus(device, level, &count, &deviceArray[0]) + return convertSlice[nvmlDevice, Device](deviceArray), ret +} + +// nvml.DeviceGetP2PStatus() +func (l *library) DeviceGetP2PStatus(device1 Device, device2 Device, p2pIndex GpuP2PCapsIndex) (GpuP2PStatus, Return) { + return device1.GetP2PStatus(device2, p2pIndex) +} + +func (device1 nvmlDevice) GetP2PStatus(device2 Device, p2pIndex GpuP2PCapsIndex) (GpuP2PStatus, Return) { + var p2pStatus GpuP2PStatus + ret := nvmlDeviceGetP2PStatus(device1, nvmlDeviceHandle(device2), p2pIndex, &p2pStatus) + return p2pStatus, ret +} + +// nvml.DeviceGetUUID() +func (l *library) DeviceGetUUID(device Device) (string, Return) { + return device.GetUUID() +} + +func (device nvmlDevice) GetUUID() (string, Return) { + uuid := make([]byte, DEVICE_UUID_V2_BUFFER_SIZE) + ret := nvmlDeviceGetUUID(device, &uuid[0], DEVICE_UUID_V2_BUFFER_SIZE) + return string(uuid[:clen(uuid)]), ret +} + +// nvml.DeviceGetMinorNumber() +func (l *library) DeviceGetMinorNumber(device Device) (int, Return) { + return device.GetMinorNumber() +} + +func (device nvmlDevice) GetMinorNumber() (int, Return) { + var minorNumber uint32 + ret := nvmlDeviceGetMinorNumber(device, &minorNumber) + return int(minorNumber), ret +} + +// nvml.DeviceGetBoardPartNumber() +func (l *library) DeviceGetBoardPartNumber(device Device) (string, Return) { + return device.GetBoardPartNumber() +} + +func (device nvmlDevice) GetBoardPartNumber() (string, Return) { + partNumber := make([]byte, DEVICE_PART_NUMBER_BUFFER_SIZE) + ret := nvmlDeviceGetBoardPartNumber(device, &partNumber[0], DEVICE_PART_NUMBER_BUFFER_SIZE) + return string(partNumber[:clen(partNumber)]), ret +} + +// nvml.DeviceGetInforomVersion() +func (l *library) DeviceGetInforomVersion(device Device, object InforomObject) (string, Return) { + return device.GetInforomVersion(object) +} + +func (device nvmlDevice) GetInforomVersion(object InforomObject) (string, Return) { + version := make([]byte, DEVICE_INFOROM_VERSION_BUFFER_SIZE) + ret := nvmlDeviceGetInforomVersion(device, object, &version[0], DEVICE_INFOROM_VERSION_BUFFER_SIZE) + return string(version[:clen(version)]), ret +} + +// nvml.DeviceGetInforomImageVersion() +func (l *library) DeviceGetInforomImageVersion(device Device) (string, Return) { + return device.GetInforomImageVersion() +} + +func (device nvmlDevice) GetInforomImageVersion() (string, Return) { + version := make([]byte, DEVICE_INFOROM_VERSION_BUFFER_SIZE) + ret := nvmlDeviceGetInforomImageVersion(device, &version[0], DEVICE_INFOROM_VERSION_BUFFER_SIZE) + return string(version[:clen(version)]), ret +} + +// nvml.DeviceGetInforomConfigurationChecksum() +func (l *library) DeviceGetInforomConfigurationChecksum(device Device) (uint32, Return) { + return device.GetInforomConfigurationChecksum() +} + +func (device nvmlDevice) GetInforomConfigurationChecksum() (uint32, Return) { + var checksum uint32 + ret := nvmlDeviceGetInforomConfigurationChecksum(device, &checksum) + return checksum, ret +} + +// nvml.DeviceValidateInforom() +func (l *library) DeviceValidateInforom(device Device) Return { + return device.ValidateInforom() +} + +func (device nvmlDevice) ValidateInforom() Return { + return nvmlDeviceValidateInforom(device) +} + +// nvml.DeviceGetDisplayMode() +func (l *library) DeviceGetDisplayMode(device Device) (EnableState, Return) { + return device.GetDisplayMode() +} + +func (device nvmlDevice) GetDisplayMode() (EnableState, Return) { + var display EnableState + ret := nvmlDeviceGetDisplayMode(device, &display) + return display, ret +} + +// nvml.DeviceGetDisplayActive() +func (l *library) DeviceGetDisplayActive(device Device) (EnableState, Return) { + return device.GetDisplayActive() +} + +func (device nvmlDevice) GetDisplayActive() (EnableState, Return) { + var isActive EnableState + ret := nvmlDeviceGetDisplayActive(device, &isActive) + return isActive, ret +} + +// nvml.DeviceGetPersistenceMode() +func (l *library) DeviceGetPersistenceMode(device Device) (EnableState, Return) { + return device.GetPersistenceMode() +} + +func (device nvmlDevice) GetPersistenceMode() (EnableState, Return) { + var mode EnableState + ret := nvmlDeviceGetPersistenceMode(device, &mode) + return mode, ret +} + +// nvml.DeviceGetPciInfo() +func (l *library) DeviceGetPciInfo(device Device) (PciInfo, Return) { + return device.GetPciInfo() +} + +func (device nvmlDevice) GetPciInfo() (PciInfo, Return) { + var pci PciInfo + ret := nvmlDeviceGetPciInfo(device, &pci) + return pci, ret +} + +// nvml.DeviceGetMaxPcieLinkGeneration() +func (l *library) DeviceGetMaxPcieLinkGeneration(device Device) (int, Return) { + return device.GetMaxPcieLinkGeneration() +} + +func (device nvmlDevice) GetMaxPcieLinkGeneration() (int, Return) { + var maxLinkGen uint32 + ret := nvmlDeviceGetMaxPcieLinkGeneration(device, &maxLinkGen) + return int(maxLinkGen), ret +} + +// nvml.DeviceGetMaxPcieLinkWidth() +func (l *library) DeviceGetMaxPcieLinkWidth(device Device) (int, Return) { + return device.GetMaxPcieLinkWidth() +} + +func (device nvmlDevice) GetMaxPcieLinkWidth() (int, Return) { + var maxLinkWidth uint32 + ret := nvmlDeviceGetMaxPcieLinkWidth(device, &maxLinkWidth) + return int(maxLinkWidth), ret +} + +// nvml.DeviceGetCurrPcieLinkGeneration() +func (l *library) DeviceGetCurrPcieLinkGeneration(device Device) (int, Return) { + return device.GetCurrPcieLinkGeneration() +} + +func (device nvmlDevice) GetCurrPcieLinkGeneration() (int, Return) { + var currLinkGen uint32 + ret := nvmlDeviceGetCurrPcieLinkGeneration(device, &currLinkGen) + return int(currLinkGen), ret +} + +// nvml.DeviceGetCurrPcieLinkWidth() +func (l *library) DeviceGetCurrPcieLinkWidth(device Device) (int, Return) { + return device.GetCurrPcieLinkWidth() +} + +func (device nvmlDevice) GetCurrPcieLinkWidth() (int, Return) { + var currLinkWidth uint32 + ret := nvmlDeviceGetCurrPcieLinkWidth(device, &currLinkWidth) + return int(currLinkWidth), ret +} + +// nvml.DeviceGetPcieThroughput() +func (l *library) DeviceGetPcieThroughput(device Device, counter PcieUtilCounter) (uint32, Return) { + return device.GetPcieThroughput(counter) +} + +func (device nvmlDevice) GetPcieThroughput(counter PcieUtilCounter) (uint32, Return) { + var value uint32 + ret := nvmlDeviceGetPcieThroughput(device, counter, &value) + return value, ret +} + +// nvml.DeviceGetPcieReplayCounter() +func (l *library) DeviceGetPcieReplayCounter(device Device) (int, Return) { + return device.GetPcieReplayCounter() +} + +func (device nvmlDevice) GetPcieReplayCounter() (int, Return) { + var value uint32 + ret := nvmlDeviceGetPcieReplayCounter(device, &value) + return int(value), ret +} + +// nvml.nvmlDeviceGetClockInfo() +func (l *library) DeviceGetClockInfo(device Device, clockType ClockType) (uint32, Return) { + return device.GetClockInfo(clockType) +} + +func (device nvmlDevice) GetClockInfo(clockType ClockType) (uint32, Return) { + var clock uint32 + ret := nvmlDeviceGetClockInfo(device, clockType, &clock) + return clock, ret +} + +// nvml.DeviceGetMaxClockInfo() +func (l *library) DeviceGetMaxClockInfo(device Device, clockType ClockType) (uint32, Return) { + return device.GetMaxClockInfo(clockType) +} + +func (device nvmlDevice) GetMaxClockInfo(clockType ClockType) (uint32, Return) { + var clock uint32 + ret := nvmlDeviceGetMaxClockInfo(device, clockType, &clock) + return clock, ret +} + +// nvml.DeviceGetApplicationsClock() +// +// Deprecated: Applications clocks are deprecated and will be removed in CUDA 14.0. +func (l *library) DeviceGetApplicationsClock(device Device, clockType ClockType) (uint32, Return) { + return device.GetApplicationsClock(clockType) +} + +// Deprecated: Applications clocks are deprecated and will be removed in CUDA 14.0. +func (device nvmlDevice) GetApplicationsClock(clockType ClockType) (uint32, Return) { + var clockMHz uint32 + ret := nvmlDeviceGetApplicationsClock(device, clockType, &clockMHz) + return clockMHz, ret +} + +// nvml.DeviceGetDefaultApplicationsClock() +// +// Deprecated: Applications clocks are deprecated and will be removed in CUDA 14.0. +func (l *library) DeviceGetDefaultApplicationsClock(device Device, clockType ClockType) (uint32, Return) { + return device.GetDefaultApplicationsClock(clockType) +} + +// Deprecated: Applications clocks are deprecated and will be removed in CUDA 14.0. +func (device nvmlDevice) GetDefaultApplicationsClock(clockType ClockType) (uint32, Return) { + var clockMHz uint32 + ret := nvmlDeviceGetDefaultApplicationsClock(device, clockType, &clockMHz) + return clockMHz, ret +} + +// nvml.DeviceResetApplicationsClocks() +// +// Deprecated: Use DeviceResetMemoryLockedClocks for Memory Clocks and DeviceResetGpuLockedClocks for Graphics Clocks instead +func (l *library) DeviceResetApplicationsClocks(device Device) Return { + return device.ResetApplicationsClocks() +} + +// Deprecated: Use DeviceResetMemoryLockedClocks for Memory Clocks and DeviceResetGpuLockedClocks for Graphics Clocks instead +func (device nvmlDevice) ResetApplicationsClocks() Return { + return nvmlDeviceResetApplicationsClocks(device) +} + +// nvml.DeviceGetClock() +func (l *library) DeviceGetClock(device Device, clockType ClockType, clockId ClockId) (uint32, Return) { + return device.GetClock(clockType, clockId) +} + +func (device nvmlDevice) GetClock(clockType ClockType, clockId ClockId) (uint32, Return) { + var clockMHz uint32 + ret := nvmlDeviceGetClock(device, clockType, clockId, &clockMHz) + return clockMHz, ret +} + +// nvml.DeviceGetMaxCustomerBoostClock() +func (l *library) DeviceGetMaxCustomerBoostClock(device Device, clockType ClockType) (uint32, Return) { + return device.GetMaxCustomerBoostClock(clockType) +} + +func (device nvmlDevice) GetMaxCustomerBoostClock(clockType ClockType) (uint32, Return) { + var clockMHz uint32 + ret := nvmlDeviceGetMaxCustomerBoostClock(device, clockType, &clockMHz) + return clockMHz, ret +} + +// nvml.DeviceGetSupportedMemoryClocks() +func (l *library) DeviceGetSupportedMemoryClocks(device Device) (int, uint32, Return) { + return device.GetSupportedMemoryClocks() +} + +func (device nvmlDevice) GetSupportedMemoryClocks() (int, uint32, Return) { + var count, clocksMHz uint32 + ret := nvmlDeviceGetSupportedMemoryClocks(device, &count, &clocksMHz) + return int(count), clocksMHz, ret +} + +// nvml.DeviceGetSupportedGraphicsClocks() +func (l *library) DeviceGetSupportedGraphicsClocks(device Device, memoryClockMHz int) (int, uint32, Return) { + return device.GetSupportedGraphicsClocks(memoryClockMHz) +} + +func (device nvmlDevice) GetSupportedGraphicsClocks(memoryClockMHz int) (int, uint32, Return) { + var count, clocksMHz uint32 + ret := nvmlDeviceGetSupportedGraphicsClocks(device, uint32(memoryClockMHz), &count, &clocksMHz) + return int(count), clocksMHz, ret +} + +// nvml.DeviceGetAutoBoostedClocksEnabled() +func (l *library) DeviceGetAutoBoostedClocksEnabled(device Device) (EnableState, EnableState, Return) { + return device.GetAutoBoostedClocksEnabled() +} + +func (device nvmlDevice) GetAutoBoostedClocksEnabled() (EnableState, EnableState, Return) { + var isEnabled, defaultIsEnabled EnableState + ret := nvmlDeviceGetAutoBoostedClocksEnabled(device, &isEnabled, &defaultIsEnabled) + return isEnabled, defaultIsEnabled, ret +} + +// nvml.DeviceSetAutoBoostedClocksEnabled() +func (l *library) DeviceSetAutoBoostedClocksEnabled(device Device, enabled EnableState) Return { + return device.SetAutoBoostedClocksEnabled(enabled) +} + +func (device nvmlDevice) SetAutoBoostedClocksEnabled(enabled EnableState) Return { + return nvmlDeviceSetAutoBoostedClocksEnabled(device, enabled) +} + +// nvml.DeviceSetDefaultAutoBoostedClocksEnabled() +func (l *library) DeviceSetDefaultAutoBoostedClocksEnabled(device Device, enabled EnableState, flags uint32) Return { + return device.SetDefaultAutoBoostedClocksEnabled(enabled, flags) +} + +func (device nvmlDevice) SetDefaultAutoBoostedClocksEnabled(enabled EnableState, flags uint32) Return { + return nvmlDeviceSetDefaultAutoBoostedClocksEnabled(device, enabled, flags) +} + +// nvml.DeviceGetFanSpeed() +func (l *library) DeviceGetFanSpeed(device Device) (uint32, Return) { + return device.GetFanSpeed() +} + +func (device nvmlDevice) GetFanSpeed() (uint32, Return) { + var speed uint32 + ret := nvmlDeviceGetFanSpeed(device, &speed) + return speed, ret +} + +// nvml.DeviceGetFanSpeed_v2() +func (l *library) DeviceGetFanSpeed_v2(device Device, fan int) (uint32, Return) { + return device.GetFanSpeed_v2(fan) +} + +func (device nvmlDevice) GetFanSpeed_v2(fan int) (uint32, Return) { + var speed uint32 + ret := nvmlDeviceGetFanSpeed_v2(device, uint32(fan), &speed) + return speed, ret +} + +// nvml.DeviceGetNumFans() +func (l *library) DeviceGetNumFans(device Device) (int, Return) { + return device.GetNumFans() +} + +func (device nvmlDevice) GetNumFans() (int, Return) { + var numFans uint32 + ret := nvmlDeviceGetNumFans(device, &numFans) + return int(numFans), ret +} + +// nvml.DeviceGetTemperature() +// +// Deprecated: Use DeviceGetTemperatureV instead. +func (l *library) DeviceGetTemperature(device Device, sensorType TemperatureSensors) (uint32, Return) { + return device.GetTemperature(sensorType) +} + +// Deprecated: Use DeviceGetTemperatureV instead. +func (device nvmlDevice) GetTemperature(sensorType TemperatureSensors) (uint32, Return) { + var temp uint32 + ret := nvmlDeviceGetTemperature(device, sensorType, &temp) + return temp, ret +} + +// nvml.DeviceGetTemperatureThreshold() +func (l *library) DeviceGetTemperatureThreshold(device Device, thresholdType TemperatureThresholds) (uint32, Return) { + return device.GetTemperatureThreshold(thresholdType) +} + +func (device nvmlDevice) GetTemperatureThreshold(thresholdType TemperatureThresholds) (uint32, Return) { + var temp uint32 + ret := nvmlDeviceGetTemperatureThreshold(device, thresholdType, &temp) + return temp, ret +} + +// nvml.DeviceSetTemperatureThreshold() +func (l *library) DeviceSetTemperatureThreshold(device Device, thresholdType TemperatureThresholds, temp int) Return { + return device.SetTemperatureThreshold(thresholdType, temp) +} + +func (device nvmlDevice) SetTemperatureThreshold(thresholdType TemperatureThresholds, temp int) Return { + t := int32(temp) + ret := nvmlDeviceSetTemperatureThreshold(device, thresholdType, &t) + return ret +} + +// nvml.DeviceGetPerformanceState() +func (l *library) DeviceGetPerformanceState(device Device) (Pstates, Return) { + return device.GetPerformanceState() +} + +func (device nvmlDevice) GetPerformanceState() (Pstates, Return) { + var pState Pstates + ret := nvmlDeviceGetPerformanceState(device, &pState) + return pState, ret +} + +// nvml.DeviceGetCurrentClocksThrottleReasons() +// +// Deprecated: Use DeviceGetCurrentClocksEventReasons instead +func (l *library) DeviceGetCurrentClocksThrottleReasons(device Device) (uint64, Return) { + return device.GetCurrentClocksThrottleReasons() +} + +// Deprecated: Use DeviceGetCurrentClocksEventReasons instead +func (device nvmlDevice) GetCurrentClocksThrottleReasons() (uint64, Return) { + var clocksThrottleReasons uint64 + ret := nvmlDeviceGetCurrentClocksThrottleReasons(device, &clocksThrottleReasons) + return clocksThrottleReasons, ret +} + +// nvml.DeviceGetSupportedClocksThrottleReasons() +// +// Deprecated: Use DeviceGetSupportedClocksEventReasons instead +func (l *library) DeviceGetSupportedClocksThrottleReasons(device Device) (uint64, Return) { + return device.GetSupportedClocksThrottleReasons() +} + +// Deprecated: Use DeviceGetSupportedClocksEventReasons instead +func (device nvmlDevice) GetSupportedClocksThrottleReasons() (uint64, Return) { + var supportedClocksThrottleReasons uint64 + ret := nvmlDeviceGetSupportedClocksThrottleReasons(device, &supportedClocksThrottleReasons) + return supportedClocksThrottleReasons, ret +} + +// nvml.DeviceGetPowerState() +// +// Deprecated: Use DeviceGetPerformanceState instead +func (l *library) DeviceGetPowerState(device Device) (Pstates, Return) { + return device.GetPowerState() +} + +// Deprecated: Use DeviceGetPerformanceState instead +func (device nvmlDevice) GetPowerState() (Pstates, Return) { + var pState Pstates + ret := nvmlDeviceGetPowerState(device, &pState) + return pState, ret +} + +// nvml.DeviceGetPowerManagementMode() +// +// Deprecated: This will be removed in a future version +func (l *library) DeviceGetPowerManagementMode(device Device) (EnableState, Return) { + return device.GetPowerManagementMode() +} + +// Deprecated: This will be removed in a future version +func (device nvmlDevice) GetPowerManagementMode() (EnableState, Return) { + var mode EnableState + ret := nvmlDeviceGetPowerManagementMode(device, &mode) + return mode, ret +} + +// nvml.DeviceGetPowerManagementLimit() +func (l *library) DeviceGetPowerManagementLimit(device Device) (uint32, Return) { + return device.GetPowerManagementLimit() +} + +func (device nvmlDevice) GetPowerManagementLimit() (uint32, Return) { + var limit uint32 + ret := nvmlDeviceGetPowerManagementLimit(device, &limit) + return limit, ret +} + +// nvml.DeviceGetPowerManagementLimitConstraints() +func (l *library) DeviceGetPowerManagementLimitConstraints(device Device) (uint32, uint32, Return) { + return device.GetPowerManagementLimitConstraints() +} + +func (device nvmlDevice) GetPowerManagementLimitConstraints() (uint32, uint32, Return) { + var minLimit, maxLimit uint32 + ret := nvmlDeviceGetPowerManagementLimitConstraints(device, &minLimit, &maxLimit) + return minLimit, maxLimit, ret +} + +// nvml.DeviceGetPowerManagementDefaultLimit() +func (l *library) DeviceGetPowerManagementDefaultLimit(device Device) (uint32, Return) { + return device.GetPowerManagementDefaultLimit() +} + +func (device nvmlDevice) GetPowerManagementDefaultLimit() (uint32, Return) { + var defaultLimit uint32 + ret := nvmlDeviceGetPowerManagementDefaultLimit(device, &defaultLimit) + return defaultLimit, ret +} + +// nvml.DeviceGetPowerUsage() +func (l *library) DeviceGetPowerUsage(device Device) (uint32, Return) { + return device.GetPowerUsage() +} + +func (device nvmlDevice) GetPowerUsage() (uint32, Return) { + var power uint32 + ret := nvmlDeviceGetPowerUsage(device, &power) + return power, ret +} + +func (l *library) DeviceGetPowerMizerMode_v1(device Device) (DevicePowerMizerModes_v1, Return) { + return device.GetPowerMizerMode_v1() +} + +func (device nvmlDevice) GetPowerMizerMode_v1() (DevicePowerMizerModes_v1, Return) { + var devicePowerMizerModes DevicePowerMizerModes_v1 + ret := nvmlDeviceGetPowerMizerMode_v1(device, &devicePowerMizerModes) + return devicePowerMizerModes, ret +} + +// nvml.DeviceGetTotalEnergyConsumption() +func (l *library) DeviceGetTotalEnergyConsumption(device Device) (uint64, Return) { + return device.GetTotalEnergyConsumption() +} + +func (device nvmlDevice) GetTotalEnergyConsumption() (uint64, Return) { + var energy uint64 + ret := nvmlDeviceGetTotalEnergyConsumption(device, &energy) + return energy, ret +} + +// nvml.DeviceGetEnforcedPowerLimit() +func (l *library) DeviceGetEnforcedPowerLimit(device Device) (uint32, Return) { + return device.GetEnforcedPowerLimit() +} + +func (device nvmlDevice) GetEnforcedPowerLimit() (uint32, Return) { + var limit uint32 + ret := nvmlDeviceGetEnforcedPowerLimit(device, &limit) + return limit, ret +} + +// nvml.DeviceGetGpuOperationMode() +func (l *library) DeviceGetGpuOperationMode(device Device) (GpuOperationMode, GpuOperationMode, Return) { + return device.GetGpuOperationMode() +} + +func (device nvmlDevice) GetGpuOperationMode() (GpuOperationMode, GpuOperationMode, Return) { + var current, pending GpuOperationMode + ret := nvmlDeviceGetGpuOperationMode(device, ¤t, &pending) + return current, pending, ret +} + +// nvml.DeviceGetMemoryInfo() +func (l *library) DeviceGetMemoryInfo(device Device) (Memory, Return) { + return device.GetMemoryInfo() +} + +func (device nvmlDevice) GetMemoryInfo() (Memory, Return) { + var memory Memory + ret := nvmlDeviceGetMemoryInfo(device, &memory) + return memory, ret +} + +// nvml.DeviceGetMemoryInfo_v2() +func (l *library) DeviceGetMemoryInfo_v2(device Device) (Memory_v2, Return) { + return device.GetMemoryInfo_v2() +} + +func (device nvmlDevice) GetMemoryInfo_v2() (Memory_v2, Return) { + var memory Memory_v2 + memory.Version = STRUCT_VERSION(memory, 2) + ret := nvmlDeviceGetMemoryInfo_v2(device, &memory) + return memory, ret +} + +// nvml.DeviceGetComputeMode() +func (l *library) DeviceGetComputeMode(device Device) (ComputeMode, Return) { + return device.GetComputeMode() +} + +func (device nvmlDevice) GetComputeMode() (ComputeMode, Return) { + var mode ComputeMode + ret := nvmlDeviceGetComputeMode(device, &mode) + return mode, ret +} + +// nvml.DeviceGetCudaComputeCapability() +func (l *library) DeviceGetCudaComputeCapability(device Device) (int, int, Return) { + return device.GetCudaComputeCapability() +} + +func (device nvmlDevice) GetCudaComputeCapability() (int, int, Return) { + var major, minor int32 + ret := nvmlDeviceGetCudaComputeCapability(device, &major, &minor) + return int(major), int(minor), ret +} + +// nvml.DeviceGetEccMode() +func (l *library) DeviceGetEccMode(device Device) (EnableState, EnableState, Return) { + return device.GetEccMode() +} + +func (device nvmlDevice) GetEccMode() (EnableState, EnableState, Return) { + var current, pending EnableState + ret := nvmlDeviceGetEccMode(device, ¤t, &pending) + return current, pending, ret +} + +// nvml.DeviceGetBoardId() +func (l *library) DeviceGetBoardId(device Device) (uint32, Return) { + return device.GetBoardId() +} + +func (device nvmlDevice) GetBoardId() (uint32, Return) { + var boardId uint32 + ret := nvmlDeviceGetBoardId(device, &boardId) + return boardId, ret +} + +// nvml.DeviceGetMultiGpuBoard() +func (l *library) DeviceGetMultiGpuBoard(device Device) (int, Return) { + return device.GetMultiGpuBoard() +} + +func (device nvmlDevice) GetMultiGpuBoard() (int, Return) { + var multiGpuBool uint32 + ret := nvmlDeviceGetMultiGpuBoard(device, &multiGpuBool) + return int(multiGpuBool), ret +} + +// nvml.DeviceGetTotalEccErrors() +func (l *library) DeviceGetTotalEccErrors(device Device, errorType MemoryErrorType, counterType EccCounterType) (uint64, Return) { + return device.GetTotalEccErrors(errorType, counterType) +} + +func (device nvmlDevice) GetTotalEccErrors(errorType MemoryErrorType, counterType EccCounterType) (uint64, Return) { + var eccCounts uint64 + ret := nvmlDeviceGetTotalEccErrors(device, errorType, counterType, &eccCounts) + return eccCounts, ret +} + +// nvml.DeviceGetDetailedEccErrors() +// +// Deprecated: See DeviceGetMemoryErrorCounter +func (l *library) DeviceGetDetailedEccErrors(device Device, errorType MemoryErrorType, counterType EccCounterType) (EccErrorCounts, Return) { + return device.GetDetailedEccErrors(errorType, counterType) +} + +// Deprecated: See DeviceGetMemoryErrorCounter +func (device nvmlDevice) GetDetailedEccErrors(errorType MemoryErrorType, counterType EccCounterType) (EccErrorCounts, Return) { + var eccCounts EccErrorCounts + ret := nvmlDeviceGetDetailedEccErrors(device, errorType, counterType, &eccCounts) + return eccCounts, ret +} + +// nvml.DeviceGetMemoryErrorCounter() +func (l *library) DeviceGetMemoryErrorCounter(device Device, errorType MemoryErrorType, counterType EccCounterType, locationType MemoryLocation) (uint64, Return) { + return device.GetMemoryErrorCounter(errorType, counterType, locationType) +} + +func (device nvmlDevice) GetMemoryErrorCounter(errorType MemoryErrorType, counterType EccCounterType, locationType MemoryLocation) (uint64, Return) { + var count uint64 + ret := nvmlDeviceGetMemoryErrorCounter(device, errorType, counterType, locationType, &count) + return count, ret +} + +// nvml.DeviceGetUtilizationRates() +func (l *library) DeviceGetUtilizationRates(device Device) (Utilization, Return) { + return device.GetUtilizationRates() +} + +func (device nvmlDevice) GetUtilizationRates() (Utilization, Return) { + var utilization Utilization + ret := nvmlDeviceGetUtilizationRates(device, &utilization) + return utilization, ret +} + +// nvml.DeviceGetEncoderUtilization() +func (l *library) DeviceGetEncoderUtilization(device Device) (uint32, uint32, Return) { + return device.GetEncoderUtilization() +} + +func (device nvmlDevice) GetEncoderUtilization() (uint32, uint32, Return) { + var utilization, samplingPeriodUs uint32 + ret := nvmlDeviceGetEncoderUtilization(device, &utilization, &samplingPeriodUs) + return utilization, samplingPeriodUs, ret +} + +// nvml.DeviceGetEncoderCapacity() +func (l *library) DeviceGetEncoderCapacity(device Device, encoderQueryType EncoderType) (int, Return) { + return device.GetEncoderCapacity(encoderQueryType) +} + +func (device nvmlDevice) GetEncoderCapacity(encoderQueryType EncoderType) (int, Return) { + var encoderCapacity uint32 + ret := nvmlDeviceGetEncoderCapacity(device, encoderQueryType, &encoderCapacity) + return int(encoderCapacity), ret +} + +// nvml.DeviceGetEncoderStats() +func (l *library) DeviceGetEncoderStats(device Device) (int, uint32, uint32, Return) { + return device.GetEncoderStats() +} + +func (device nvmlDevice) GetEncoderStats() (int, uint32, uint32, Return) { + var sessionCount, averageFps, averageLatency uint32 + ret := nvmlDeviceGetEncoderStats(device, &sessionCount, &averageFps, &averageLatency) + return int(sessionCount), averageFps, averageLatency, ret +} + +// nvml.DeviceGetEncoderSessions() +func (l *library) DeviceGetEncoderSessions(device Device) ([]EncoderSessionInfo, Return) { + return device.GetEncoderSessions() +} + +func (device nvmlDevice) GetEncoderSessions() ([]EncoderSessionInfo, Return) { + var sessionCount uint32 = 1 // Will be reduced upon returning + for { + sessionInfos := make([]EncoderSessionInfo, sessionCount) + ret := nvmlDeviceGetEncoderSessions(device, &sessionCount, &sessionInfos[0]) + if ret == SUCCESS { + return sessionInfos[:sessionCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + sessionCount *= 2 + } +} + +// nvml.DeviceGetDecoderUtilization() +func (l *library) DeviceGetDecoderUtilization(device Device) (uint32, uint32, Return) { + return device.GetDecoderUtilization() +} + +func (device nvmlDevice) GetDecoderUtilization() (uint32, uint32, Return) { + var utilization, samplingPeriodUs uint32 + ret := nvmlDeviceGetDecoderUtilization(device, &utilization, &samplingPeriodUs) + return utilization, samplingPeriodUs, ret +} + +// nvml.DeviceGetFBCStats() +func (l *library) DeviceGetFBCStats(device Device) (FBCStats, Return) { + return device.GetFBCStats() +} + +func (device nvmlDevice) GetFBCStats() (FBCStats, Return) { + var fbcStats FBCStats + ret := nvmlDeviceGetFBCStats(device, &fbcStats) + return fbcStats, ret +} + +// nvml.DeviceGetFBCSessions() +func (l *library) DeviceGetFBCSessions(device Device) ([]FBCSessionInfo, Return) { + return device.GetFBCSessions() +} + +func (device nvmlDevice) GetFBCSessions() ([]FBCSessionInfo, Return) { + var sessionCount uint32 = 1 // Will be reduced upon returning + for { + sessionInfo := make([]FBCSessionInfo, sessionCount) + ret := nvmlDeviceGetFBCSessions(device, &sessionCount, &sessionInfo[0]) + if ret == SUCCESS { + return sessionInfo[:sessionCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + sessionCount *= 2 + } +} + +// nvml.DeviceGetDriverModel() +func (l *library) DeviceGetDriverModel(device Device) (DriverModel, DriverModel, Return) { + return device.GetDriverModel() +} + +func (device nvmlDevice) GetDriverModel() (DriverModel, DriverModel, Return) { + var current, pending DriverModel + ret := nvmlDeviceGetDriverModel(device, ¤t, &pending) + return current, pending, ret +} + +// nvml.DeviceGetVbiosVersion() +func (l *library) DeviceGetVbiosVersion(device Device) (string, Return) { + return device.GetVbiosVersion() +} + +func (device nvmlDevice) GetVbiosVersion() (string, Return) { + version := make([]byte, DEVICE_VBIOS_VERSION_BUFFER_SIZE) + ret := nvmlDeviceGetVbiosVersion(device, &version[0], DEVICE_VBIOS_VERSION_BUFFER_SIZE) + return string(version[:clen(version)]), ret +} + +// nvml.DeviceGetBridgeChipInfo() +func (l *library) DeviceGetBridgeChipInfo(device Device) (BridgeChipHierarchy, Return) { + return device.GetBridgeChipInfo() +} + +func (device nvmlDevice) GetBridgeChipInfo() (BridgeChipHierarchy, Return) { + var bridgeHierarchy BridgeChipHierarchy + ret := nvmlDeviceGetBridgeChipInfo(device, &bridgeHierarchy) + return bridgeHierarchy, ret +} + +// nvml.DeviceGetComputeRunningProcesses() +func deviceGetComputeRunningProcesses_v1(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo_v1, infoCount) + ret := nvmlDeviceGetComputeRunningProcesses_v1(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return ProcessInfo_v1Slice(infos[:infoCount]).ToProcessInfoSlice(), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func deviceGetComputeRunningProcesses_v2(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo_v2, infoCount) + ret := nvmlDeviceGetComputeRunningProcesses_v2(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return ProcessInfo_v2Slice(infos[:infoCount]).ToProcessInfoSlice(), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func deviceGetComputeRunningProcesses_v3(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo, infoCount) + ret := nvmlDeviceGetComputeRunningProcesses_v3(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return infos[:infoCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func (l *library) DeviceGetComputeRunningProcesses(device Device) ([]ProcessInfo, Return) { + return device.GetComputeRunningProcesses() +} + +func (device nvmlDevice) GetComputeRunningProcesses() ([]ProcessInfo, Return) { + return deviceGetComputeRunningProcesses(device) +} + +// nvml.DeviceGetGraphicsRunningProcesses() +func deviceGetGraphicsRunningProcesses_v1(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo_v1, infoCount) + ret := nvmlDeviceGetGraphicsRunningProcesses_v1(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return ProcessInfo_v1Slice(infos[:infoCount]).ToProcessInfoSlice(), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func deviceGetGraphicsRunningProcesses_v2(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo_v2, infoCount) + ret := nvmlDeviceGetGraphicsRunningProcesses_v2(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return ProcessInfo_v2Slice(infos[:infoCount]).ToProcessInfoSlice(), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func deviceGetGraphicsRunningProcesses_v3(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo, infoCount) + ret := nvmlDeviceGetGraphicsRunningProcesses_v3(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return infos[:infoCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func (l *library) DeviceGetGraphicsRunningProcesses(device Device) ([]ProcessInfo, Return) { + return device.GetGraphicsRunningProcesses() +} + +func (device nvmlDevice) GetGraphicsRunningProcesses() ([]ProcessInfo, Return) { + return deviceGetGraphicsRunningProcesses(device) +} + +// nvml.DeviceGetMPSComputeRunningProcesses() +func deviceGetMPSComputeRunningProcesses_v1(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo_v1, infoCount) + ret := nvmlDeviceGetMPSComputeRunningProcesses_v1(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return ProcessInfo_v1Slice(infos[:infoCount]).ToProcessInfoSlice(), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func deviceGetMPSComputeRunningProcesses_v2(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo_v2, infoCount) + ret := nvmlDeviceGetMPSComputeRunningProcesses_v2(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return ProcessInfo_v2Slice(infos[:infoCount]).ToProcessInfoSlice(), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func deviceGetMPSComputeRunningProcesses_v3(device nvmlDevice) ([]ProcessInfo, Return) { + var infoCount uint32 = 1 // Will be reduced upon returning + for { + infos := make([]ProcessInfo, infoCount) + ret := nvmlDeviceGetMPSComputeRunningProcesses_v3(device, &infoCount, &infos[0]) + if ret == SUCCESS { + return infos[:infoCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + infoCount *= 2 + } +} + +func (l *library) DeviceGetMPSComputeRunningProcesses(device Device) ([]ProcessInfo, Return) { + return device.GetMPSComputeRunningProcesses() +} + +func (device nvmlDevice) GetMPSComputeRunningProcesses() ([]ProcessInfo, Return) { + return deviceGetMPSComputeRunningProcesses(device) +} + +// nvml.DeviceOnSameBoard() +func (l *library) DeviceOnSameBoard(device1 Device, device2 Device) (int, Return) { + return device1.OnSameBoard(device2) +} + +func (device1 nvmlDevice) OnSameBoard(device2 Device) (int, Return) { + var onSameBoard int32 + ret := nvmlDeviceOnSameBoard(device1, nvmlDeviceHandle(device2), &onSameBoard) + return int(onSameBoard), ret +} + +// nvml.DeviceGetAPIRestriction() +func (l *library) DeviceGetAPIRestriction(device Device, apiType RestrictedAPI) (EnableState, Return) { + return device.GetAPIRestriction(apiType) +} + +func (device nvmlDevice) GetAPIRestriction(apiType RestrictedAPI) (EnableState, Return) { + var isRestricted EnableState + ret := nvmlDeviceGetAPIRestriction(device, apiType, &isRestricted) + return isRestricted, ret +} + +// nvml.DeviceGetSamples() +func (l *library) DeviceGetSamples(device Device, samplingType SamplingType, lastSeenTimestamp uint64) (ValueType, []Sample, Return) { + return device.GetSamples(samplingType, lastSeenTimestamp) +} + +func (device nvmlDevice) GetSamples(samplingType SamplingType, lastSeenTimestamp uint64) (ValueType, []Sample, Return) { + var sampleValType ValueType + var sampleCount uint32 + ret := nvmlDeviceGetSamples(device, samplingType, lastSeenTimestamp, &sampleValType, &sampleCount, nil) + if ret != SUCCESS { + return sampleValType, nil, ret + } + if sampleCount == 0 { + return sampleValType, []Sample{}, ret + } + samples := make([]Sample, sampleCount) + ret = nvmlDeviceGetSamples(device, samplingType, lastSeenTimestamp, &sampleValType, &sampleCount, &samples[0]) + return sampleValType, samples, ret +} + +// nvml.DeviceGetBAR1MemoryInfo() +func (l *library) DeviceGetBAR1MemoryInfo(device Device) (BAR1Memory, Return) { + return device.GetBAR1MemoryInfo() +} + +func (device nvmlDevice) GetBAR1MemoryInfo() (BAR1Memory, Return) { + var bar1Memory BAR1Memory + ret := nvmlDeviceGetBAR1MemoryInfo(device, &bar1Memory) + return bar1Memory, ret +} + +// nvml.DeviceGetViolationStatus() +// +// Deprecated: Use DeviceGetFieldValues instead. +func (l *library) DeviceGetViolationStatus(device Device, perfPolicyType PerfPolicyType) (ViolationTime, Return) { + return device.GetViolationStatus(perfPolicyType) +} + +// Deprecated: Use DeviceGetFieldValues instead. +func (device nvmlDevice) GetViolationStatus(perfPolicyType PerfPolicyType) (ViolationTime, Return) { + var violTime ViolationTime + ret := nvmlDeviceGetViolationStatus(device, perfPolicyType, &violTime) + return violTime, ret +} + +// nvml.DeviceGetIrqNum() +func (l *library) DeviceGetIrqNum(device Device) (int, Return) { + return device.GetIrqNum() +} + +func (device nvmlDevice) GetIrqNum() (int, Return) { + var irqNum uint32 + ret := nvmlDeviceGetIrqNum(device, &irqNum) + return int(irqNum), ret +} + +// nvml.DeviceGetNumGpuCores() +func (l *library) DeviceGetNumGpuCores(device Device) (int, Return) { + return device.GetNumGpuCores() +} + +func (device nvmlDevice) GetNumGpuCores() (int, Return) { + var numCores uint32 + ret := nvmlDeviceGetNumGpuCores(device, &numCores) + return int(numCores), ret +} + +// nvml.DeviceGetPowerSource() +func (l *library) DeviceGetPowerSource(device Device) (PowerSource, Return) { + return device.GetPowerSource() +} + +func (device nvmlDevice) GetPowerSource() (PowerSource, Return) { + var powerSource PowerSource + ret := nvmlDeviceGetPowerSource(device, &powerSource) + return powerSource, ret +} + +// nvml.DeviceGetMemoryBusWidth() +func (l *library) DeviceGetMemoryBusWidth(device Device) (uint32, Return) { + return device.GetMemoryBusWidth() +} + +func (device nvmlDevice) GetMemoryBusWidth() (uint32, Return) { + var busWidth uint32 + ret := nvmlDeviceGetMemoryBusWidth(device, &busWidth) + return busWidth, ret +} + +// nvml.DeviceGetPcieLinkMaxSpeed() +func (l *library) DeviceGetPcieLinkMaxSpeed(device Device) (uint32, Return) { + return device.GetPcieLinkMaxSpeed() +} + +func (device nvmlDevice) GetPcieLinkMaxSpeed() (uint32, Return) { + var maxSpeed uint32 + ret := nvmlDeviceGetPcieLinkMaxSpeed(device, &maxSpeed) + return maxSpeed, ret +} + +// nvml.DeviceGetAdaptiveClockInfoStatus() +func (l *library) DeviceGetAdaptiveClockInfoStatus(device Device) (uint32, Return) { + return device.GetAdaptiveClockInfoStatus() +} + +func (device nvmlDevice) GetAdaptiveClockInfoStatus() (uint32, Return) { + var adaptiveClockStatus uint32 + ret := nvmlDeviceGetAdaptiveClockInfoStatus(device, &adaptiveClockStatus) + return adaptiveClockStatus, ret +} + +// nvml.DeviceGetAccountingMode() +func (l *library) DeviceGetAccountingMode(device Device) (EnableState, Return) { + return device.GetAccountingMode() +} + +func (device nvmlDevice) GetAccountingMode() (EnableState, Return) { + var mode EnableState + ret := nvmlDeviceGetAccountingMode(device, &mode) + return mode, ret +} + +func (l *library) DeviceGetPdi(device Device) (Pdi, Return) { + return device.GetPdi() +} + +func (device nvmlDevice) GetPdi() (Pdi, Return) { + var pdi Pdi + pdi.Version = STRUCT_VERSION(pdi, 1) + ret := nvmlDeviceGetPdi(device, &pdi) + return pdi, ret +} + +// nvml.DeviceGetAccountingStats() +func (l *library) DeviceGetAccountingStats(device Device, pid uint32) (AccountingStats, Return) { + return device.GetAccountingStats(pid) +} + +func (device nvmlDevice) GetAccountingStats(pid uint32) (AccountingStats, Return) { + var stats AccountingStats + ret := nvmlDeviceGetAccountingStats(device, pid, &stats) + return stats, ret +} + +// nvml.DeviceGetAccountingPids() +func (l *library) DeviceGetAccountingPids(device Device) ([]int, Return) { + return device.GetAccountingPids() +} + +func (device nvmlDevice) GetAccountingPids() ([]int, Return) { + var count uint32 = 1 // Will be reduced upon returning + for { + pids := make([]uint32, count) + ret := nvmlDeviceGetAccountingPids(device, &count, &pids[0]) + if ret == SUCCESS { + return uint32SliceToIntSlice(pids[:count]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + count *= 2 + } +} + +// nvml.DeviceGetAccountingBufferSize() +func (l *library) DeviceGetAccountingBufferSize(device Device) (int, Return) { + return device.GetAccountingBufferSize() +} + +func (device nvmlDevice) GetAccountingBufferSize() (int, Return) { + var bufferSize uint32 + ret := nvmlDeviceGetAccountingBufferSize(device, &bufferSize) + return int(bufferSize), ret +} + +// nvml.DeviceGetRetiredPages() +func (l *library) DeviceGetRetiredPages(device Device, cause PageRetirementCause) ([]uint64, Return) { + return device.GetRetiredPages(cause) +} + +func (device nvmlDevice) GetRetiredPages(cause PageRetirementCause) ([]uint64, Return) { + var pageCount uint32 = 1 // Will be reduced upon returning + for { + addresses := make([]uint64, pageCount) + ret := nvmlDeviceGetRetiredPages(device, cause, &pageCount, &addresses[0]) + if ret == SUCCESS { + return addresses[:pageCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + pageCount *= 2 + } +} + +// nvml.DeviceGetRetiredPages_v2() +func (l *library) DeviceGetRetiredPages_v2(device Device, cause PageRetirementCause) ([]uint64, []uint64, Return) { + return device.GetRetiredPages_v2(cause) +} + +func (device nvmlDevice) GetRetiredPages_v2(cause PageRetirementCause) ([]uint64, []uint64, Return) { + var pageCount uint32 = 1 // Will be reduced upon returning + for { + addresses := make([]uint64, pageCount) + timestamps := make([]uint64, pageCount) + ret := nvmlDeviceGetRetiredPages_v2(device, cause, &pageCount, &addresses[0], ×tamps[0]) + if ret == SUCCESS { + return addresses[:pageCount], timestamps[:pageCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, nil, ret + } + pageCount *= 2 + } +} + +// nvml.DeviceGetRetiredPagesPendingStatus() +func (l *library) DeviceGetRetiredPagesPendingStatus(device Device) (EnableState, Return) { + return device.GetRetiredPagesPendingStatus() +} + +func (device nvmlDevice) GetRetiredPagesPendingStatus() (EnableState, Return) { + var isPending EnableState + ret := nvmlDeviceGetRetiredPagesPendingStatus(device, &isPending) + return isPending, ret +} + +// nvml.DeviceSetPersistenceMode() +func (l *library) DeviceSetPersistenceMode(device Device, mode EnableState) Return { + return device.SetPersistenceMode(mode) +} + +func (device nvmlDevice) SetPersistenceMode(mode EnableState) Return { + return nvmlDeviceSetPersistenceMode(device, mode) +} + +// nvml.DeviceSetComputeMode() +func (l *library) DeviceSetComputeMode(device Device, mode ComputeMode) Return { + return device.SetComputeMode(mode) +} + +func (device nvmlDevice) SetComputeMode(mode ComputeMode) Return { + return nvmlDeviceSetComputeMode(device, mode) +} + +// nvml.DeviceSetEccMode() +func (l *library) DeviceSetEccMode(device Device, ecc EnableState) Return { + return device.SetEccMode(ecc) +} + +func (device nvmlDevice) SetEccMode(ecc EnableState) Return { + return nvmlDeviceSetEccMode(device, ecc) +} + +// nvml.DeviceClearEccErrorCounts() +func (l *library) DeviceClearEccErrorCounts(device Device, counterType EccCounterType) Return { + return device.ClearEccErrorCounts(counterType) +} + +func (device nvmlDevice) ClearEccErrorCounts(counterType EccCounterType) Return { + return nvmlDeviceClearEccErrorCounts(device, counterType) +} + +// nvml.DeviceSetDriverModel() +func (l *library) DeviceSetDriverModel(device Device, driverModel DriverModel, flags uint32) Return { + return device.SetDriverModel(driverModel, flags) +} + +func (device nvmlDevice) SetDriverModel(driverModel DriverModel, flags uint32) Return { + return nvmlDeviceSetDriverModel(device, driverModel, flags) +} + +// nvml.DeviceSetGpuLockedClocks() +func (l *library) DeviceSetGpuLockedClocks(device Device, minGpuClockMHz uint32, maxGpuClockMHz uint32) Return { + return device.SetGpuLockedClocks(minGpuClockMHz, maxGpuClockMHz) +} + +func (device nvmlDevice) SetGpuLockedClocks(minGpuClockMHz uint32, maxGpuClockMHz uint32) Return { + return nvmlDeviceSetGpuLockedClocks(device, minGpuClockMHz, maxGpuClockMHz) +} + +// nvml.DeviceResetGpuLockedClocks() +func (l *library) DeviceResetGpuLockedClocks(device Device) Return { + return device.ResetGpuLockedClocks() +} + +func (device nvmlDevice) ResetGpuLockedClocks() Return { + return nvmlDeviceResetGpuLockedClocks(device) +} + +// nvmlDeviceSetMemoryLockedClocks() +func (l *library) DeviceSetMemoryLockedClocks(device Device, minMemClockMHz uint32, maxMemClockMHz uint32) Return { + return device.SetMemoryLockedClocks(minMemClockMHz, maxMemClockMHz) +} + +func (device nvmlDevice) SetMemoryLockedClocks(minMemClockMHz uint32, maxMemClockMHz uint32) Return { + return nvmlDeviceSetMemoryLockedClocks(device, minMemClockMHz, maxMemClockMHz) +} + +// nvmlDeviceResetMemoryLockedClocks() +func (l *library) DeviceResetMemoryLockedClocks(device Device) Return { + return device.ResetMemoryLockedClocks() +} + +func (device nvmlDevice) ResetMemoryLockedClocks() Return { + return nvmlDeviceResetMemoryLockedClocks(device) +} + +// nvml.DeviceGetClkMonStatus() +func (l *library) DeviceGetClkMonStatus(device Device) (ClkMonStatus, Return) { + return device.GetClkMonStatus() +} + +func (device nvmlDevice) GetClkMonStatus() (ClkMonStatus, Return) { + var status ClkMonStatus + ret := nvmlDeviceGetClkMonStatus(device, &status) + return status, ret +} + +// nvml.DeviceSetApplicationsClocks() +// +// Deprecated: Use DeviceSetMemoryLockedClocks for Memory Clocks and DeviceSetGpuLockedClocks for Graphics Clocks instead +func (l *library) DeviceSetApplicationsClocks(device Device, memClockMHz uint32, graphicsClockMHz uint32) Return { + return device.SetApplicationsClocks(memClockMHz, graphicsClockMHz) +} + +// Deprecated: Use DeviceSetMemoryLockedClocks for Memory Clocks and DeviceSetGpuLockedClocks for Graphics Clocks instead +func (device nvmlDevice) SetApplicationsClocks(memClockMHz uint32, graphicsClockMHz uint32) Return { + return nvmlDeviceSetApplicationsClocks(device, memClockMHz, graphicsClockMHz) +} + +// nvml.DeviceSetPowerManagementLimit() +func (l *library) DeviceSetPowerManagementLimit(device Device, limit uint32) Return { + return device.SetPowerManagementLimit(limit) +} + +func (device nvmlDevice) SetPowerManagementLimit(limit uint32) Return { + return nvmlDeviceSetPowerManagementLimit(device, limit) +} + +// nvml.DeviceSetPowerManagementLimit_v2() +func (l *library) DeviceSetPowerManagementLimit_v2(device Device, powerValue *PowerValue_v2) Return { + return device.SetPowerManagementLimit_v2(powerValue) +} + +func (device nvmlDevice) SetPowerManagementLimit_v2(powerValue *PowerValue_v2) Return { + return nvmlDeviceSetPowerManagementLimit_v2(device, powerValue) +} + +// nvml.DeviceSetGpuOperationMode() +func (l *library) DeviceSetGpuOperationMode(device Device, mode GpuOperationMode) Return { + return device.SetGpuOperationMode(mode) +} + +func (device nvmlDevice) SetGpuOperationMode(mode GpuOperationMode) Return { + return nvmlDeviceSetGpuOperationMode(device, mode) +} + +// nvml.DeviceSetAPIRestriction() +func (l *library) DeviceSetAPIRestriction(device Device, apiType RestrictedAPI, isRestricted EnableState) Return { + return device.SetAPIRestriction(apiType, isRestricted) +} + +func (device nvmlDevice) SetAPIRestriction(apiType RestrictedAPI, isRestricted EnableState) Return { + return nvmlDeviceSetAPIRestriction(device, apiType, isRestricted) +} + +// nvml.DeviceSetAccountingMode() +func (l *library) DeviceSetAccountingMode(device Device, mode EnableState) Return { + return device.SetAccountingMode(mode) +} + +func (device nvmlDevice) SetAccountingMode(mode EnableState) Return { + return nvmlDeviceSetAccountingMode(device, mode) +} + +// nvml.DeviceClearAccountingPids() +func (l *library) DeviceClearAccountingPids(device Device) Return { + return device.ClearAccountingPids() +} + +func (device nvmlDevice) ClearAccountingPids() Return { + return nvmlDeviceClearAccountingPids(device) +} + +// nvml.DeviceGetNvLinkState() +func (l *library) DeviceGetNvLinkState(device Device, link int) (EnableState, Return) { + return device.GetNvLinkState(link) +} + +func (device nvmlDevice) GetNvLinkState(link int) (EnableState, Return) { + var isActive EnableState + ret := nvmlDeviceGetNvLinkState(device, uint32(link), &isActive) + return isActive, ret +} + +// nvml.DeviceGetNvLinkVersion() +func (l *library) DeviceGetNvLinkVersion(device Device, link int) (uint32, Return) { + return device.GetNvLinkVersion(link) +} + +func (device nvmlDevice) GetNvLinkVersion(link int) (uint32, Return) { + var version uint32 + ret := nvmlDeviceGetNvLinkVersion(device, uint32(link), &version) + return version, ret +} + +// nvml.DeviceGetNvLinkCapability() +func (l *library) DeviceGetNvLinkCapability(device Device, link int, capability NvLinkCapability) (uint32, Return) { + return device.GetNvLinkCapability(link, capability) +} + +func (device nvmlDevice) GetNvLinkCapability(link int, capability NvLinkCapability) (uint32, Return) { + var capResult uint32 + ret := nvmlDeviceGetNvLinkCapability(device, uint32(link), capability, &capResult) + return capResult, ret +} + +// nvml.DeviceGetNvLinkRemotePciInfo() +func (l *library) DeviceGetNvLinkRemotePciInfo(device Device, link int) (PciInfo, Return) { + return device.GetNvLinkRemotePciInfo(link) +} + +func (device nvmlDevice) GetNvLinkRemotePciInfo(link int) (PciInfo, Return) { + var pci PciInfo + ret := nvmlDeviceGetNvLinkRemotePciInfo(device, uint32(link), &pci) + return pci, ret +} + +// nvml.DeviceGetNvLinkErrorCounter() +func (l *library) DeviceGetNvLinkErrorCounter(device Device, link int, counter NvLinkErrorCounter) (uint64, Return) { + return device.GetNvLinkErrorCounter(link, counter) +} + +func (device nvmlDevice) GetNvLinkErrorCounter(link int, counter NvLinkErrorCounter) (uint64, Return) { + var counterValue uint64 + ret := nvmlDeviceGetNvLinkErrorCounter(device, uint32(link), counter, &counterValue) + return counterValue, ret +} + +// nvml.DeviceResetNvLinkErrorCounters() +func (l *library) DeviceResetNvLinkErrorCounters(device Device, link int) Return { + return device.ResetNvLinkErrorCounters(link) +} + +func (device nvmlDevice) ResetNvLinkErrorCounters(link int) Return { + return nvmlDeviceResetNvLinkErrorCounters(device, uint32(link)) +} + +// nvml.DeviceSetNvLinkUtilizationControl() +// +// Deprecated: Setting utilization counter control is no longer supported. +func (l *library) DeviceSetNvLinkUtilizationControl(device Device, link int, counter int, control *NvLinkUtilizationControl, reset bool) Return { + return device.SetNvLinkUtilizationControl(link, counter, control, reset) +} + +// Deprecated: Setting utilization counter control is no longer supported. +func (device nvmlDevice) SetNvLinkUtilizationControl(link int, counter int, control *NvLinkUtilizationControl, reset bool) Return { + resetValue := uint32(0) + if reset { + resetValue = 1 + } + return nvmlDeviceSetNvLinkUtilizationControl(device, uint32(link), uint32(counter), control, resetValue) +} + +// nvml.DeviceGetNvLinkUtilizationControl() +// +// Deprecated: Getting utilization counter control is no longer supported. +func (l *library) DeviceGetNvLinkUtilizationControl(device Device, link int, counter int) (NvLinkUtilizationControl, Return) { + return device.GetNvLinkUtilizationControl(link, counter) +} + +// Deprecated: Getting utilization counter control is no longer supported. +func (device nvmlDevice) GetNvLinkUtilizationControl(link int, counter int) (NvLinkUtilizationControl, Return) { + var control NvLinkUtilizationControl + ret := nvmlDeviceGetNvLinkUtilizationControl(device, uint32(link), uint32(counter), &control) + return control, ret +} + +// nvml.DeviceGetNvLinkUtilizationCounter() +// +// Deprecated: Use DeviceGetFieldValues with NVML_FI_DEV_NVLINK_THROUGHPUT_* as field values instead. +func (l *library) DeviceGetNvLinkUtilizationCounter(device Device, link int, counter int) (uint64, uint64, Return) { + return device.GetNvLinkUtilizationCounter(link, counter) +} + +// Deprecated: Use DeviceGetFieldValues with NVML_FI_DEV_NVLINK_THROUGHPUT_* as field values instead. +func (device nvmlDevice) GetNvLinkUtilizationCounter(link int, counter int) (uint64, uint64, Return) { + var rxCounter, txCounter uint64 + ret := nvmlDeviceGetNvLinkUtilizationCounter(device, uint32(link), uint32(counter), &rxCounter, &txCounter) + return rxCounter, txCounter, ret +} + +// nvml.DeviceFreezeNvLinkUtilizationCounter() +// +// Deprecated: Freezing NVLINK utilization counters is no longer supported. +func (l *library) DeviceFreezeNvLinkUtilizationCounter(device Device, link int, counter int, freeze EnableState) Return { + return device.FreezeNvLinkUtilizationCounter(link, counter, freeze) +} + +// Deprecated: Freezing NVLINK utilization counters is no longer supported. +func (device nvmlDevice) FreezeNvLinkUtilizationCounter(link int, counter int, freeze EnableState) Return { + return nvmlDeviceFreezeNvLinkUtilizationCounter(device, uint32(link), uint32(counter), freeze) +} + +// nvml.DeviceResetNvLinkUtilizationCounter() +// +// Deprecated: Resetting NVLINK utilization counters is no longer supported. +func (l *library) DeviceResetNvLinkUtilizationCounter(device Device, link int, counter int) Return { + return device.ResetNvLinkUtilizationCounter(link, counter) +} + +// Deprecated: Resetting NVLINK utilization counters is no longer supported. +func (device nvmlDevice) ResetNvLinkUtilizationCounter(link int, counter int) Return { + return nvmlDeviceResetNvLinkUtilizationCounter(device, uint32(link), uint32(counter)) +} + +// nvml.DeviceGetNvLinkRemoteDeviceType() +func (l *library) DeviceGetNvLinkRemoteDeviceType(device Device, link int) (IntNvLinkDeviceType, Return) { + return device.GetNvLinkRemoteDeviceType(link) +} + +func (device nvmlDevice) GetNvLinkRemoteDeviceType(link int) (IntNvLinkDeviceType, Return) { + var nvLinkDeviceType IntNvLinkDeviceType + ret := nvmlDeviceGetNvLinkRemoteDeviceType(device, uint32(link), &nvLinkDeviceType) + return nvLinkDeviceType, ret +} + +// nvml.DeviceRegisterEvents() +func (l *library) DeviceRegisterEvents(device Device, eventTypes uint64, set EventSet) Return { + return device.RegisterEvents(eventTypes, set) +} + +func (device nvmlDevice) RegisterEvents(eventTypes uint64, set EventSet) Return { + return nvmlDeviceRegisterEvents(device, eventTypes, set.(nvmlEventSet)) +} + +// nvmlDeviceGetSupportedEventTypes() +func (l *library) DeviceGetSupportedEventTypes(device Device) (uint64, Return) { + return device.GetSupportedEventTypes() +} + +func (device nvmlDevice) GetSupportedEventTypes() (uint64, Return) { + var eventTypes uint64 + ret := nvmlDeviceGetSupportedEventTypes(device, &eventTypes) + return eventTypes, ret +} + +// nvml.DeviceModifyDrainState() +func (l *library) DeviceModifyDrainState(pciInfo *PciInfo, newState EnableState) Return { + return nvmlDeviceModifyDrainState(pciInfo, newState) +} + +// nvml.DeviceQueryDrainState() +func (l *library) DeviceQueryDrainState(pciInfo *PciInfo) (EnableState, Return) { + var currentState EnableState + ret := nvmlDeviceQueryDrainState(pciInfo, ¤tState) + return currentState, ret +} + +// nvml.DeviceRemoveGpu() +func (l *library) DeviceRemoveGpu(pciInfo *PciInfo) Return { + return nvmlDeviceRemoveGpu(pciInfo) +} + +// nvml.DeviceRemoveGpu_v2() +func (l *library) DeviceRemoveGpu_v2(pciInfo *PciInfo, gpuState DetachGpuState, linkState PcieLinkState) Return { + return nvmlDeviceRemoveGpu_v2(pciInfo, gpuState, linkState) +} + +// nvml.DeviceDiscoverGpus() +func (l *library) DeviceDiscoverGpus() (PciInfo, Return) { + var pciInfo PciInfo + ret := nvmlDeviceDiscoverGpus(&pciInfo) + return pciInfo, ret +} + +// nvml.DeviceGetFieldValues() +func (l *library) DeviceGetFieldValues(device Device, values []FieldValue) Return { + return device.GetFieldValues(values) +} + +func (device nvmlDevice) GetFieldValues(values []FieldValue) Return { + valuesCount := len(values) + return nvmlDeviceGetFieldValues(device, int32(valuesCount), &values[0]) +} + +// nvml.DeviceGetVirtualizationMode() +func (l *library) DeviceGetVirtualizationMode(device Device) (GpuVirtualizationMode, Return) { + return device.GetVirtualizationMode() +} + +func (device nvmlDevice) GetVirtualizationMode() (GpuVirtualizationMode, Return) { + var pVirtualMode GpuVirtualizationMode + ret := nvmlDeviceGetVirtualizationMode(device, &pVirtualMode) + return pVirtualMode, ret +} + +// nvml.DeviceGetHostVgpuMode() +func (l *library) DeviceGetHostVgpuMode(device Device) (HostVgpuMode, Return) { + return device.GetHostVgpuMode() +} + +func (device nvmlDevice) GetHostVgpuMode() (HostVgpuMode, Return) { + var pHostVgpuMode HostVgpuMode + ret := nvmlDeviceGetHostVgpuMode(device, &pHostVgpuMode) + return pHostVgpuMode, ret +} + +// nvml.DeviceSetVirtualizationMode() +func (l *library) DeviceSetVirtualizationMode(device Device, virtualMode GpuVirtualizationMode) Return { + return device.SetVirtualizationMode(virtualMode) +} + +func (device nvmlDevice) SetVirtualizationMode(virtualMode GpuVirtualizationMode) Return { + return nvmlDeviceSetVirtualizationMode(device, virtualMode) +} + +// nvml.DeviceGetGridLicensableFeatures() +func (l *library) DeviceGetGridLicensableFeatures(device Device) (GridLicensableFeatures, Return) { + return device.GetGridLicensableFeatures() +} + +func (device nvmlDevice) GetGridLicensableFeatures() (GridLicensableFeatures, Return) { + var pGridLicensableFeatures GridLicensableFeatures + ret := nvmlDeviceGetGridLicensableFeatures(device, &pGridLicensableFeatures) + return pGridLicensableFeatures, ret +} + +// nvml.DeviceGetProcessUtilization() +func (l *library) DeviceGetProcessUtilization(device Device, lastSeenTimestamp uint64) ([]ProcessUtilizationSample, Return) { + return device.GetProcessUtilization(lastSeenTimestamp) +} + +func (device nvmlDevice) GetProcessUtilization(lastSeenTimestamp uint64) ([]ProcessUtilizationSample, Return) { + var processSamplesCount uint32 + ret := nvmlDeviceGetProcessUtilization(device, nil, &processSamplesCount, lastSeenTimestamp) + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + if processSamplesCount == 0 { + return []ProcessUtilizationSample{}, ret + } + utilization := make([]ProcessUtilizationSample, processSamplesCount) + ret = nvmlDeviceGetProcessUtilization(device, &utilization[0], &processSamplesCount, lastSeenTimestamp) + return utilization[:processSamplesCount], ret +} + +// nvml.DeviceGetSupportedVgpus() +func (l *library) DeviceGetSupportedVgpus(device Device) ([]VgpuTypeId, Return) { + return device.GetSupportedVgpus() +} + +func (device nvmlDevice) GetSupportedVgpus() ([]VgpuTypeId, Return) { + var vgpuCount uint32 = 1 // Will be reduced upon returning + for { + vgpuTypeIds := make([]nvmlVgpuTypeId, vgpuCount) + ret := nvmlDeviceGetSupportedVgpus(device, &vgpuCount, &vgpuTypeIds[0]) + if ret == SUCCESS { + return convertSlice[nvmlVgpuTypeId, VgpuTypeId](vgpuTypeIds[:vgpuCount]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + vgpuCount *= 2 + } +} + +// nvml.DeviceGetCreatableVgpus() +func (l *library) DeviceGetCreatableVgpus(device Device) ([]VgpuTypeId, Return) { + return device.GetCreatableVgpus() +} + +func (device nvmlDevice) GetCreatableVgpus() ([]VgpuTypeId, Return) { + var vgpuCount uint32 = 1 // Will be reduced upon returning + for { + vgpuTypeIds := make([]nvmlVgpuTypeId, vgpuCount) + ret := nvmlDeviceGetCreatableVgpus(device, &vgpuCount, &vgpuTypeIds[0]) + if ret == SUCCESS { + return convertSlice[nvmlVgpuTypeId, VgpuTypeId](vgpuTypeIds[:vgpuCount]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + vgpuCount *= 2 + } +} + +// nvml.DeviceGetActiveVgpus() +func (l *library) DeviceGetActiveVgpus(device Device) ([]VgpuInstance, Return) { + return device.GetActiveVgpus() +} + +func (device nvmlDevice) GetActiveVgpus() ([]VgpuInstance, Return) { + var vgpuCount uint32 = 1 // Will be reduced upon returning + for { + vgpuInstances := make([]nvmlVgpuInstance, vgpuCount) + ret := nvmlDeviceGetActiveVgpus(device, &vgpuCount, &vgpuInstances[0]) + if ret == SUCCESS { + return convertSlice[nvmlVgpuInstance, VgpuInstance](vgpuInstances[:vgpuCount]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + vgpuCount *= 2 + } +} + +// nvml.DeviceGetVgpuMetadata() +func (l *library) DeviceGetVgpuMetadata(device Device) (VgpuPgpuMetadata, Return) { + return device.GetVgpuMetadata() +} + +func (device nvmlDevice) GetVgpuMetadata() (VgpuPgpuMetadata, Return) { + var vgpuPgpuMetadata VgpuPgpuMetadata + opaqueDataSize := unsafe.Sizeof(vgpuPgpuMetadata.nvmlVgpuPgpuMetadata.OpaqueData) + vgpuPgpuMetadataSize := unsafe.Sizeof(vgpuPgpuMetadata.nvmlVgpuPgpuMetadata) - opaqueDataSize + for { + bufferSize := uint32(vgpuPgpuMetadataSize + opaqueDataSize) + buffer := make([]byte, bufferSize) + nvmlVgpuPgpuMetadataPtr := (*nvmlVgpuPgpuMetadata)(unsafe.Pointer(&buffer[0])) + ret := nvmlDeviceGetVgpuMetadata(device, nvmlVgpuPgpuMetadataPtr, &bufferSize) + if ret == SUCCESS { + vgpuPgpuMetadata.nvmlVgpuPgpuMetadata = *nvmlVgpuPgpuMetadataPtr + vgpuPgpuMetadata.OpaqueData = buffer[vgpuPgpuMetadataSize:bufferSize] + return vgpuPgpuMetadata, ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return vgpuPgpuMetadata, ret + } + opaqueDataSize = 2 * opaqueDataSize + } +} + +// nvml.DeviceGetPgpuMetadataString() +func (l *library) DeviceGetPgpuMetadataString(device Device) (string, Return) { + return device.GetPgpuMetadataString() +} + +func (device nvmlDevice) GetPgpuMetadataString() (string, Return) { + var bufferSize uint32 = 1 // Will be reduced upon returning + for { + pgpuMetadata := make([]byte, bufferSize) + ret := nvmlDeviceGetPgpuMetadataString(device, &pgpuMetadata[0], &bufferSize) + if ret == SUCCESS { + return string(pgpuMetadata[:clen(pgpuMetadata)]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return "", ret + } + bufferSize *= 2 + } +} + +// nvml.DeviceGetVgpuUtilization() +func (l *library) DeviceGetVgpuUtilization(device Device, lastSeenTimestamp uint64) (ValueType, []VgpuInstanceUtilizationSample, Return) { + return device.GetVgpuUtilization(lastSeenTimestamp) +} + +func (device nvmlDevice) GetVgpuUtilization(lastSeenTimestamp uint64) (ValueType, []VgpuInstanceUtilizationSample, Return) { + var sampleValType ValueType + var vgpuInstanceSamplesCount uint32 = 1 // Will be reduced upon returning + for { + utilizationSamples := make([]VgpuInstanceUtilizationSample, vgpuInstanceSamplesCount) + ret := nvmlDeviceGetVgpuUtilization(device, lastSeenTimestamp, &sampleValType, &vgpuInstanceSamplesCount, &utilizationSamples[0]) + if ret == SUCCESS { + return sampleValType, utilizationSamples[:vgpuInstanceSamplesCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return sampleValType, nil, ret + } + vgpuInstanceSamplesCount *= 2 + } +} + +// nvml.DeviceGetAttributes() +func (l *library) DeviceGetAttributes(device Device) (DeviceAttributes, Return) { + return device.GetAttributes() +} + +func (device nvmlDevice) GetAttributes() (DeviceAttributes, Return) { + var attributes DeviceAttributes + ret := nvmlDeviceGetAttributes(device, &attributes) + return attributes, ret +} + +// nvml.DeviceGetRemappedRows() +func (l *library) DeviceGetRemappedRows(device Device) (int, int, bool, bool, Return) { + return device.GetRemappedRows() +} + +func (device nvmlDevice) GetRemappedRows() (int, int, bool, bool, Return) { + var corrRows, uncRows, isPending, failureOccured uint32 + ret := nvmlDeviceGetRemappedRows(device, &corrRows, &uncRows, &isPending, &failureOccured) + return int(corrRows), int(uncRows), (isPending != 0), (failureOccured != 0), ret +} + +// nvml.DeviceGetRowRemapperHistogram() +func (l *library) DeviceGetRowRemapperHistogram(device Device) (RowRemapperHistogramValues, Return) { + return device.GetRowRemapperHistogram() +} + +func (device nvmlDevice) GetRowRemapperHistogram() (RowRemapperHistogramValues, Return) { + var values RowRemapperHistogramValues + ret := nvmlDeviceGetRowRemapperHistogram(device, &values) + return values, ret +} + +// nvml.DeviceGetArchitecture() +func (l *library) DeviceGetArchitecture(device Device) (DeviceArchitecture, Return) { + return device.GetArchitecture() +} + +func (device nvmlDevice) GetArchitecture() (DeviceArchitecture, Return) { + var arch DeviceArchitecture + ret := nvmlDeviceGetArchitecture(device, &arch) + return arch, ret +} + +// nvml.DeviceGetVgpuProcessUtilization() +func (l *library) DeviceGetVgpuProcessUtilization(device Device, lastSeenTimestamp uint64) ([]VgpuProcessUtilizationSample, Return) { + return device.GetVgpuProcessUtilization(lastSeenTimestamp) +} + +func (device nvmlDevice) GetVgpuProcessUtilization(lastSeenTimestamp uint64) ([]VgpuProcessUtilizationSample, Return) { + var vgpuProcessSamplesCount uint32 = 1 // Will be reduced upon returning + for { + utilizationSamples := make([]VgpuProcessUtilizationSample, vgpuProcessSamplesCount) + ret := nvmlDeviceGetVgpuProcessUtilization(device, lastSeenTimestamp, &vgpuProcessSamplesCount, &utilizationSamples[0]) + if ret == SUCCESS { + return utilizationSamples[:vgpuProcessSamplesCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + vgpuProcessSamplesCount *= 2 + } +} + +// nvml.GetExcludedDeviceCount() +func (l *library) GetExcludedDeviceCount() (int, Return) { + var deviceCount uint32 + ret := nvmlGetExcludedDeviceCount(&deviceCount) + return int(deviceCount), ret +} + +// nvml.GetExcludedDeviceInfoByIndex() +func (l *library) GetExcludedDeviceInfoByIndex(index int) (ExcludedDeviceInfo, Return) { + var info ExcludedDeviceInfo + ret := nvmlGetExcludedDeviceInfoByIndex(uint32(index), &info) + return info, ret +} + +func (l *library) DeviceReadWritePRM_v1(device Device, buffer *PRMTLV_v1) Return { + return device.ReadWritePRM_v1(buffer) +} + +func (device nvmlDevice) ReadWritePRM_v1(buffer *PRMTLV_v1) Return { + return nvmlDeviceReadWritePRM_v1(device, buffer) +} + +// nvml.DeviceSetMigMode() +func (l *library) DeviceSetMigMode(device Device, mode int) (Return, Return) { + return device.SetMigMode(mode) +} + +func (device nvmlDevice) SetMigMode(mode int) (Return, Return) { + var activationStatus Return + ret := nvmlDeviceSetMigMode(device, uint32(mode), &activationStatus) + return activationStatus, ret +} + +// nvml.DeviceGetMigMode() +func (l *library) DeviceGetMigMode(device Device) (int, int, Return) { + return device.GetMigMode() +} + +func (device nvmlDevice) GetMigMode() (int, int, Return) { + var currentMode, pendingMode uint32 + ret := nvmlDeviceGetMigMode(device, ¤tMode, &pendingMode) + return int(currentMode), int(pendingMode), ret +} + +// nvml.DeviceGetGpuInstanceProfileInfo() +func (l *library) DeviceGetGpuInstanceProfileInfo(device Device, profile int) (GpuInstanceProfileInfo, Return) { + return device.GetGpuInstanceProfileInfo(profile) +} + +func (device nvmlDevice) GetGpuInstanceProfileInfo(profile int) (GpuInstanceProfileInfo, Return) { + var info GpuInstanceProfileInfo + ret := nvmlDeviceGetGpuInstanceProfileInfo(device, uint32(profile), &info) + return info, ret +} + +// nvml.DeviceGetGpuInstanceProfileInfoV() +type GpuInstanceProfileInfoHandler struct { + device nvmlDevice + profile int +} + +func (handler GpuInstanceProfileInfoHandler) V1() (GpuInstanceProfileInfo, Return) { + return DeviceGetGpuInstanceProfileInfo(handler.device, handler.profile) +} + +func (handler GpuInstanceProfileInfoHandler) V2() (GpuInstanceProfileInfo_v2, Return) { + var info GpuInstanceProfileInfo_v2 + info.Version = STRUCT_VERSION(info, 2) + ret := nvmlDeviceGetGpuInstanceProfileInfoV(handler.device, uint32(handler.profile), &info) + return info, ret +} + +func (handler GpuInstanceProfileInfoHandler) V3() (GpuInstanceProfileInfo_v3, Return) { + var info GpuInstanceProfileInfo_v3 + info.Version = STRUCT_VERSION(info, 3) + ret := nvmlDeviceGetGpuInstanceProfileInfoV(handler.device, uint32(handler.profile), (*GpuInstanceProfileInfo_v2)(unsafe.Pointer(&info))) + return info, ret +} + +func (l *library) DeviceGetGpuInstanceProfileInfoV(device Device, profile int) GpuInstanceProfileInfoHandler { + return device.GetGpuInstanceProfileInfoV(profile) +} + +func (device nvmlDevice) GetGpuInstanceProfileInfoV(profile int) GpuInstanceProfileInfoHandler { + return GpuInstanceProfileInfoHandler{device, profile} +} + +type GpuInstanceProfileInfoByIdHandler struct { + device nvmlDevice + profileId int +} + +func (handler GpuInstanceProfileInfoByIdHandler) V2() (GpuInstanceProfileInfo_v2, Return) { + var info GpuInstanceProfileInfo_v2 + info.Version = STRUCT_VERSION(info, 2) + ret := nvmlDeviceGetGpuInstanceProfileInfoByIdV(handler.device, uint32(handler.profileId), &info) + return info, ret +} + +func (handler GpuInstanceProfileInfoByIdHandler) V3() (GpuInstanceProfileInfo_v3, Return) { + var info GpuInstanceProfileInfo_v3 + info.Version = STRUCT_VERSION(info, 3) + ret := nvmlDeviceGetGpuInstanceProfileInfoV(handler.device, uint32(handler.profileId), (*GpuInstanceProfileInfo_v2)(unsafe.Pointer(&info))) + return info, ret +} + +func (l *library) DeviceGetGpuInstanceProfileInfoByIdV(device Device, profileId int) GpuInstanceProfileInfoByIdHandler { + return device.GetGpuInstanceProfileInfoByIdV(profileId) +} + +func (device nvmlDevice) GetGpuInstanceProfileInfoByIdV(profileId int) GpuInstanceProfileInfoByIdHandler { + return GpuInstanceProfileInfoByIdHandler{device, profileId} +} + +// nvml.DeviceGetGpuInstancePossiblePlacements() +func (l *library) DeviceGetGpuInstancePossiblePlacements(device Device, info *GpuInstanceProfileInfo) ([]GpuInstancePlacement, Return) { + return device.GetGpuInstancePossiblePlacements(info) +} + +func (device nvmlDevice) GetGpuInstancePossiblePlacements(info *GpuInstanceProfileInfo) ([]GpuInstancePlacement, Return) { + if info == nil { + return nil, ERROR_INVALID_ARGUMENT + } + var count uint32 + ret := nvmlDeviceGetGpuInstancePossiblePlacements(device, info.Id, nil, &count) + if ret != SUCCESS { + return nil, ret + } + if count == 0 { + return []GpuInstancePlacement{}, ret + } + placements := make([]GpuInstancePlacement, count) + ret = nvmlDeviceGetGpuInstancePossiblePlacements(device, info.Id, &placements[0], &count) + return placements[:count], ret +} + +// nvml.DeviceGetGpuInstanceRemainingCapacity() +func (l *library) DeviceGetGpuInstanceRemainingCapacity(device Device, info *GpuInstanceProfileInfo) (int, Return) { + return device.GetGpuInstanceRemainingCapacity(info) +} + +func (device nvmlDevice) GetGpuInstanceRemainingCapacity(info *GpuInstanceProfileInfo) (int, Return) { + if info == nil { + return 0, ERROR_INVALID_ARGUMENT + } + var count uint32 + ret := nvmlDeviceGetGpuInstanceRemainingCapacity(device, info.Id, &count) + return int(count), ret +} + +// nvml.DeviceCreateGpuInstance() +func (l *library) DeviceCreateGpuInstance(device Device, info *GpuInstanceProfileInfo) (GpuInstance, Return) { + return device.CreateGpuInstance(info) +} + +func (device nvmlDevice) CreateGpuInstance(info *GpuInstanceProfileInfo) (GpuInstance, Return) { + if info == nil { + return nil, ERROR_INVALID_ARGUMENT + } + var gpuInstance nvmlGpuInstance + ret := nvmlDeviceCreateGpuInstance(device, info.Id, &gpuInstance) + return gpuInstance, ret +} + +// nvml.DeviceCreateGpuInstanceWithPlacement() +func (l *library) DeviceCreateGpuInstanceWithPlacement(device Device, info *GpuInstanceProfileInfo, placement *GpuInstancePlacement) (GpuInstance, Return) { + return device.CreateGpuInstanceWithPlacement(info, placement) +} + +func (device nvmlDevice) CreateGpuInstanceWithPlacement(info *GpuInstanceProfileInfo, placement *GpuInstancePlacement) (GpuInstance, Return) { + if info == nil { + return nil, ERROR_INVALID_ARGUMENT + } + var gpuInstance nvmlGpuInstance + ret := nvmlDeviceCreateGpuInstanceWithPlacement(device, info.Id, placement, &gpuInstance) + return gpuInstance, ret +} + +// nvml.GpuInstanceDestroy() +func (l *library) GpuInstanceDestroy(gpuInstance GpuInstance) Return { + return gpuInstance.Destroy() +} + +func (gpuInstance nvmlGpuInstance) Destroy() Return { + return nvmlGpuInstanceDestroy(gpuInstance) +} + +// nvml.DeviceGetGpuInstances() +func (l *library) DeviceGetGpuInstances(device Device, info *GpuInstanceProfileInfo) ([]GpuInstance, Return) { + return device.GetGpuInstances(info) +} + +func (device nvmlDevice) GetGpuInstances(info *GpuInstanceProfileInfo) ([]GpuInstance, Return) { + if info == nil { + return nil, ERROR_INVALID_ARGUMENT + } + var count = info.InstanceCount + gpuInstances := make([]nvmlGpuInstance, count) + ret := nvmlDeviceGetGpuInstances(device, info.Id, &gpuInstances[0], &count) + return convertSlice[nvmlGpuInstance, GpuInstance](gpuInstances[:count]), ret +} + +// nvml.DeviceGetGpuInstanceById() +func (l *library) DeviceGetGpuInstanceById(device Device, id int) (GpuInstance, Return) { + return device.GetGpuInstanceById(id) +} + +func (device nvmlDevice) GetGpuInstanceById(id int) (GpuInstance, Return) { + var gpuInstance nvmlGpuInstance + ret := nvmlDeviceGetGpuInstanceById(device, uint32(id), &gpuInstance) + return gpuInstance, ret +} + +// nvml.GpuInstanceGetInfo() +func (l *library) GpuInstanceGetInfo(gpuInstance GpuInstance) (GpuInstanceInfo, Return) { + return gpuInstance.GetInfo() +} + +func (gpuInstance nvmlGpuInstance) GetInfo() (GpuInstanceInfo, Return) { + var info nvmlGpuInstanceInfo + ret := nvmlGpuInstanceGetInfo(gpuInstance, &info) + return info.convert(), ret +} + +// nvml.GpuInstanceGetComputeInstanceProfileInfo() +func (l *library) GpuInstanceGetComputeInstanceProfileInfo(gpuInstance GpuInstance, profile int, engProfile int) (ComputeInstanceProfileInfo, Return) { + return gpuInstance.GetComputeInstanceProfileInfo(profile, engProfile) +} + +func (gpuInstance nvmlGpuInstance) GetComputeInstanceProfileInfo(profile int, engProfile int) (ComputeInstanceProfileInfo, Return) { + var info ComputeInstanceProfileInfo + ret := nvmlGpuInstanceGetComputeInstanceProfileInfo(gpuInstance, uint32(profile), uint32(engProfile), &info) + return info, ret +} + +// nvml.GpuInstanceGetComputeInstanceProfileInfoV() +type ComputeInstanceProfileInfoHandler struct { + gpuInstance nvmlGpuInstance + profile int + engProfile int +} + +func (handler ComputeInstanceProfileInfoHandler) V1() (ComputeInstanceProfileInfo, Return) { + return GpuInstanceGetComputeInstanceProfileInfo(handler.gpuInstance, handler.profile, handler.engProfile) +} + +func (handler ComputeInstanceProfileInfoHandler) V2() (ComputeInstanceProfileInfo_v2, Return) { + var info ComputeInstanceProfileInfo_v2 + info.Version = STRUCT_VERSION(info, 2) + ret := nvmlGpuInstanceGetComputeInstanceProfileInfoV(handler.gpuInstance, uint32(handler.profile), uint32(handler.engProfile), &info) + return info, ret +} + +func (handler ComputeInstanceProfileInfoHandler) V3() (ComputeInstanceProfileInfo_v3, Return) { + var info ComputeInstanceProfileInfo_v3 + info.Version = STRUCT_VERSION(info, 3) + ret := nvmlGpuInstanceGetComputeInstanceProfileInfoV(handler.gpuInstance, uint32(handler.profile), uint32(handler.engProfile), (*ComputeInstanceProfileInfo_v2)(unsafe.Pointer(&info))) + return info, ret +} + +func (l *library) GpuInstanceGetComputeInstanceProfileInfoV(gpuInstance GpuInstance, profile int, engProfile int) ComputeInstanceProfileInfoHandler { + return gpuInstance.GetComputeInstanceProfileInfoV(profile, engProfile) +} + +func (gpuInstance nvmlGpuInstance) GetComputeInstanceProfileInfoV(profile int, engProfile int) ComputeInstanceProfileInfoHandler { + return ComputeInstanceProfileInfoHandler{gpuInstance, profile, engProfile} +} + +// nvml.GpuInstanceGetComputeInstanceRemainingCapacity() +func (l *library) GpuInstanceGetComputeInstanceRemainingCapacity(gpuInstance GpuInstance, info *ComputeInstanceProfileInfo) (int, Return) { + return gpuInstance.GetComputeInstanceRemainingCapacity(info) +} + +func (gpuInstance nvmlGpuInstance) GetComputeInstanceRemainingCapacity(info *ComputeInstanceProfileInfo) (int, Return) { + if info == nil { + return 0, ERROR_INVALID_ARGUMENT + } + var count uint32 + ret := nvmlGpuInstanceGetComputeInstanceRemainingCapacity(gpuInstance, info.Id, &count) + return int(count), ret +} + +// nvml.GpuInstanceCreateComputeInstance() +func (l *library) GpuInstanceCreateComputeInstance(gpuInstance GpuInstance, info *ComputeInstanceProfileInfo) (ComputeInstance, Return) { + return gpuInstance.CreateComputeInstance(info) +} + +func (gpuInstance nvmlGpuInstance) CreateComputeInstance(info *ComputeInstanceProfileInfo) (ComputeInstance, Return) { + if info == nil { + return nil, ERROR_INVALID_ARGUMENT + } + var computeInstance nvmlComputeInstance + ret := nvmlGpuInstanceCreateComputeInstance(gpuInstance, info.Id, &computeInstance) + return computeInstance, ret +} + +// nvml.ComputeInstanceDestroy() +func (l *library) ComputeInstanceDestroy(computeInstance ComputeInstance) Return { + return computeInstance.Destroy() +} + +func (computeInstance nvmlComputeInstance) Destroy() Return { + return nvmlComputeInstanceDestroy(computeInstance) +} + +// nvml.GpuInstanceGetComputeInstances() +func (l *library) GpuInstanceGetComputeInstances(gpuInstance GpuInstance, info *ComputeInstanceProfileInfo) ([]ComputeInstance, Return) { + return gpuInstance.GetComputeInstances(info) +} + +func (gpuInstance nvmlGpuInstance) GetComputeInstances(info *ComputeInstanceProfileInfo) ([]ComputeInstance, Return) { + if info == nil { + return nil, ERROR_INVALID_ARGUMENT + } + var count = info.InstanceCount + computeInstances := make([]nvmlComputeInstance, count) + ret := nvmlGpuInstanceGetComputeInstances(gpuInstance, info.Id, &computeInstances[0], &count) + return convertSlice[nvmlComputeInstance, ComputeInstance](computeInstances[:count]), ret +} + +// nvml.GpuInstanceGetComputeInstanceById() +func (l *library) GpuInstanceGetComputeInstanceById(gpuInstance GpuInstance, id int) (ComputeInstance, Return) { + return gpuInstance.GetComputeInstanceById(id) +} + +func (gpuInstance nvmlGpuInstance) GetComputeInstanceById(id int) (ComputeInstance, Return) { + var computeInstance nvmlComputeInstance + ret := nvmlGpuInstanceGetComputeInstanceById(gpuInstance, uint32(id), &computeInstance) + return computeInstance, ret +} + +// nvml.ComputeInstanceGetInfo() +func (l *library) ComputeInstanceGetInfo(computeInstance ComputeInstance) (ComputeInstanceInfo, Return) { + return computeInstance.GetInfo() +} + +func (computeInstance nvmlComputeInstance) GetInfo() (ComputeInstanceInfo, Return) { + var info nvmlComputeInstanceInfo + ret := nvmlComputeInstanceGetInfo(computeInstance, &info) + return info.convert(), ret +} + +// nvml.DeviceIsMigDeviceHandle() +func (l *library) DeviceIsMigDeviceHandle(device Device) (bool, Return) { + return device.IsMigDeviceHandle() +} + +func (device nvmlDevice) IsMigDeviceHandle() (bool, Return) { + var isMigDevice uint32 + ret := nvmlDeviceIsMigDeviceHandle(device, &isMigDevice) + return (isMigDevice != 0), ret +} + +// nvml DeviceGetGpuInstanceId() +func (l *library) DeviceGetGpuInstanceId(device Device) (int, Return) { + return device.GetGpuInstanceId() +} + +func (device nvmlDevice) GetGpuInstanceId() (int, Return) { + var id uint32 + ret := nvmlDeviceGetGpuInstanceId(device, &id) + return int(id), ret +} + +// nvml.DeviceGetComputeInstanceId() +func (l *library) DeviceGetComputeInstanceId(device Device) (int, Return) { + return device.GetComputeInstanceId() +} + +func (device nvmlDevice) GetComputeInstanceId() (int, Return) { + var id uint32 + ret := nvmlDeviceGetComputeInstanceId(device, &id) + return int(id), ret +} + +// nvml.DeviceGetMaxMigDeviceCount() +func (l *library) DeviceGetMaxMigDeviceCount(device Device) (int, Return) { + return device.GetMaxMigDeviceCount() +} + +func (device nvmlDevice) GetMaxMigDeviceCount() (int, Return) { + var count uint32 + ret := nvmlDeviceGetMaxMigDeviceCount(device, &count) + return int(count), ret +} + +// nvml.DeviceGetMigDeviceHandleByIndex() +func (l *library) DeviceGetMigDeviceHandleByIndex(device Device, index int) (Device, Return) { + return device.GetMigDeviceHandleByIndex(index) +} + +func (device nvmlDevice) GetMigDeviceHandleByIndex(index int) (Device, Return) { + var migDevice nvmlDevice + ret := nvmlDeviceGetMigDeviceHandleByIndex(device, uint32(index), &migDevice) + return migDevice, ret +} + +// nvml.DeviceGetDeviceHandleFromMigDeviceHandle() +func (l *library) DeviceGetDeviceHandleFromMigDeviceHandle(migdevice Device) (Device, Return) { + return migdevice.GetDeviceHandleFromMigDeviceHandle() +} + +func (migDevice nvmlDevice) GetDeviceHandleFromMigDeviceHandle() (Device, Return) { + var device nvmlDevice + ret := nvmlDeviceGetDeviceHandleFromMigDeviceHandle(migDevice, &device) + return device, ret +} + +// nvml.DeviceGetBusType() +func (l *library) DeviceGetBusType(device Device) (BusType, Return) { + return device.GetBusType() +} + +func (device nvmlDevice) GetBusType() (BusType, Return) { + var busType BusType + ret := nvmlDeviceGetBusType(device, &busType) + return busType, ret +} + +// nvml.DeviceSetDefaultFanSpeed_v2() +func (l *library) DeviceSetDefaultFanSpeed_v2(device Device, fan int) Return { + return device.SetDefaultFanSpeed_v2(fan) +} + +func (device nvmlDevice) SetDefaultFanSpeed_v2(fan int) Return { + return nvmlDeviceSetDefaultFanSpeed_v2(device, uint32(fan)) +} + +// nvml.DeviceGetMinMaxFanSpeed() +func (l *library) DeviceGetMinMaxFanSpeed(device Device) (int, int, Return) { + return device.GetMinMaxFanSpeed() +} + +func (device nvmlDevice) GetMinMaxFanSpeed() (int, int, Return) { + var minSpeed, maxSpeed uint32 + ret := nvmlDeviceGetMinMaxFanSpeed(device, &minSpeed, &maxSpeed) + return int(minSpeed), int(maxSpeed), ret +} + +// nvml.DeviceGetThermalSettings() +func (l *library) DeviceGetThermalSettings(device Device, sensorIndex uint32) (GpuThermalSettings, Return) { + return device.GetThermalSettings(sensorIndex) +} + +func (device nvmlDevice) GetThermalSettings(sensorIndex uint32) (GpuThermalSettings, Return) { + var pThermalSettings GpuThermalSettings + ret := nvmlDeviceGetThermalSettings(device, sensorIndex, &pThermalSettings) + return pThermalSettings, ret +} + +// nvml.DeviceGetDefaultEccMode() +func (l *library) DeviceGetDefaultEccMode(device Device) (EnableState, Return) { + return device.GetDefaultEccMode() +} + +func (device nvmlDevice) GetDefaultEccMode() (EnableState, Return) { + var defaultMode EnableState + ret := nvmlDeviceGetDefaultEccMode(device, &defaultMode) + return defaultMode, ret +} + +// nvml.DeviceGetPcieSpeed() +func (l *library) DeviceGetPcieSpeed(device Device) (int, Return) { + return device.GetPcieSpeed() +} + +func (device nvmlDevice) GetPcieSpeed() (int, Return) { + var pcieSpeed uint32 + ret := nvmlDeviceGetPcieSpeed(device, &pcieSpeed) + return int(pcieSpeed), ret +} + +// nvml.DeviceGetGspFirmwareVersion() +func (l *library) DeviceGetGspFirmwareVersion(device Device) (string, Return) { + return device.GetGspFirmwareVersion() +} + +func (device nvmlDevice) GetGspFirmwareVersion() (string, Return) { + version := make([]byte, GSP_FIRMWARE_VERSION_BUF_SIZE) + ret := nvmlDeviceGetGspFirmwareVersion(device, &version[0]) + return string(version[:clen(version)]), ret +} + +// nvml.DeviceGetGspFirmwareMode() +func (l *library) DeviceGetGspFirmwareMode(device Device) (bool, bool, Return) { + return device.GetGspFirmwareMode() +} + +func (device nvmlDevice) GetGspFirmwareMode() (bool, bool, Return) { + var isEnabled, defaultMode uint32 + ret := nvmlDeviceGetGspFirmwareMode(device, &isEnabled, &defaultMode) + return (isEnabled != 0), (defaultMode != 0), ret +} + +// nvml.DeviceGetDynamicPstatesInfo() +func (l *library) DeviceGetDynamicPstatesInfo(device Device) (GpuDynamicPstatesInfo, Return) { + return device.GetDynamicPstatesInfo() +} + +func (device nvmlDevice) GetDynamicPstatesInfo() (GpuDynamicPstatesInfo, Return) { + var pDynamicPstatesInfo GpuDynamicPstatesInfo + ret := nvmlDeviceGetDynamicPstatesInfo(device, &pDynamicPstatesInfo) + return pDynamicPstatesInfo, ret +} + +// nvml.DeviceSetFanSpeed_v2() +func (l *library) DeviceSetFanSpeed_v2(device Device, fan int, speed int) Return { + return device.SetFanSpeed_v2(fan, speed) +} + +func (device nvmlDevice) SetFanSpeed_v2(fan int, speed int) Return { + return nvmlDeviceSetFanSpeed_v2(device, uint32(fan), uint32(speed)) +} + +// nvml.DeviceGetGpcClkVfOffset() +func (l *library) DeviceGetGpcClkVfOffset(device Device) (int, Return) { + return device.GetGpcClkVfOffset() +} + +func (device nvmlDevice) GetGpcClkVfOffset() (int, Return) { + var offset int32 + ret := nvmlDeviceGetGpcClkVfOffset(device, &offset) + return int(offset), ret +} + +// nvml.DeviceSetGpcClkVfOffset() +// +// Deprecated: Use DeviceSetClockOffsets instead. +func (l *library) DeviceSetGpcClkVfOffset(device Device, offset int) Return { + return device.SetGpcClkVfOffset(offset) +} + +// Deprecated: Use DeviceSetClockOffsets instead. +func (device nvmlDevice) SetGpcClkVfOffset(offset int) Return { + return nvmlDeviceSetGpcClkVfOffset(device, int32(offset)) +} + +// nvml.DeviceGetMinMaxClockOfPState() +func (l *library) DeviceGetMinMaxClockOfPState(device Device, clockType ClockType, pstate Pstates) (uint32, uint32, Return) { + return device.GetMinMaxClockOfPState(clockType, pstate) +} + +func (device nvmlDevice) GetMinMaxClockOfPState(clockType ClockType, pstate Pstates) (uint32, uint32, Return) { + var minClockMHz, maxClockMHz uint32 + ret := nvmlDeviceGetMinMaxClockOfPState(device, clockType, pstate, &minClockMHz, &maxClockMHz) + return minClockMHz, maxClockMHz, ret +} + +// nvml.DeviceGetSupportedPerformanceStates() +func (l *library) DeviceGetSupportedPerformanceStates(device Device) ([]Pstates, Return) { + return device.GetSupportedPerformanceStates() +} + +func (device nvmlDevice) GetSupportedPerformanceStates() ([]Pstates, Return) { + pstates := make([]Pstates, MAX_GPU_PERF_PSTATES) + ret := nvmlDeviceGetSupportedPerformanceStates(device, &pstates[0], MAX_GPU_PERF_PSTATES) + for i := 0; i < MAX_GPU_PERF_PSTATES; i++ { + if pstates[i] == PSTATE_UNKNOWN { + return pstates[0:i], ret + } + } + return pstates, ret +} + +// nvml.DeviceGetTargetFanSpeed() +func (l *library) DeviceGetTargetFanSpeed(device Device, fan int) (int, Return) { + return device.GetTargetFanSpeed(fan) +} + +func (device nvmlDevice) GetTargetFanSpeed(fan int) (int, Return) { + var targetSpeed uint32 + ret := nvmlDeviceGetTargetFanSpeed(device, uint32(fan), &targetSpeed) + return int(targetSpeed), ret +} + +// nvml.DeviceGetMemClkVfOffset() +func (l *library) DeviceGetMemClkVfOffset(device Device) (int, Return) { + return device.GetMemClkVfOffset() +} + +func (device nvmlDevice) GetMemClkVfOffset() (int, Return) { + var offset int32 + ret := nvmlDeviceGetMemClkVfOffset(device, &offset) + return int(offset), ret +} + +// nvml.DeviceSetMemClkVfOffset() +// +// Deprecated: Use DeviceSetMemClkVfOffset instead +func (l *library) DeviceSetMemClkVfOffset(device Device, offset int) Return { + return device.SetMemClkVfOffset(offset) +} + +// Deprecated: Use DeviceSetMemClkVfOffset instead +func (device nvmlDevice) SetMemClkVfOffset(offset int) Return { + return nvmlDeviceSetMemClkVfOffset(device, int32(offset)) +} + +// nvml.DeviceGetGpcClkMinMaxVfOffset() +func (l *library) DeviceGetGpcClkMinMaxVfOffset(device Device) (int, int, Return) { + return device.GetGpcClkMinMaxVfOffset() +} + +func (device nvmlDevice) GetGpcClkMinMaxVfOffset() (int, int, Return) { + var minOffset, maxOffset int32 + ret := nvmlDeviceGetGpcClkMinMaxVfOffset(device, &minOffset, &maxOffset) + return int(minOffset), int(maxOffset), ret +} + +// nvml.DeviceGetMemClkMinMaxVfOffset() +func (l *library) DeviceGetMemClkMinMaxVfOffset(device Device) (int, int, Return) { + return device.GetMemClkMinMaxVfOffset() +} + +func (device nvmlDevice) GetMemClkMinMaxVfOffset() (int, int, Return) { + var minOffset, maxOffset int32 + ret := nvmlDeviceGetMemClkMinMaxVfOffset(device, &minOffset, &maxOffset) + return int(minOffset), int(maxOffset), ret +} + +// nvml.DeviceGetGpuMaxPcieLinkGeneration() +func (l *library) DeviceGetGpuMaxPcieLinkGeneration(device Device) (int, Return) { + return device.GetGpuMaxPcieLinkGeneration() +} + +func (device nvmlDevice) GetGpuMaxPcieLinkGeneration() (int, Return) { + var maxLinkGenDevice uint32 + ret := nvmlDeviceGetGpuMaxPcieLinkGeneration(device, &maxLinkGenDevice) + return int(maxLinkGenDevice), ret +} + +// nvml.DeviceGetFanControlPolicy_v2() +func (l *library) DeviceGetFanControlPolicy_v2(device Device, fan int) (FanControlPolicy, Return) { + return device.GetFanControlPolicy_v2(fan) +} + +func (device nvmlDevice) GetFanControlPolicy_v2(fan int) (FanControlPolicy, Return) { + var policy FanControlPolicy + ret := nvmlDeviceGetFanControlPolicy_v2(device, uint32(fan), &policy) + return policy, ret +} + +// nvml.DeviceSetFanControlPolicy() +func (l *library) DeviceSetFanControlPolicy(device Device, fan int, policy FanControlPolicy) Return { + return device.SetFanControlPolicy(fan, policy) +} + +func (device nvmlDevice) SetFanControlPolicy(fan int, policy FanControlPolicy) Return { + return nvmlDeviceSetFanControlPolicy(device, uint32(fan), policy) +} + +// nvml.DeviceClearFieldValues() +func (l *library) DeviceClearFieldValues(device Device, values []FieldValue) Return { + return device.ClearFieldValues(values) +} + +func (device nvmlDevice) ClearFieldValues(values []FieldValue) Return { + valuesCount := len(values) + return nvmlDeviceClearFieldValues(device, int32(valuesCount), &values[0]) +} + +// nvml.DeviceGetVgpuCapabilities() +func (l *library) DeviceGetVgpuCapabilities(device Device, capability DeviceVgpuCapability) (bool, Return) { + return device.GetVgpuCapabilities(capability) +} + +func (device nvmlDevice) GetVgpuCapabilities(capability DeviceVgpuCapability) (bool, Return) { + var capResult uint32 + ret := nvmlDeviceGetVgpuCapabilities(device, capability, &capResult) + return (capResult != 0), ret +} + +// nvml.DeviceGetVgpuSchedulerLog() +func (l *library) DeviceGetVgpuSchedulerLog(device Device) (VgpuSchedulerLog, Return) { + return device.GetVgpuSchedulerLog() +} + +func (device nvmlDevice) GetVgpuSchedulerLog() (VgpuSchedulerLog, Return) { + var pSchedulerLog VgpuSchedulerLog + ret := nvmlDeviceGetVgpuSchedulerLog(device, &pSchedulerLog) + return pSchedulerLog, ret +} + +// nvml.DeviceGetVgpuSchedulerState() +func (l *library) DeviceGetVgpuSchedulerState(device Device) (VgpuSchedulerGetState, Return) { + return device.GetVgpuSchedulerState() +} + +func (device nvmlDevice) GetVgpuSchedulerState() (VgpuSchedulerGetState, Return) { + var pSchedulerState VgpuSchedulerGetState + ret := nvmlDeviceGetVgpuSchedulerState(device, &pSchedulerState) + return pSchedulerState, ret +} + +// nvml.DeviceSetVgpuSchedulerState() +func (l *library) DeviceSetVgpuSchedulerState(device Device, pSchedulerState *VgpuSchedulerSetState) Return { + return device.SetVgpuSchedulerState(pSchedulerState) +} + +func (device nvmlDevice) SetVgpuSchedulerState(pSchedulerState *VgpuSchedulerSetState) Return { + return nvmlDeviceSetVgpuSchedulerState(device, pSchedulerState) +} + +// nvml.DeviceGetVgpuSchedulerCapabilities() +func (l *library) DeviceGetVgpuSchedulerCapabilities(device Device) (VgpuSchedulerCapabilities, Return) { + return device.GetVgpuSchedulerCapabilities() +} + +func (device nvmlDevice) GetVgpuSchedulerCapabilities() (VgpuSchedulerCapabilities, Return) { + var pCapabilities VgpuSchedulerCapabilities + ret := nvmlDeviceGetVgpuSchedulerCapabilities(device, &pCapabilities) + return pCapabilities, ret +} + +// nvml.GpuInstanceGetComputeInstancePossiblePlacements() +func (l *library) GpuInstanceGetComputeInstancePossiblePlacements(gpuInstance GpuInstance, info *ComputeInstanceProfileInfo) ([]ComputeInstancePlacement, Return) { + return gpuInstance.GetComputeInstancePossiblePlacements(info) +} + +func (gpuInstance nvmlGpuInstance) GetComputeInstancePossiblePlacements(info *ComputeInstanceProfileInfo) ([]ComputeInstancePlacement, Return) { + var count uint32 + ret := nvmlGpuInstanceGetComputeInstancePossiblePlacements(gpuInstance, info.Id, nil, &count) + if ret != SUCCESS { + return nil, ret + } + if count == 0 { + return []ComputeInstancePlacement{}, ret + } + placementArray := make([]ComputeInstancePlacement, count) + ret = nvmlGpuInstanceGetComputeInstancePossiblePlacements(gpuInstance, info.Id, &placementArray[0], &count) + return placementArray, ret +} + +// nvml.GpuInstanceCreateComputeInstanceWithPlacement() +func (l *library) GpuInstanceCreateComputeInstanceWithPlacement(gpuInstance GpuInstance, info *ComputeInstanceProfileInfo, placement *ComputeInstancePlacement) (ComputeInstance, Return) { + return gpuInstance.CreateComputeInstanceWithPlacement(info, placement) +} + +func (gpuInstance nvmlGpuInstance) CreateComputeInstanceWithPlacement(info *ComputeInstanceProfileInfo, placement *ComputeInstancePlacement) (ComputeInstance, Return) { + var computeInstance nvmlComputeInstance + ret := nvmlGpuInstanceCreateComputeInstanceWithPlacement(gpuInstance, info.Id, placement, &computeInstance) + return computeInstance, ret +} + +// nvml.DeviceGetGpuFabricInfo() +// +// Deprecated: Use DeviceGetGpuFabricInfoV instead +func (l *library) DeviceGetGpuFabricInfo(device Device) (GpuFabricInfo, Return) { + return device.GetGpuFabricInfo() +} + +// Deprecated: Use DeviceGetGpuFabricInfoV instead +func (device nvmlDevice) GetGpuFabricInfo() (GpuFabricInfo, Return) { + var gpuFabricInfo GpuFabricInfo + ret := nvmlDeviceGetGpuFabricInfo(device, &gpuFabricInfo) + return gpuFabricInfo, ret +} + +// nvml.DeviceSetNvLinkDeviceLowPowerThreshold() +func (l *library) DeviceSetNvLinkDeviceLowPowerThreshold(device Device, info *NvLinkPowerThres) Return { + return device.SetNvLinkDeviceLowPowerThreshold(info) +} + +func (device nvmlDevice) SetNvLinkDeviceLowPowerThreshold(info *NvLinkPowerThres) Return { + return nvmlDeviceSetNvLinkDeviceLowPowerThreshold(device, info) +} + +// nvml.DeviceGetModuleId() +func (l *library) DeviceGetModuleId(device Device) (int, Return) { + return device.GetModuleId() +} + +func (device nvmlDevice) GetModuleId() (int, Return) { + var moduleID uint32 + ret := nvmlDeviceGetModuleId(device, &moduleID) + return int(moduleID), ret +} + +// nvml.DeviceGetCurrentClocksEventReasons() +func (l *library) DeviceGetCurrentClocksEventReasons(device Device) (uint64, Return) { + return device.GetCurrentClocksEventReasons() +} + +func (device nvmlDevice) GetCurrentClocksEventReasons() (uint64, Return) { + var clocksEventReasons uint64 + ret := nvmlDeviceGetCurrentClocksEventReasons(device, &clocksEventReasons) + return clocksEventReasons, ret +} + +// nvml.DeviceGetSupportedClocksEventReasons() +func (l *library) DeviceGetSupportedClocksEventReasons(device Device) (uint64, Return) { + return device.GetSupportedClocksEventReasons() +} + +func (device nvmlDevice) GetSupportedClocksEventReasons() (uint64, Return) { + var supportedClocksEventReasons uint64 + ret := nvmlDeviceGetSupportedClocksEventReasons(device, &supportedClocksEventReasons) + return supportedClocksEventReasons, ret +} + +// nvml.DeviceGetJpgUtilization() +func (l *library) DeviceGetJpgUtilization(device Device) (uint32, uint32, Return) { + return device.GetJpgUtilization() +} + +func (device nvmlDevice) GetJpgUtilization() (uint32, uint32, Return) { + var utilization, samplingPeriodUs uint32 + ret := nvmlDeviceGetJpgUtilization(device, &utilization, &samplingPeriodUs) + return utilization, samplingPeriodUs, ret +} + +// nvml.DeviceGetOfaUtilization() +func (l *library) DeviceGetOfaUtilization(device Device) (uint32, uint32, Return) { + return device.GetOfaUtilization() +} + +func (device nvmlDevice) GetOfaUtilization() (uint32, uint32, Return) { + var utilization, samplingPeriodUs uint32 + ret := nvmlDeviceGetOfaUtilization(device, &utilization, &samplingPeriodUs) + return utilization, samplingPeriodUs, ret +} + +// nvml.DeviceGetRunningProcessDetailList() +func (l *library) DeviceGetRunningProcessDetailList(device Device) (ProcessDetailList, Return) { + return device.GetRunningProcessDetailList() +} + +func (device nvmlDevice) GetRunningProcessDetailList() (ProcessDetailList, Return) { + var plist ProcessDetailList + plist.Version = STRUCT_VERSION(plist, 1) + ret := nvmlDeviceGetRunningProcessDetailList(device, &plist) + return plist, ret +} + +// nvml.DeviceGetConfComputeMemSizeInfo() +func (l *library) DeviceGetConfComputeMemSizeInfo(device Device) (ConfComputeMemSizeInfo, Return) { + return device.GetConfComputeMemSizeInfo() +} + +func (device nvmlDevice) GetConfComputeMemSizeInfo() (ConfComputeMemSizeInfo, Return) { + var memInfo ConfComputeMemSizeInfo + ret := nvmlDeviceGetConfComputeMemSizeInfo(device, &memInfo) + return memInfo, ret +} + +// nvml.DeviceGetConfComputeProtectedMemoryUsage() +func (l *library) DeviceGetConfComputeProtectedMemoryUsage(device Device) (Memory, Return) { + return device.GetConfComputeProtectedMemoryUsage() +} + +func (device nvmlDevice) GetConfComputeProtectedMemoryUsage() (Memory, Return) { + var memory Memory + ret := nvmlDeviceGetConfComputeProtectedMemoryUsage(device, &memory) + return memory, ret +} + +// nvml.DeviceGetConfComputeGpuCertificate() +func (l *library) DeviceGetConfComputeGpuCertificate(device Device) (ConfComputeGpuCertificate, Return) { + return device.GetConfComputeGpuCertificate() +} + +func (device nvmlDevice) GetConfComputeGpuCertificate() (ConfComputeGpuCertificate, Return) { + var gpuCert ConfComputeGpuCertificate + ret := nvmlDeviceGetConfComputeGpuCertificate(device, &gpuCert) + return gpuCert, ret +} + +// nvml.DeviceGetConfComputeGpuAttestationReport() +func (l *library) DeviceGetConfComputeGpuAttestationReport(device Device, gpuAtstReport *ConfComputeGpuAttestationReport) Return { + return device.GetConfComputeGpuAttestationReport(gpuAtstReport) +} + +func (device nvmlDevice) GetConfComputeGpuAttestationReport(gpuAtstReport *ConfComputeGpuAttestationReport) Return { + return nvmlDeviceGetConfComputeGpuAttestationReport(device, gpuAtstReport) +} + +// nvml.DeviceSetConfComputeUnprotectedMemSize() +func (l *library) DeviceSetConfComputeUnprotectedMemSize(device Device, sizeKiB uint64) Return { + return device.SetConfComputeUnprotectedMemSize(sizeKiB) +} + +func (device nvmlDevice) SetConfComputeUnprotectedMemSize(sizeKiB uint64) Return { + return nvmlDeviceSetConfComputeUnprotectedMemSize(device, sizeKiB) +} + +// nvml.DeviceGetC2cModeInfoV() +type C2cModeInfoHandler struct { + device nvmlDevice +} + +func (handler C2cModeInfoHandler) V1() (C2cModeInfo_v1, Return) { + var c2cModeInfo C2cModeInfo_v1 + ret := nvmlDeviceGetC2cModeInfoV(handler.device, &c2cModeInfo) + return c2cModeInfo, ret +} + +func (l *library) DeviceGetC2cModeInfoV(device Device) C2cModeInfoHandler { + return device.GetC2cModeInfoV() +} + +func (device nvmlDevice) GetC2cModeInfoV() C2cModeInfoHandler { + return C2cModeInfoHandler{device} +} + +// nvml.DeviceGetLastBBXFlushTime() +func (l *library) DeviceGetLastBBXFlushTime(device Device) (uint64, uint, Return) { + return device.GetLastBBXFlushTime() +} + +func (device nvmlDevice) GetLastBBXFlushTime() (uint64, uint, Return) { + var timestamp uint64 + var durationUs uint + ret := nvmlDeviceGetLastBBXFlushTime(device, ×tamp, &durationUs) + return timestamp, durationUs, ret +} + +// nvml.DeviceGetNumaNodeId() +func (l *library) DeviceGetNumaNodeId(device Device) (int, Return) { + return device.GetNumaNodeId() +} + +func (device nvmlDevice) GetNumaNodeId() (int, Return) { + var node uint32 + ret := nvmlDeviceGetNumaNodeId(device, &node) + return int(node), ret +} + +func (l *library) DeviceGetAddressingMode(device Device) (DeviceAddressingMode, Return) { + return device.GetAddressingMode() +} + +func (device nvmlDevice) GetAddressingMode() (DeviceAddressingMode, Return) { + var deviceAddressingMode DeviceAddressingMode + deviceAddressingMode.Version = STRUCT_VERSION(deviceAddressingMode, 1) + ret := nvmlDeviceGetAddressingMode(device, &deviceAddressingMode) + return deviceAddressingMode, ret +} + +func (l *library) DeviceGetRepairStatus(device Device) (RepairStatus, Return) { + return device.GetRepairStatus() +} + +func (device nvmlDevice) GetRepairStatus() (RepairStatus, Return) { + var repairStatus RepairStatus + repairStatus.Version = STRUCT_VERSION(repairStatus, 1) + ret := nvmlDeviceGetRepairStatus(device, &repairStatus) + return repairStatus, ret +} + +// nvml.DeviceGetPciInfoExt() +func (l *library) DeviceGetPciInfoExt(device Device) (PciInfoExt, Return) { + return device.GetPciInfoExt() +} + +func (device nvmlDevice) GetPciInfoExt() (PciInfoExt, Return) { + var pciInfo PciInfoExt + pciInfo.Version = STRUCT_VERSION(pciInfo, 1) + ret := nvmlDeviceGetPciInfoExt(device, &pciInfo) + return pciInfo, ret +} + +// nvml.DeviceGetGpuFabricInfoV() +type GpuFabricInfoHandler struct { + device nvmlDevice +} + +func (handler GpuFabricInfoHandler) V1() (GpuFabricInfo, Return) { + return handler.device.GetGpuFabricInfo() +} + +func (handler GpuFabricInfoHandler) V2() (GpuFabricInfo_v2, Return) { + var info GpuFabricInfo_v2 + info.Version = STRUCT_VERSION(info, 2) + ret := nvmlDeviceGetGpuFabricInfoV(handler.device, (*GpuFabricInfoV)(unsafe.Pointer(&info))) + return info, ret +} + +func (handler GpuFabricInfoHandler) V3() (GpuFabricInfo_v3, Return) { + var info GpuFabricInfo_v3 + info.Version = STRUCT_VERSION(info, 3) + ret := nvmlDeviceGetGpuFabricInfoV(handler.device, (*GpuFabricInfoV)(unsafe.Pointer(&info))) + return info, ret +} + +func (l *library) DeviceGetGpuFabricInfoV(device Device) GpuFabricInfoHandler { + return device.GetGpuFabricInfoV() +} + +func (device nvmlDevice) GetGpuFabricInfoV() GpuFabricInfoHandler { + return GpuFabricInfoHandler{device} +} + +// nvml.DeviceGetProcessesUtilizationInfo() +func (l *library) DeviceGetProcessesUtilizationInfo(device Device) (ProcessesUtilizationInfo, Return) { + return device.GetProcessesUtilizationInfo() +} + +func (device nvmlDevice) GetProcessesUtilizationInfo() (ProcessesUtilizationInfo, Return) { + var processesUtilInfo ProcessesUtilizationInfo + ret := nvmlDeviceGetProcessesUtilizationInfo(device, &processesUtilInfo) + return processesUtilInfo, ret +} + +// nvml.DeviceGetVgpuHeterogeneousMode() +func (l *library) DeviceGetVgpuHeterogeneousMode(device Device) (VgpuHeterogeneousMode, Return) { + return device.GetVgpuHeterogeneousMode() +} + +func (device nvmlDevice) GetVgpuHeterogeneousMode() (VgpuHeterogeneousMode, Return) { + var heterogeneousMode VgpuHeterogeneousMode + heterogeneousMode.Version = STRUCT_VERSION(heterogeneousMode, 1) + ret := nvmlDeviceGetVgpuHeterogeneousMode(device, &heterogeneousMode) + return heterogeneousMode, ret +} + +// nvml.DeviceSetVgpuHeterogeneousMode() +func (l *library) DeviceSetVgpuHeterogeneousMode(device Device, heterogeneousMode VgpuHeterogeneousMode) Return { + return device.SetVgpuHeterogeneousMode(heterogeneousMode) +} + +func (device nvmlDevice) SetVgpuHeterogeneousMode(heterogeneousMode VgpuHeterogeneousMode) Return { + ret := nvmlDeviceSetVgpuHeterogeneousMode(device, &heterogeneousMode) + return ret +} + +// nvml.DeviceGetVgpuTypeSupportedPlacements() +func (l *library) DeviceGetVgpuTypeSupportedPlacements(device Device, vgpuTypeId VgpuTypeId) (VgpuPlacementList, Return) { + return device.GetVgpuTypeSupportedPlacements(vgpuTypeId) +} + +func (device nvmlDevice) GetVgpuTypeSupportedPlacements(vgpuTypeId VgpuTypeId) (VgpuPlacementList, Return) { + return vgpuTypeId.GetSupportedPlacements(device) +} + +func (vgpuTypeId nvmlVgpuTypeId) GetSupportedPlacements(device Device) (VgpuPlacementList, Return) { + var placementList VgpuPlacementList + placementList.Version = STRUCT_VERSION(placementList, 1) + ret := nvmlDeviceGetVgpuTypeSupportedPlacements(nvmlDeviceHandle(device), vgpuTypeId, &placementList) + return placementList, ret +} + +// nvml.DeviceGetVgpuTypeCreatablePlacements() +func (l *library) DeviceGetVgpuTypeCreatablePlacements(device Device, vgpuTypeId VgpuTypeId) (VgpuPlacementList, Return) { + return device.GetVgpuTypeCreatablePlacements(vgpuTypeId) +} + +func (device nvmlDevice) GetVgpuTypeCreatablePlacements(vgpuTypeId VgpuTypeId) (VgpuPlacementList, Return) { + return vgpuTypeId.GetCreatablePlacements(device) +} + +func (vgpuTypeId nvmlVgpuTypeId) GetCreatablePlacements(device Device) (VgpuPlacementList, Return) { + var placementList VgpuPlacementList + placementList.Version = STRUCT_VERSION(placementList, 1) + ret := nvmlDeviceGetVgpuTypeCreatablePlacements(nvmlDeviceHandle(device), vgpuTypeId, &placementList) + return placementList, ret +} + +// nvml.DeviceSetVgpuCapabilities() +func (l *library) DeviceSetVgpuCapabilities(device Device, capability DeviceVgpuCapability, state EnableState) Return { + return device.SetVgpuCapabilities(capability, state) +} + +func (device nvmlDevice) SetVgpuCapabilities(capability DeviceVgpuCapability, state EnableState) Return { + ret := nvmlDeviceSetVgpuCapabilities(device, capability, state) + return ret +} + +// nvml.DeviceGetVgpuInstancesUtilizationInfo() +func (l *library) DeviceGetVgpuInstancesUtilizationInfo(device Device) (VgpuInstancesUtilizationInfo, Return) { + return device.GetVgpuInstancesUtilizationInfo() +} + +func (device nvmlDevice) GetVgpuInstancesUtilizationInfo() (VgpuInstancesUtilizationInfo, Return) { + var vgpuUtilInfo VgpuInstancesUtilizationInfo + ret := nvmlDeviceGetVgpuInstancesUtilizationInfo(device, &vgpuUtilInfo) + return vgpuUtilInfo, ret +} + +// nvml.DeviceGetVgpuProcessesUtilizationInfo() +func (l *library) DeviceGetVgpuProcessesUtilizationInfo(device Device) (VgpuProcessesUtilizationInfo, Return) { + return device.GetVgpuProcessesUtilizationInfo() +} + +func (device nvmlDevice) GetVgpuProcessesUtilizationInfo() (VgpuProcessesUtilizationInfo, Return) { + var vgpuProcUtilInfo VgpuProcessesUtilizationInfo + vgpuProcUtilInfo.Version = STRUCT_VERSION(vgpuProcUtilInfo, 1) + ret := nvmlDeviceGetVgpuProcessesUtilizationInfo(device, &vgpuProcUtilInfo) + return vgpuProcUtilInfo, ret +} + +// nvml.DeviceGetSramEccErrorStatus() +func (l *library) DeviceGetSramEccErrorStatus(device Device) (EccSramErrorStatus, Return) { + return device.GetSramEccErrorStatus() +} + +func (device nvmlDevice) GetSramEccErrorStatus() (EccSramErrorStatus, Return) { + var status EccSramErrorStatus + status.Version = STRUCT_VERSION(status, 1) + ret := nvmlDeviceGetSramEccErrorStatus(device, &status) + return status, ret +} + +// nvml.DeviceGetClockOffsets() +func (l *library) DeviceGetClockOffsets(device Device) (ClockOffset, Return) { + return device.GetClockOffsets() +} + +func (device nvmlDevice) GetClockOffsets() (ClockOffset, Return) { + var info ClockOffset + info.Version = STRUCT_VERSION(info, 1) + ret := nvmlDeviceGetClockOffsets(device, &info) + return info, ret +} + +// nvml.DeviceSetClockOffsets() +func (l *library) DeviceSetClockOffsets(device Device, info ClockOffset) Return { + return device.SetClockOffsets(info) +} + +func (device nvmlDevice) SetClockOffsets(info ClockOffset) Return { + return nvmlDeviceSetClockOffsets(device, &info) +} + +// nvml.DeviceGetDriverModel_v2() +func (l *library) DeviceGetDriverModel_v2(device Device) (DriverModel, DriverModel, Return) { + return device.GetDriverModel_v2() +} + +func (device nvmlDevice) GetDriverModel_v2() (DriverModel, DriverModel, Return) { + var current, pending DriverModel + ret := nvmlDeviceGetDriverModel_v2(device, ¤t, &pending) + return current, pending, ret +} + +// nvml.DeviceGetCapabilities() +func (l *library) DeviceGetCapabilities(device Device) (DeviceCapabilities, Return) { + return device.GetCapabilities() +} + +func (device nvmlDevice) GetCapabilities() (DeviceCapabilities, Return) { + var caps DeviceCapabilities + caps.Version = STRUCT_VERSION(caps, 1) + ret := nvmlDeviceGetCapabilities(device, &caps) + return caps, ret +} + +// nvml.DeviceGetFanSpeedRPM() +func (l *library) DeviceGetFanSpeedRPM(device Device) (FanSpeedInfo, Return) { + return device.GetFanSpeedRPM() +} + +func (device nvmlDevice) GetFanSpeedRPM() (FanSpeedInfo, Return) { + var fanSpeed FanSpeedInfo + fanSpeed.Version = STRUCT_VERSION(fanSpeed, 1) + ret := nvmlDeviceGetFanSpeedRPM(device, &fanSpeed) + return fanSpeed, ret +} + +// nvml.DeviceGetCoolerInfo() +func (l *library) DeviceGetCoolerInfo(device Device) (CoolerInfo, Return) { + return device.GetCoolerInfo() +} + +func (device nvmlDevice) GetCoolerInfo() (CoolerInfo, Return) { + var coolerInfo CoolerInfo + coolerInfo.Version = STRUCT_VERSION(coolerInfo, 1) + ret := nvmlDeviceGetCoolerInfo(device, &coolerInfo) + return coolerInfo, ret +} + +// nvml.DeviceGetTemperatureV() +type TemperatureHandler struct { + device nvmlDevice +} + +func (handler TemperatureHandler) V1() (Temperature, Return) { + var temperature Temperature + temperature.Version = STRUCT_VERSION(temperature, 1) + ret := nvmlDeviceGetTemperatureV(handler.device, &temperature) + return temperature, ret +} + +func (l *library) DeviceGetTemperatureV(device Device) TemperatureHandler { + return device.GetTemperatureV() +} + +func (device nvmlDevice) GetTemperatureV() TemperatureHandler { + return TemperatureHandler{device} +} + +// nvml.DeviceGetMarginTemperature() +func (l *library) DeviceGetMarginTemperature(device Device) (MarginTemperature, Return) { + return device.GetMarginTemperature() +} + +func (device nvmlDevice) GetMarginTemperature() (MarginTemperature, Return) { + var marginTemp MarginTemperature + marginTemp.Version = STRUCT_VERSION(marginTemp, 1) + ret := nvmlDeviceGetMarginTemperature(device, &marginTemp) + return marginTemp, ret +} + +// nvml.DeviceGetPerformanceModes() +func (l *library) DeviceGetPerformanceModes(device Device) (DevicePerfModes, Return) { + return device.GetPerformanceModes() +} + +func (device nvmlDevice) GetPerformanceModes() (DevicePerfModes, Return) { + var perfModes DevicePerfModes + perfModes.Version = STRUCT_VERSION(perfModes, 1) + ret := nvmlDeviceGetPerformanceModes(device, &perfModes) + return perfModes, ret +} + +// nvml.DeviceGetCurrentClockFreqs() +func (l *library) DeviceGetCurrentClockFreqs(device Device) (DeviceCurrentClockFreqs, Return) { + return device.GetCurrentClockFreqs() +} + +func (device nvmlDevice) GetCurrentClockFreqs() (DeviceCurrentClockFreqs, Return) { + var currentClockFreqs DeviceCurrentClockFreqs + currentClockFreqs.Version = STRUCT_VERSION(currentClockFreqs, 1) + ret := nvmlDeviceGetCurrentClockFreqs(device, ¤tClockFreqs) + return currentClockFreqs, ret +} + +// nvml.DeviceGetDramEncryptionMode() +func (l *library) DeviceGetDramEncryptionMode(device Device) (DramEncryptionInfo, DramEncryptionInfo, Return) { + return device.GetDramEncryptionMode() +} + +func (device nvmlDevice) GetDramEncryptionMode() (DramEncryptionInfo, DramEncryptionInfo, Return) { + var current, pending DramEncryptionInfo + current.Version = STRUCT_VERSION(current, 1) + pending.Version = STRUCT_VERSION(pending, 1) + ret := nvmlDeviceGetDramEncryptionMode(device, ¤t, &pending) + return current, pending, ret +} + +// nvml.DeviceSetDramEncryptionMode() +func (l *library) DeviceSetDramEncryptionMode(device Device, dramEncryption *DramEncryptionInfo) Return { + return device.SetDramEncryptionMode(dramEncryption) +} + +func (device nvmlDevice) SetDramEncryptionMode(dramEncryption *DramEncryptionInfo) Return { + return nvmlDeviceSetDramEncryptionMode(device, dramEncryption) +} + +// nvml.DeviceGetPlatformInfo() +func (l *library) DeviceGetPlatformInfo(device Device) (PlatformInfo, Return) { + return device.GetPlatformInfo() +} + +func (device nvmlDevice) GetPlatformInfo() (PlatformInfo, Return) { + var platformInfo PlatformInfo + platformInfo.Version = STRUCT_VERSION(platformInfo, 1) + ret := nvmlDeviceGetPlatformInfo(device, &platformInfo) + return platformInfo, ret +} + +// nvml.DeviceGetNvlinkSupportedBwModes() +func (l *library) DeviceGetNvlinkSupportedBwModes(device Device) (NvlinkSupportedBwModes, Return) { + return device.GetNvlinkSupportedBwModes() +} + +func (device nvmlDevice) GetNvlinkSupportedBwModes() (NvlinkSupportedBwModes, Return) { + var supportedBwMode NvlinkSupportedBwModes + supportedBwMode.Version = STRUCT_VERSION(supportedBwMode, 1) + ret := nvmlDeviceGetNvlinkSupportedBwModes(device, &supportedBwMode) + return supportedBwMode, ret +} + +// nvml.DeviceGetNvlinkBwMode() +func (l *library) DeviceGetNvlinkBwMode(device Device) (NvlinkGetBwMode, Return) { + return device.GetNvlinkBwMode() +} + +func (device nvmlDevice) GetNvlinkBwMode() (NvlinkGetBwMode, Return) { + var getBwMode NvlinkGetBwMode + getBwMode.Version = STRUCT_VERSION(getBwMode, 1) + ret := nvmlDeviceGetNvlinkBwMode(device, &getBwMode) + return getBwMode, ret +} + +// nvml.DeviceSetNvlinkBwMode() +func (l *library) DeviceSetNvlinkBwMode(device Device, setBwMode *NvlinkSetBwMode) Return { + return device.SetNvlinkBwMode(setBwMode) +} + +func (device nvmlDevice) SetNvlinkBwMode(setBwMode *NvlinkSetBwMode) Return { + return nvmlDeviceSetNvlinkBwMode(device, setBwMode) +} + +func (l *library) DeviceGetNvLinkInfo(device Device) NvLinkInfoHandler { + return device.GetNvLinkInfo() +} + +func (device nvmlDevice) GetNvLinkInfo() NvLinkInfoHandler { + return NvLinkInfoHandler{device} +} + +type NvLinkInfoHandler struct { + device nvmlDevice +} + +func (handler NvLinkInfoHandler) V1() (NvLinkInfo_v1, Return) { + var info NvLinkInfo_v1 + info.Version = STRUCT_VERSION(info, 1) + ret := nvmlDeviceGetNvLinkInfo(handler.device, (*NvLinkInfo)(unsafe.Pointer(&info))) + + return info, ret +} + +func (handler NvLinkInfoHandler) V2() (NvLinkInfo_v2, Return) { + var info NvLinkInfo_v2 + info.Version = STRUCT_VERSION(info, 2) + ret := nvmlDeviceGetNvLinkInfo(handler.device, (*NvLinkInfo)(unsafe.Pointer(&info))) + + return info, ret +} + +// nvml.DeviceWorkloadPowerProfileGetProfilesInfo() +func (l *library) DeviceWorkloadPowerProfileGetProfilesInfo(device Device) (WorkloadPowerProfileProfilesInfo, Return) { + return device.WorkloadPowerProfileGetProfilesInfo() +} + +func (device nvmlDevice) WorkloadPowerProfileGetProfilesInfo() (WorkloadPowerProfileProfilesInfo, Return) { + var profilesInfo WorkloadPowerProfileProfilesInfo + profilesInfo.Version = STRUCT_VERSION(profilesInfo, 1) + ret := nvmlDeviceWorkloadPowerProfileGetProfilesInfo(device, &profilesInfo) + return profilesInfo, ret +} + +// nvml.DeviceWorkloadPowerProfileGetCurrentProfiles() +func (l *library) DeviceWorkloadPowerProfileGetCurrentProfiles(device Device) (WorkloadPowerProfileCurrentProfiles, Return) { + return device.WorkloadPowerProfileGetCurrentProfiles() +} + +func (device nvmlDevice) WorkloadPowerProfileGetCurrentProfiles() (WorkloadPowerProfileCurrentProfiles, Return) { + var currentProfiles WorkloadPowerProfileCurrentProfiles + currentProfiles.Version = STRUCT_VERSION(currentProfiles, 1) + ret := nvmlDeviceWorkloadPowerProfileGetCurrentProfiles(device, ¤tProfiles) + return currentProfiles, ret +} + +// nvml.DeviceWorkloadPowerProfileSetRequestedProfiles() +func (l *library) DeviceWorkloadPowerProfileSetRequestedProfiles(device Device, requestedProfiles *WorkloadPowerProfileRequestedProfiles) Return { + return device.WorkloadPowerProfileSetRequestedProfiles(requestedProfiles) +} + +func (device nvmlDevice) WorkloadPowerProfileSetRequestedProfiles(requestedProfiles *WorkloadPowerProfileRequestedProfiles) Return { + return nvmlDeviceWorkloadPowerProfileSetRequestedProfiles(device, requestedProfiles) +} + +// nvml.DeviceWorkloadPowerProfileClearRequestedProfiles() +func (l *library) DeviceWorkloadPowerProfileClearRequestedProfiles(device Device, requestedProfiles *WorkloadPowerProfileRequestedProfiles) Return { + return device.WorkloadPowerProfileClearRequestedProfiles(requestedProfiles) +} + +func (device nvmlDevice) WorkloadPowerProfileClearRequestedProfiles(requestedProfiles *WorkloadPowerProfileRequestedProfiles) Return { + return nvmlDeviceWorkloadPowerProfileClearRequestedProfiles(device, requestedProfiles) +} + +// nvml.DevicePowerSmoothingActivatePresetProfile() +func (l *library) DevicePowerSmoothingActivatePresetProfile(device Device, profile *PowerSmoothingProfile) Return { + return device.PowerSmoothingActivatePresetProfile(profile) +} + +func (device nvmlDevice) PowerSmoothingActivatePresetProfile(profile *PowerSmoothingProfile) Return { + return nvmlDevicePowerSmoothingActivatePresetProfile(device, profile) +} + +// nvml.DevicePowerSmoothingUpdatePresetProfileParam() +func (l *library) DevicePowerSmoothingUpdatePresetProfileParam(device Device, profile *PowerSmoothingProfile) Return { + return device.PowerSmoothingUpdatePresetProfileParam(profile) +} + +func (device nvmlDevice) PowerSmoothingUpdatePresetProfileParam(profile *PowerSmoothingProfile) Return { + return nvmlDevicePowerSmoothingUpdatePresetProfileParam(device, profile) +} + +// nvml.DevicePowerSmoothingSetState() +func (l *library) DevicePowerSmoothingSetState(device Device, state *PowerSmoothingState) Return { + return device.PowerSmoothingSetState(state) +} + +func (device nvmlDevice) PowerSmoothingSetState(state *PowerSmoothingState) Return { + return nvmlDevicePowerSmoothingSetState(device, state) +} + +func (l *library) DeviceGetSramUniqueUncorrectedEccErrorCounts(device Device, errorCounts *EccSramUniqueUncorrectedErrorCounts) Return { + return device.GetSramUniqueUncorrectedEccErrorCounts(errorCounts) +} + +func (device nvmlDevice) GetSramUniqueUncorrectedEccErrorCounts(errorCounts *EccSramUniqueUncorrectedErrorCounts) Return { + return nvmlDeviceGetSramUniqueUncorrectedEccErrorCounts(device, errorCounts) +} + +// nvml.GpuInstanceGetCreatableVgpus() +func (l *library) GpuInstanceGetCreatableVgpus(gpuInstance GpuInstance) (VgpuTypeIdInfo, Return) { + return gpuInstance.GetCreatableVgpus() +} + +func (gpuInstance nvmlGpuInstance) GetCreatableVgpus() (VgpuTypeIdInfo, Return) { + var vgpuTypeIdInfo VgpuTypeIdInfo + vgpuTypeIdInfo.Version = STRUCT_VERSION(vgpuTypeIdInfo, 1) + ret := nvmlGpuInstanceGetCreatableVgpus(gpuInstance, &vgpuTypeIdInfo) + return vgpuTypeIdInfo, ret +} + +// nvml.GpuInstanceGetActiveVgpus() +func (l *library) GpuInstanceGetActiveVgpus(gpuInstance GpuInstance) (ActiveVgpuInstanceInfo, Return) { + return gpuInstance.GetActiveVgpus() +} + +func (gpuInstance nvmlGpuInstance) GetActiveVgpus() (ActiveVgpuInstanceInfo, Return) { + var activeVgpuInstanceInfo ActiveVgpuInstanceInfo + activeVgpuInstanceInfo.Version = STRUCT_VERSION(activeVgpuInstanceInfo, 1) + ret := nvmlGpuInstanceGetActiveVgpus(gpuInstance, &activeVgpuInstanceInfo) + return activeVgpuInstanceInfo, ret +} + +// nvml.GpuInstanceSetVgpuSchedulerState() +func (l *library) GpuInstanceSetVgpuSchedulerState(gpuInstance GpuInstance, scheduler *VgpuSchedulerState) Return { + return gpuInstance.SetVgpuSchedulerState(scheduler) +} + +func (gpuInstance nvmlGpuInstance) SetVgpuSchedulerState(scheduler *VgpuSchedulerState) Return { + return nvmlGpuInstanceSetVgpuSchedulerState(gpuInstance, scheduler) +} + +// nvml.GpuInstanceGetVgpuSchedulerState() +func (l *library) GpuInstanceGetVgpuSchedulerState(gpuInstance GpuInstance) (VgpuSchedulerStateInfo, Return) { + return gpuInstance.GetVgpuSchedulerState() +} + +func (gpuInstance nvmlGpuInstance) GetVgpuSchedulerState() (VgpuSchedulerStateInfo, Return) { + var schedulerStateInfo VgpuSchedulerStateInfo + schedulerStateInfo.Version = STRUCT_VERSION(schedulerStateInfo, 1) + ret := nvmlGpuInstanceGetVgpuSchedulerState(gpuInstance, &schedulerStateInfo) + return schedulerStateInfo, ret +} + +// nvml.GpuInstanceGetVgpuSchedulerLog() +func (l *library) GpuInstanceGetVgpuSchedulerLog(gpuInstance GpuInstance) (VgpuSchedulerLogInfo, Return) { + return gpuInstance.GetVgpuSchedulerLog() +} + +func (gpuInstance nvmlGpuInstance) GetVgpuSchedulerLog() (VgpuSchedulerLogInfo, Return) { + var schedulerLogInfo VgpuSchedulerLogInfo + schedulerLogInfo.Version = STRUCT_VERSION(schedulerLogInfo, 1) + ret := nvmlGpuInstanceGetVgpuSchedulerLog(gpuInstance, &schedulerLogInfo) + return schedulerLogInfo, ret +} + +// nvml.GpuInstanceGetVgpuTypeCreatablePlacements() +func (l *library) GpuInstanceGetVgpuTypeCreatablePlacements(gpuInstance GpuInstance) (VgpuCreatablePlacementInfo, Return) { + return gpuInstance.GetVgpuTypeCreatablePlacements() +} + +func (gpuInstance nvmlGpuInstance) GetVgpuTypeCreatablePlacements() (VgpuCreatablePlacementInfo, Return) { + var creatablePlacementInfo VgpuCreatablePlacementInfo + creatablePlacementInfo.Version = STRUCT_VERSION(creatablePlacementInfo, 1) + ret := nvmlGpuInstanceGetVgpuTypeCreatablePlacements(gpuInstance, &creatablePlacementInfo) + return creatablePlacementInfo, ret +} + +// nvml.GpuInstanceGetVgpuHeterogeneousMode() +func (l *library) GpuInstanceGetVgpuHeterogeneousMode(gpuInstance GpuInstance) (VgpuHeterogeneousMode, Return) { + return gpuInstance.GetVgpuHeterogeneousMode() +} + +func (gpuInstance nvmlGpuInstance) GetVgpuHeterogeneousMode() (VgpuHeterogeneousMode, Return) { + var heterogeneousMode VgpuHeterogeneousMode + heterogeneousMode.Version = STRUCT_VERSION(heterogeneousMode, 1) + ret := nvmlGpuInstanceGetVgpuHeterogeneousMode(gpuInstance, &heterogeneousMode) + return heterogeneousMode, ret +} + +// nvml.GpuInstanceSetVgpuHeterogeneousMode() +func (l *library) GpuInstanceSetVgpuHeterogeneousMode(gpuInstance GpuInstance, heterogeneousMode *VgpuHeterogeneousMode) Return { + return gpuInstance.SetVgpuHeterogeneousMode(heterogeneousMode) +} + +func (gpuInstance nvmlGpuInstance) SetVgpuHeterogeneousMode(heterogeneousMode *VgpuHeterogeneousMode) Return { + return nvmlGpuInstanceSetVgpuHeterogeneousMode(gpuInstance, heterogeneousMode) +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/doc.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/doc.go new file mode 100644 index 00000000..c2ce2e37 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/doc.go @@ -0,0 +1,21 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +// WARNING: THIS FILE WAS AUTOMATICALLY GENERATED. +// Code generated by https://git.io/c-for-go. DO NOT EDIT. + +/* +Package NVML bindings +*/ +package nvml diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/dynamicLibrary_mock.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/dynamicLibrary_mock.go new file mode 100644 index 00000000..b785431c --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/dynamicLibrary_mock.go @@ -0,0 +1,157 @@ +// Code generated by moq; DO NOT EDIT. +// github.com/matryer/moq + +package nvml + +import ( + "sync" +) + +// Ensure, that dynamicLibraryMock does implement dynamicLibrary. +// If this is not the case, regenerate this file with moq. +var _ dynamicLibrary = &dynamicLibraryMock{} + +// dynamicLibraryMock is a mock implementation of dynamicLibrary. +// +// func TestSomethingThatUsesdynamicLibrary(t *testing.T) { +// +// // make and configure a mocked dynamicLibrary +// mockeddynamicLibrary := &dynamicLibraryMock{ +// CloseFunc: func() error { +// panic("mock out the Close method") +// }, +// LookupFunc: func(s string) error { +// panic("mock out the Lookup method") +// }, +// OpenFunc: func() error { +// panic("mock out the Open method") +// }, +// } +// +// // use mockeddynamicLibrary in code that requires dynamicLibrary +// // and then make assertions. +// +// } +type dynamicLibraryMock struct { + // CloseFunc mocks the Close method. + CloseFunc func() error + + // LookupFunc mocks the Lookup method. + LookupFunc func(s string) error + + // OpenFunc mocks the Open method. + OpenFunc func() error + + // calls tracks calls to the methods. + calls struct { + // Close holds details about calls to the Close method. + Close []struct { + } + // Lookup holds details about calls to the Lookup method. + Lookup []struct { + // S is the s argument value. + S string + } + // Open holds details about calls to the Open method. + Open []struct { + } + } + lockClose sync.RWMutex + lockLookup sync.RWMutex + lockOpen sync.RWMutex +} + +// Close calls CloseFunc. +func (mock *dynamicLibraryMock) Close() error { + callInfo := struct { + }{} + mock.lockClose.Lock() + mock.calls.Close = append(mock.calls.Close, callInfo) + mock.lockClose.Unlock() + if mock.CloseFunc == nil { + var ( + errOut error + ) + return errOut + } + return mock.CloseFunc() +} + +// CloseCalls gets all the calls that were made to Close. +// Check the length with: +// +// len(mockeddynamicLibrary.CloseCalls()) +func (mock *dynamicLibraryMock) CloseCalls() []struct { +} { + var calls []struct { + } + mock.lockClose.RLock() + calls = mock.calls.Close + mock.lockClose.RUnlock() + return calls +} + +// Lookup calls LookupFunc. +func (mock *dynamicLibraryMock) Lookup(s string) error { + callInfo := struct { + S string + }{ + S: s, + } + mock.lockLookup.Lock() + mock.calls.Lookup = append(mock.calls.Lookup, callInfo) + mock.lockLookup.Unlock() + if mock.LookupFunc == nil { + var ( + errOut error + ) + return errOut + } + return mock.LookupFunc(s) +} + +// LookupCalls gets all the calls that were made to Lookup. +// Check the length with: +// +// len(mockeddynamicLibrary.LookupCalls()) +func (mock *dynamicLibraryMock) LookupCalls() []struct { + S string +} { + var calls []struct { + S string + } + mock.lockLookup.RLock() + calls = mock.calls.Lookup + mock.lockLookup.RUnlock() + return calls +} + +// Open calls OpenFunc. +func (mock *dynamicLibraryMock) Open() error { + callInfo := struct { + }{} + mock.lockOpen.Lock() + mock.calls.Open = append(mock.calls.Open, callInfo) + mock.lockOpen.Unlock() + if mock.OpenFunc == nil { + var ( + errOut error + ) + return errOut + } + return mock.OpenFunc() +} + +// OpenCalls gets all the calls that were made to Open. +// Check the length with: +// +// len(mockeddynamicLibrary.OpenCalls()) +func (mock *dynamicLibraryMock) OpenCalls() []struct { +} { + var calls []struct { + } + mock.lockOpen.RLock() + calls = mock.calls.Open + mock.lockOpen.RUnlock() + return calls +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/event_set.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/event_set.go new file mode 100644 index 00000000..b772d57f --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/event_set.go @@ -0,0 +1,82 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +// EventData includes an interface type for Device instead of nvmlDevice +type EventData struct { + Device Device + EventType uint64 + EventData uint64 + GpuInstanceId uint32 + ComputeInstanceId uint32 +} + +func (e nvmlEventData) convert() EventData { + out := EventData{ + Device: e.Device, + EventType: e.EventType, + EventData: e.EventData, + GpuInstanceId: e.GpuInstanceId, + ComputeInstanceId: e.ComputeInstanceId, + } + return out +} + +// nvml.EventSetCreate() +func (l *library) EventSetCreate() (EventSet, Return) { + var Set nvmlEventSet + ret := nvmlEventSetCreate(&Set) + return Set, ret +} + +// nvml.EventSetWait() +func (l *library) EventSetWait(set EventSet, timeoutms uint32) (EventData, Return) { + return set.Wait(timeoutms) +} + +func (set nvmlEventSet) Wait(timeoutms uint32) (EventData, Return) { + var data nvmlEventData + ret := nvmlEventSetWait(set, &data, timeoutms) + return data.convert(), ret +} + +// nvml.EventSetFree() +func (l *library) EventSetFree(set EventSet) Return { + return set.Free() +} + +func (set nvmlEventSet) Free() Return { + return nvmlEventSetFree(set) +} + +// nvml.SystemEventSetCreate() +func (l *library) SystemEventSetCreate(request *SystemEventSetCreateRequest) Return { + return nvmlSystemEventSetCreate(request) +} + +// nvml.SystemEventSetFree() +func (l *library) SystemEventSetFree(request *SystemEventSetFreeRequest) Return { + return nvmlSystemEventSetFree(request) +} + +// nvml.SystemRegisterEvents() +func (l *library) SystemRegisterEvents(request *SystemRegisterEventRequest) Return { + return nvmlSystemRegisterEvents(request) +} + +// nvml.SystemEventSetWait() +func (l *library) SystemEventSetWait(request *SystemEventSetWaitRequest) Return { + return nvmlSystemEventSetWait(request) +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/gpm.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/gpm.go new file mode 100644 index 00000000..563bc593 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/gpm.go @@ -0,0 +1,170 @@ +// Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +// GpmMetricsGetType includes interface types for GpmSample instead of nvmlGpmSample +type GpmMetricsGetType struct { + Version uint32 + NumMetrics uint32 + Sample1 GpmSample + Sample2 GpmSample + Metrics [210]GpmMetric +} + +func (g *GpmMetricsGetType) convert() *nvmlGpmMetricsGetType { + out := &nvmlGpmMetricsGetType{ + Version: g.Version, + NumMetrics: g.NumMetrics, + Sample1: g.Sample1.(nvmlGpmSample), + Sample2: g.Sample2.(nvmlGpmSample), + } + copy(out.Metrics[:], g.Metrics[:]) + + return out +} + +func (g *nvmlGpmMetricsGetType) convert() *GpmMetricsGetType { + out := &GpmMetricsGetType{ + Version: g.Version, + NumMetrics: g.NumMetrics, + Sample1: g.Sample1, + Sample2: g.Sample2, + } + copy(out.Metrics[:], g.Metrics[:]) + + return out +} + +// nvml.GpmMetricsGet() +type GpmMetricsGetVType struct { + metricsGet *GpmMetricsGetType +} + +func (l *library) GpmMetricsGetV(metricsGet *GpmMetricsGetType) GpmMetricsGetVType { + return GpmMetricsGetVType{metricsGet} +} + +// nvmlGpmMetricsGetStub is a stub function that can be overridden for testing. +var nvmlGpmMetricsGetStub = nvmlGpmMetricsGet + +func (metricsGetV GpmMetricsGetVType) V1() Return { + metricsGetV.metricsGet.Version = 1 + return gpmMetricsGet(metricsGetV.metricsGet) +} + +func (l *library) GpmMetricsGet(metricsGet *GpmMetricsGetType) Return { + metricsGet.Version = GPM_METRICS_GET_VERSION + return gpmMetricsGet(metricsGet) +} + +func gpmMetricsGet(metricsGet *GpmMetricsGetType) Return { + nvmlMetricsGet := metricsGet.convert() + ret := nvmlGpmMetricsGetStub(nvmlMetricsGet) + *metricsGet = *nvmlMetricsGet.convert() + return ret +} + +// nvml.GpmSampleFree() +func (l *library) GpmSampleFree(gpmSample GpmSample) Return { + return gpmSample.Free() +} + +func (gpmSample nvmlGpmSample) Free() Return { + return nvmlGpmSampleFree(gpmSample) +} + +// nvml.GpmSampleAlloc() +func (l *library) GpmSampleAlloc() (GpmSample, Return) { + var gpmSample nvmlGpmSample + ret := nvmlGpmSampleAlloc(&gpmSample) + return gpmSample, ret +} + +// nvml.GpmSampleGet() +func (l *library) GpmSampleGet(device Device, gpmSample GpmSample) Return { + return gpmSample.Get(device) +} + +func (device nvmlDevice) GpmSampleGet(gpmSample GpmSample) Return { + return gpmSample.Get(device) +} + +func (gpmSample nvmlGpmSample) Get(device Device) Return { + return nvmlGpmSampleGet(nvmlDeviceHandle(device), gpmSample) +} + +// nvml.GpmQueryDeviceSupport() +type GpmSupportV struct { + device nvmlDevice +} + +func (l *library) GpmQueryDeviceSupportV(device Device) GpmSupportV { + return device.GpmQueryDeviceSupportV() +} + +func (device nvmlDevice) GpmQueryDeviceSupportV() GpmSupportV { + return GpmSupportV{device} +} + +func (gpmSupportV GpmSupportV) V1() (GpmSupport, Return) { + var gpmSupport GpmSupport + gpmSupport.Version = STRUCT_VERSION(gpmSupport, 1) + ret := nvmlGpmQueryDeviceSupport(gpmSupportV.device, &gpmSupport) + return gpmSupport, ret +} + +func (l *library) GpmQueryDeviceSupport(device Device) (GpmSupport, Return) { + return device.GpmQueryDeviceSupport() +} + +func (device nvmlDevice) GpmQueryDeviceSupport() (GpmSupport, Return) { + var gpmSupport GpmSupport + gpmSupport.Version = STRUCT_VERSION(gpmSupport, GPM_SUPPORT_VERSION) + ret := nvmlGpmQueryDeviceSupport(device, &gpmSupport) + return gpmSupport, ret +} + +// nvml.GpmMigSampleGet() +func (l *library) GpmMigSampleGet(device Device, gpuInstanceId int, gpmSample GpmSample) Return { + return gpmSample.MigGet(device, gpuInstanceId) +} + +func (device nvmlDevice) GpmMigSampleGet(gpuInstanceId int, gpmSample GpmSample) Return { + return gpmSample.MigGet(device, gpuInstanceId) +} + +func (gpmSample nvmlGpmSample) MigGet(device Device, gpuInstanceId int) Return { + return nvmlGpmMigSampleGet(nvmlDeviceHandle(device), uint32(gpuInstanceId), gpmSample) +} + +// nvml.GpmQueryIfStreamingEnabled() +func (l *library) GpmQueryIfStreamingEnabled(device Device) (uint32, Return) { + return device.GpmQueryIfStreamingEnabled() +} + +func (device nvmlDevice) GpmQueryIfStreamingEnabled() (uint32, Return) { + var state uint32 + ret := nvmlGpmQueryIfStreamingEnabled(device, &state) + return state, ret +} + +// nvml.GpmSetStreamingEnabled() +func (l *library) GpmSetStreamingEnabled(device Device, state uint32) Return { + return device.GpmSetStreamingEnabled(state) +} + +func (device nvmlDevice) GpmSetStreamingEnabled(state uint32) Return { + return nvmlGpmSetStreamingEnabled(device, state) +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/init.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/init.go new file mode 100644 index 00000000..06e64441 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/init.go @@ -0,0 +1,48 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +import "C" + +// nvml.Init() +func (l *library) Init() Return { + if err := l.load(); err != nil { + return ERROR_LIBRARY_NOT_FOUND + } + return nvmlInit() +} + +// nvml.InitWithFlags() +func (l *library) InitWithFlags(flags uint32) Return { + if err := l.load(); err != nil { + return ERROR_LIBRARY_NOT_FOUND + } + return nvmlInitWithFlags(flags) +} + +// nvml.Shutdown() +func (l *library) Shutdown() Return { + ret := nvmlShutdown() + if ret != SUCCESS { + return ret + } + + err := l.close() + if err != nil { + return ERROR_UNKNOWN + } + + return ret +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/lib.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/lib.go new file mode 100644 index 00000000..5a7e6882 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/lib.go @@ -0,0 +1,296 @@ +/** +# Copyright 2023 NVIDIA CORPORATION +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +**/ + +package nvml + +import ( + "errors" + "fmt" + "sync" + + "github.com/NVIDIA/go-nvml/pkg/dl" +) + +import "C" + +const ( + defaultNvmlLibraryName = "libnvidia-ml.so.1" + defaultNvmlLibraryLoadFlags = dl.RTLD_LAZY | dl.RTLD_GLOBAL +) + +var errLibraryNotLoaded = errors.New("library not loaded") +var errLibraryAlreadyLoaded = errors.New("library already loaded") + +// dynamicLibrary is an interface for abstacting the underlying library. +// This also allows for mocking and testing. + +//go:generate moq -stub -out dynamicLibrary_mock.go . dynamicLibrary +type dynamicLibrary interface { + Lookup(string) error + Open() error + Close() error +} + +// library represents an nvml library. +// This includes a reference to the underlying DynamicLibrary +type library struct { + sync.Mutex + path string + refcount refcount + dl dynamicLibrary +} + +var _ Interface = (*library)(nil) + +// libnvml is a global instance of the nvml library. +var libnvml = newLibrary() + +func New(opts ...LibraryOption) Interface { + return newLibrary(opts...) +} + +func newLibrary(opts ...LibraryOption) *library { + l := &library{} + l.init(opts...) + return l +} + +func (l *library) init(opts ...LibraryOption) { + o := libraryOptions{} + for _, opt := range opts { + opt(&o) + } + + if o.path == "" { + o.path = defaultNvmlLibraryName + } + if o.flags == 0 { + o.flags = defaultNvmlLibraryLoadFlags + } + + l.path = o.path + l.dl = dl.New(o.path, o.flags) +} + +func (l *library) Extensions() ExtendedInterface { + return l +} + +// LookupSymbol checks whether the specified library symbol exists in the library. +// Note that this requires that the library be loaded. +func (l *library) LookupSymbol(name string) error { + if l == nil || l.refcount == 0 { + return fmt.Errorf("error looking up %s: %w", name, errLibraryNotLoaded) + } + return l.dl.Lookup(name) +} + +// load initializes the library and updates the versioned symbols. +// Multiple calls to an already loaded library will return without error. +func (l *library) load() (rerr error) { + l.Lock() + defer l.Unlock() + + defer func() { l.refcount.IncOnNoError(rerr) }() + if l.refcount > 0 { + return nil + } + + if err := l.dl.Open(); err != nil { + return fmt.Errorf("error opening %s: %w", l.path, err) + } + + // Update the errorStringFunc to point to nvml.ErrorString + errorStringFunc = nvmlErrorString + + // Update all versioned symbols + l.updateVersionedSymbols() + + return nil +} + +// close the underlying library and ensure that the global pointer to the +// library is set to nil to ensure that subsequent calls to open will reinitialize it. +// Multiple calls to an already closed nvml library will return without error. +func (l *library) close() (rerr error) { + l.Lock() + defer l.Unlock() + + defer func() { l.refcount.DecOnNoError(rerr) }() + if l.refcount != 1 { + return nil + } + + if err := l.dl.Close(); err != nil { + return fmt.Errorf("error closing %s: %w", l.path, err) + } + + // Update the errorStringFunc to point to defaultErrorStringFunc + errorStringFunc = defaultErrorStringFunc + + return nil +} + +// Default all versioned APIs to v1 (to infer the types) +var nvmlInit = nvmlInit_v1 +var nvmlDeviceGetPciInfo = nvmlDeviceGetPciInfo_v1 +var nvmlDeviceGetCount = nvmlDeviceGetCount_v1 +var nvmlDeviceGetHandleByIndex = nvmlDeviceGetHandleByIndex_v1 +var nvmlDeviceGetHandleByPciBusId = nvmlDeviceGetHandleByPciBusId_v1 +var nvmlDeviceGetNvLinkRemotePciInfo = nvmlDeviceGetNvLinkRemotePciInfo_v1 +var nvmlDeviceRemoveGpu = nvmlDeviceRemoveGpu_v1 +var nvmlDeviceGetGridLicensableFeatures = nvmlDeviceGetGridLicensableFeatures_v1 +var nvmlEventSetWait = nvmlEventSetWait_v1 +var nvmlDeviceGetAttributes = nvmlDeviceGetAttributes_v1 +var nvmlComputeInstanceGetInfo = nvmlComputeInstanceGetInfo_v1 +var deviceGetComputeRunningProcesses = deviceGetComputeRunningProcesses_v1 +var deviceGetGraphicsRunningProcesses = deviceGetGraphicsRunningProcesses_v1 +var deviceGetMPSComputeRunningProcesses = deviceGetMPSComputeRunningProcesses_v1 +var GetBlacklistDeviceCount = GetExcludedDeviceCount +var GetBlacklistDeviceInfoByIndex = GetExcludedDeviceInfoByIndex +var nvmlDeviceGetGpuInstancePossiblePlacements = nvmlDeviceGetGpuInstancePossiblePlacements_v1 +var nvmlVgpuInstanceGetLicenseInfo = nvmlVgpuInstanceGetLicenseInfo_v1 +var nvmlDeviceGetDriverModel = nvmlDeviceGetDriverModel_v1 + +// BlacklistDeviceInfo was replaced by ExcludedDeviceInfo +type BlacklistDeviceInfo = ExcludedDeviceInfo + +type ProcessInfo_v1Slice []ProcessInfo_v1 +type ProcessInfo_v2Slice []ProcessInfo_v2 + +func (pis ProcessInfo_v1Slice) ToProcessInfoSlice() []ProcessInfo { + var newInfos []ProcessInfo + for _, pi := range pis { + info := ProcessInfo{ + Pid: pi.Pid, + UsedGpuMemory: pi.UsedGpuMemory, + GpuInstanceId: 0xFFFFFFFF, // GPU instance ID is invalid in v1 + ComputeInstanceId: 0xFFFFFFFF, // Compute instance ID is invalid in v1 + } + newInfos = append(newInfos, info) + } + return newInfos +} + +func (pis ProcessInfo_v2Slice) ToProcessInfoSlice() []ProcessInfo { + var newInfos []ProcessInfo + for _, pi := range pis { + info := ProcessInfo(pi) + newInfos = append(newInfos, info) + } + return newInfos +} + +// updateVersionedSymbols checks for versioned symbols in the loaded dynamic library. +// If newer versioned symbols exist, these replace the default `v1` symbols initialized above. +// When new versioned symbols are added, these would have to be initialized above and have +// corresponding checks and subsequent assignments added below. +func (l *library) updateVersionedSymbols() { + err := l.dl.Lookup("nvmlInit_v2") + if err == nil { + nvmlInit = nvmlInit_v2 + } + err = l.dl.Lookup("nvmlDeviceGetPciInfo_v2") + if err == nil { + nvmlDeviceGetPciInfo = nvmlDeviceGetPciInfo_v2 + } + err = l.dl.Lookup("nvmlDeviceGetPciInfo_v3") + if err == nil { + nvmlDeviceGetPciInfo = nvmlDeviceGetPciInfo_v3 + } + err = l.dl.Lookup("nvmlDeviceGetCount_v2") + if err == nil { + nvmlDeviceGetCount = nvmlDeviceGetCount_v2 + } + err = l.dl.Lookup("nvmlDeviceGetHandleByIndex_v2") + if err == nil { + nvmlDeviceGetHandleByIndex = nvmlDeviceGetHandleByIndex_v2 + } + err = l.dl.Lookup("nvmlDeviceGetHandleByPciBusId_v2") + if err == nil { + nvmlDeviceGetHandleByPciBusId = nvmlDeviceGetHandleByPciBusId_v2 + } + err = l.dl.Lookup("nvmlDeviceGetNvLinkRemotePciInfo_v2") + if err == nil { + nvmlDeviceGetNvLinkRemotePciInfo = nvmlDeviceGetNvLinkRemotePciInfo_v2 + } + // Unable to overwrite nvmlDeviceRemoveGpu() because the v2 function takes + // a different set of parameters than the v1 function. + //err = l.dl.Lookup("nvmlDeviceRemoveGpu_v2") + //if err == nil { + // nvmlDeviceRemoveGpu = nvmlDeviceRemoveGpu_v2 + //} + err = l.dl.Lookup("nvmlDeviceGetGridLicensableFeatures_v2") + if err == nil { + nvmlDeviceGetGridLicensableFeatures = nvmlDeviceGetGridLicensableFeatures_v2 + } + err = l.dl.Lookup("nvmlDeviceGetGridLicensableFeatures_v3") + if err == nil { + nvmlDeviceGetGridLicensableFeatures = nvmlDeviceGetGridLicensableFeatures_v3 + } + err = l.dl.Lookup("nvmlDeviceGetGridLicensableFeatures_v4") + if err == nil { + nvmlDeviceGetGridLicensableFeatures = nvmlDeviceGetGridLicensableFeatures_v4 + } + err = l.dl.Lookup("nvmlEventSetWait_v2") + if err == nil { + nvmlEventSetWait = nvmlEventSetWait_v2 + } + err = l.dl.Lookup("nvmlDeviceGetAttributes_v2") + if err == nil { + nvmlDeviceGetAttributes = nvmlDeviceGetAttributes_v2 + } + err = l.dl.Lookup("nvmlComputeInstanceGetInfo_v2") + if err == nil { + nvmlComputeInstanceGetInfo = nvmlComputeInstanceGetInfo_v2 + } + err = l.dl.Lookup("nvmlDeviceGetComputeRunningProcesses_v2") + if err == nil { + deviceGetComputeRunningProcesses = deviceGetComputeRunningProcesses_v2 + } + err = l.dl.Lookup("nvmlDeviceGetComputeRunningProcesses_v3") + if err == nil { + deviceGetComputeRunningProcesses = deviceGetComputeRunningProcesses_v3 + } + err = l.dl.Lookup("nvmlDeviceGetGraphicsRunningProcesses_v2") + if err == nil { + deviceGetGraphicsRunningProcesses = deviceGetGraphicsRunningProcesses_v2 + } + err = l.dl.Lookup("nvmlDeviceGetGraphicsRunningProcesses_v3") + if err == nil { + deviceGetGraphicsRunningProcesses = deviceGetGraphicsRunningProcesses_v3 + } + err = l.dl.Lookup("nvmlDeviceGetMPSComputeRunningProcesses_v2") + if err == nil { + deviceGetMPSComputeRunningProcesses = deviceGetMPSComputeRunningProcesses_v2 + } + err = l.dl.Lookup("nvmlDeviceGetMPSComputeRunningProcesses_v3") + if err == nil { + deviceGetMPSComputeRunningProcesses = deviceGetMPSComputeRunningProcesses_v3 + } + err = l.dl.Lookup("nvmlDeviceGetGpuInstancePossiblePlacements_v2") + if err == nil { + nvmlDeviceGetGpuInstancePossiblePlacements = nvmlDeviceGetGpuInstancePossiblePlacements_v2 + } + err = l.dl.Lookup("nvmlVgpuInstanceGetLicenseInfo_v2") + if err == nil { + nvmlVgpuInstanceGetLicenseInfo = nvmlVgpuInstanceGetLicenseInfo_v2 + } + err = l.dl.Lookup("nvmlDeviceGetDriverModel_v2") + if err == nil { + nvmlDeviceGetDriverModel = nvmlDeviceGetDriverModel_v2 + } +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.go new file mode 100644 index 00000000..38123a96 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.go @@ -0,0 +1,3749 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +// WARNING: THIS FILE WAS AUTOMATICALLY GENERATED. +// Code generated by https://git.io/c-for-go. DO NOT EDIT. + +package nvml + +/* +#cgo linux LDFLAGS: -Wl,--export-dynamic -Wl,--unresolved-symbols=ignore-in-object-files +#cgo darwin LDFLAGS: -Wl,-undefined,dynamic_lookup +#cgo CFLAGS: -DNVML_NO_UNVERSIONED_FUNC_DEFS=1 +#include "nvml.h" +#include +#include "cgo_helpers.h" +*/ +import "C" +import "unsafe" + +// nvmlInit_v2 function as declared in nvml/nvml.h +func nvmlInit_v2() Return { + __ret := C.nvmlInit_v2() + __v := (Return)(__ret) + return __v +} + +// nvmlInitWithFlags function as declared in nvml/nvml.h +func nvmlInitWithFlags(Flags uint32) Return { + cFlags, _ := (C.uint)(Flags), cgoAllocsUnknown + __ret := C.nvmlInitWithFlags(cFlags) + __v := (Return)(__ret) + return __v +} + +// nvmlShutdown function as declared in nvml/nvml.h +func nvmlShutdown() Return { + __ret := C.nvmlShutdown() + __v := (Return)(__ret) + return __v +} + +// nvmlErrorString function as declared in nvml/nvml.h +func nvmlErrorString(Result Return) string { + cResult, _ := (C.nvmlReturn_t)(Result), cgoAllocsUnknown + __ret := C.nvmlErrorString(cResult) + __v := packPCharString(__ret) + return __v +} + +// nvmlSystemGetDriverVersion function as declared in nvml/nvml.h +func nvmlSystemGetDriverVersion(Version *byte, Length uint32) Return { + cVersion, _ := (*C.char)(unsafe.Pointer(Version)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlSystemGetDriverVersion(cVersion, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetNVMLVersion function as declared in nvml/nvml.h +func nvmlSystemGetNVMLVersion(Version *byte, Length uint32) Return { + cVersion, _ := (*C.char)(unsafe.Pointer(Version)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlSystemGetNVMLVersion(cVersion, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetCudaDriverVersion function as declared in nvml/nvml.h +func nvmlSystemGetCudaDriverVersion(CudaDriverVersion *int32) Return { + cCudaDriverVersion, _ := (*C.int)(unsafe.Pointer(CudaDriverVersion)), cgoAllocsUnknown + __ret := C.nvmlSystemGetCudaDriverVersion(cCudaDriverVersion) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetCudaDriverVersion_v2 function as declared in nvml/nvml.h +func nvmlSystemGetCudaDriverVersion_v2(CudaDriverVersion *int32) Return { + cCudaDriverVersion, _ := (*C.int)(unsafe.Pointer(CudaDriverVersion)), cgoAllocsUnknown + __ret := C.nvmlSystemGetCudaDriverVersion_v2(cCudaDriverVersion) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetProcessName function as declared in nvml/nvml.h +func nvmlSystemGetProcessName(Pid uint32, Name *byte, Length uint32) Return { + cPid, _ := (C.uint)(Pid), cgoAllocsUnknown + cName, _ := (*C.char)(unsafe.Pointer(Name)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlSystemGetProcessName(cPid, cName, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetHicVersion function as declared in nvml/nvml.h +func nvmlSystemGetHicVersion(HwbcCount *uint32, HwbcEntries *HwbcEntry) Return { + cHwbcCount, _ := (*C.uint)(unsafe.Pointer(HwbcCount)), cgoAllocsUnknown + cHwbcEntries, _ := (*C.nvmlHwbcEntry_t)(unsafe.Pointer(HwbcEntries)), cgoAllocsUnknown + __ret := C.nvmlSystemGetHicVersion(cHwbcCount, cHwbcEntries) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetTopologyGpuSet function as declared in nvml/nvml.h +func nvmlSystemGetTopologyGpuSet(CpuNumber uint32, Count *uint32, DeviceArray *nvmlDevice) Return { + cCpuNumber, _ := (C.uint)(CpuNumber), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + cDeviceArray, _ := (*C.nvmlDevice_t)(unsafe.Pointer(DeviceArray)), cgoAllocsUnknown + __ret := C.nvmlSystemGetTopologyGpuSet(cCpuNumber, cCount, cDeviceArray) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetDriverBranch function as declared in nvml/nvml.h +func nvmlSystemGetDriverBranch(BranchInfo *SystemDriverBranchInfo, Length uint32) Return { + cBranchInfo, _ := (*C.nvmlSystemDriverBranchInfo_t)(unsafe.Pointer(BranchInfo)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlSystemGetDriverBranch(cBranchInfo, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetCount function as declared in nvml/nvml.h +func nvmlUnitGetCount(UnitCount *uint32) Return { + cUnitCount, _ := (*C.uint)(unsafe.Pointer(UnitCount)), cgoAllocsUnknown + __ret := C.nvmlUnitGetCount(cUnitCount) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetHandleByIndex function as declared in nvml/nvml.h +func nvmlUnitGetHandleByIndex(Index uint32, nvmlUnit *nvmlUnit) Return { + cIndex, _ := (C.uint)(Index), cgoAllocsUnknown + cnvmlUnit, _ := (*C.nvmlUnit_t)(unsafe.Pointer(nvmlUnit)), cgoAllocsUnknown + __ret := C.nvmlUnitGetHandleByIndex(cIndex, cnvmlUnit) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetUnitInfo function as declared in nvml/nvml.h +func nvmlUnitGetUnitInfo(nvmlUnit nvmlUnit, Info *UnitInfo) Return { + cnvmlUnit, _ := *(*C.nvmlUnit_t)(unsafe.Pointer(&nvmlUnit)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlUnitInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlUnitGetUnitInfo(cnvmlUnit, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetLedState function as declared in nvml/nvml.h +func nvmlUnitGetLedState(nvmlUnit nvmlUnit, State *LedState) Return { + cnvmlUnit, _ := *(*C.nvmlUnit_t)(unsafe.Pointer(&nvmlUnit)), cgoAllocsUnknown + cState, _ := (*C.nvmlLedState_t)(unsafe.Pointer(State)), cgoAllocsUnknown + __ret := C.nvmlUnitGetLedState(cnvmlUnit, cState) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetPsuInfo function as declared in nvml/nvml.h +func nvmlUnitGetPsuInfo(nvmlUnit nvmlUnit, Psu *PSUInfo) Return { + cnvmlUnit, _ := *(*C.nvmlUnit_t)(unsafe.Pointer(&nvmlUnit)), cgoAllocsUnknown + cPsu, _ := (*C.nvmlPSUInfo_t)(unsafe.Pointer(Psu)), cgoAllocsUnknown + __ret := C.nvmlUnitGetPsuInfo(cnvmlUnit, cPsu) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetTemperature function as declared in nvml/nvml.h +func nvmlUnitGetTemperature(nvmlUnit nvmlUnit, _type uint32, Temp *uint32) Return { + cnvmlUnit, _ := *(*C.nvmlUnit_t)(unsafe.Pointer(&nvmlUnit)), cgoAllocsUnknown + c_type, _ := (C.uint)(_type), cgoAllocsUnknown + cTemp, _ := (*C.uint)(unsafe.Pointer(Temp)), cgoAllocsUnknown + __ret := C.nvmlUnitGetTemperature(cnvmlUnit, c_type, cTemp) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetFanSpeedInfo function as declared in nvml/nvml.h +func nvmlUnitGetFanSpeedInfo(nvmlUnit nvmlUnit, FanSpeeds *UnitFanSpeeds) Return { + cnvmlUnit, _ := *(*C.nvmlUnit_t)(unsafe.Pointer(&nvmlUnit)), cgoAllocsUnknown + cFanSpeeds, _ := (*C.nvmlUnitFanSpeeds_t)(unsafe.Pointer(FanSpeeds)), cgoAllocsUnknown + __ret := C.nvmlUnitGetFanSpeedInfo(cnvmlUnit, cFanSpeeds) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitGetDevices function as declared in nvml/nvml.h +func nvmlUnitGetDevices(nvmlUnit nvmlUnit, DeviceCount *uint32, Devices *nvmlDevice) Return { + cnvmlUnit, _ := *(*C.nvmlUnit_t)(unsafe.Pointer(&nvmlUnit)), cgoAllocsUnknown + cDeviceCount, _ := (*C.uint)(unsafe.Pointer(DeviceCount)), cgoAllocsUnknown + cDevices, _ := (*C.nvmlDevice_t)(unsafe.Pointer(Devices)), cgoAllocsUnknown + __ret := C.nvmlUnitGetDevices(cnvmlUnit, cDeviceCount, cDevices) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCount_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetCount_v2(DeviceCount *uint32) Return { + cDeviceCount, _ := (*C.uint)(unsafe.Pointer(DeviceCount)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCount_v2(cDeviceCount) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAttributes_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetAttributes_v2(nvmlDevice nvmlDevice, Attributes *DeviceAttributes) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cAttributes, _ := (*C.nvmlDeviceAttributes_t)(unsafe.Pointer(Attributes)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAttributes_v2(cnvmlDevice, cAttributes) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHandleByIndex_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetHandleByIndex_v2(Index uint32, nvmlDevice *nvmlDevice) Return { + cIndex, _ := (C.uint)(Index), cgoAllocsUnknown + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHandleByIndex_v2(cIndex, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHandleBySerial function as declared in nvml/nvml.h +func nvmlDeviceGetHandleBySerial(Serial string, nvmlDevice *nvmlDevice) Return { + cSerial, _ := unpackPCharString(Serial) + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHandleBySerial(cSerial, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHandleByUUID function as declared in nvml/nvml.h +func nvmlDeviceGetHandleByUUID(Uuid string, nvmlDevice *nvmlDevice) Return { + cUuid, _ := unpackPCharString(Uuid) + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHandleByUUID(cUuid, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHandleByUUIDV function as declared in nvml/nvml.h +func nvmlDeviceGetHandleByUUIDV(Uuid *UUID, nvmlDevice *nvmlDevice) Return { + cUuid, _ := (*C.nvmlUUID_t)(unsafe.Pointer(Uuid)), cgoAllocsUnknown + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHandleByUUIDV(cUuid, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHandleByPciBusId_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetHandleByPciBusId_v2(PciBusId string, nvmlDevice *nvmlDevice) Return { + cPciBusId, _ := unpackPCharString(PciBusId) + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHandleByPciBusId_v2(cPciBusId, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetName function as declared in nvml/nvml.h +func nvmlDeviceGetName(nvmlDevice nvmlDevice, Name *byte, Length uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cName, _ := (*C.char)(unsafe.Pointer(Name)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlDeviceGetName(cnvmlDevice, cName, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetBrand function as declared in nvml/nvml.h +func nvmlDeviceGetBrand(nvmlDevice nvmlDevice, _type *BrandType) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + c_type, _ := (*C.nvmlBrandType_t)(unsafe.Pointer(_type)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetBrand(cnvmlDevice, c_type) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetIndex function as declared in nvml/nvml.h +func nvmlDeviceGetIndex(nvmlDevice nvmlDevice, Index *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIndex, _ := (*C.uint)(unsafe.Pointer(Index)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetIndex(cnvmlDevice, cIndex) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSerial function as declared in nvml/nvml.h +func nvmlDeviceGetSerial(nvmlDevice nvmlDevice, Serial *byte, Length uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSerial, _ := (*C.char)(unsafe.Pointer(Serial)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSerial(cnvmlDevice, cSerial, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetModuleId function as declared in nvml/nvml.h +func nvmlDeviceGetModuleId(nvmlDevice nvmlDevice, ModuleId *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cModuleId, _ := (*C.uint)(unsafe.Pointer(ModuleId)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetModuleId(cnvmlDevice, cModuleId) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetC2cModeInfoV function as declared in nvml/nvml.h +func nvmlDeviceGetC2cModeInfoV(nvmlDevice nvmlDevice, C2cModeInfo *C2cModeInfo_v1) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cC2cModeInfo, _ := (*C.nvmlC2cModeInfo_v1_t)(unsafe.Pointer(C2cModeInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetC2cModeInfoV(cnvmlDevice, cC2cModeInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMemoryAffinity function as declared in nvml/nvml.h +func nvmlDeviceGetMemoryAffinity(nvmlDevice nvmlDevice, NodeSetSize uint32, NodeSet *uint, Scope AffinityScope) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cNodeSetSize, _ := (C.uint)(NodeSetSize), cgoAllocsUnknown + cNodeSet, _ := (*C.ulong)(unsafe.Pointer(NodeSet)), cgoAllocsUnknown + cScope, _ := (C.nvmlAffinityScope_t)(Scope), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMemoryAffinity(cnvmlDevice, cNodeSetSize, cNodeSet, cScope) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCpuAffinityWithinScope function as declared in nvml/nvml.h +func nvmlDeviceGetCpuAffinityWithinScope(nvmlDevice nvmlDevice, CpuSetSize uint32, CpuSet *uint, Scope AffinityScope) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCpuSetSize, _ := (C.uint)(CpuSetSize), cgoAllocsUnknown + cCpuSet, _ := (*C.ulong)(unsafe.Pointer(CpuSet)), cgoAllocsUnknown + cScope, _ := (C.nvmlAffinityScope_t)(Scope), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCpuAffinityWithinScope(cnvmlDevice, cCpuSetSize, cCpuSet, cScope) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCpuAffinity function as declared in nvml/nvml.h +func nvmlDeviceGetCpuAffinity(nvmlDevice nvmlDevice, CpuSetSize uint32, CpuSet *uint) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCpuSetSize, _ := (C.uint)(CpuSetSize), cgoAllocsUnknown + cCpuSet, _ := (*C.ulong)(unsafe.Pointer(CpuSet)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCpuAffinity(cnvmlDevice, cCpuSetSize, cCpuSet) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetCpuAffinity function as declared in nvml/nvml.h +func nvmlDeviceSetCpuAffinity(nvmlDevice nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetCpuAffinity(cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceClearCpuAffinity function as declared in nvml/nvml.h +func nvmlDeviceClearCpuAffinity(nvmlDevice nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceClearCpuAffinity(cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNumaNodeId function as declared in nvml/nvml.h +func nvmlDeviceGetNumaNodeId(nvmlDevice nvmlDevice, Node *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cNode, _ := (*C.uint)(unsafe.Pointer(Node)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNumaNodeId(cnvmlDevice, cNode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAddressingMode function as declared in nvml/nvml.h +func nvmlDeviceGetAddressingMode(nvmlDevice nvmlDevice, Mode *DeviceAddressingMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (*C.nvmlDeviceAddressingMode_t)(unsafe.Pointer(Mode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAddressingMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetRepairStatus function as declared in nvml/nvml.h +func nvmlDeviceGetRepairStatus(nvmlDevice nvmlDevice, RepairStatus *RepairStatus) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cRepairStatus, _ := (*C.nvmlRepairStatus_t)(unsafe.Pointer(RepairStatus)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetRepairStatus(cnvmlDevice, cRepairStatus) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTopologyCommonAncestor function as declared in nvml/nvml.h +func nvmlDeviceGetTopologyCommonAncestor(Device1 nvmlDevice, Device2 nvmlDevice, PathInfo *GpuTopologyLevel) Return { + cDevice1, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&Device1)), cgoAllocsUnknown + cDevice2, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&Device2)), cgoAllocsUnknown + cPathInfo, _ := (*C.nvmlGpuTopologyLevel_t)(unsafe.Pointer(PathInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTopologyCommonAncestor(cDevice1, cDevice2, cPathInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTopologyNearestGpus function as declared in nvml/nvml.h +func nvmlDeviceGetTopologyNearestGpus(nvmlDevice nvmlDevice, Level GpuTopologyLevel, Count *uint32, DeviceArray *nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLevel, _ := (C.nvmlGpuTopologyLevel_t)(Level), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + cDeviceArray, _ := (*C.nvmlDevice_t)(unsafe.Pointer(DeviceArray)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTopologyNearestGpus(cnvmlDevice, cLevel, cCount, cDeviceArray) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetP2PStatus function as declared in nvml/nvml.h +func nvmlDeviceGetP2PStatus(Device1 nvmlDevice, Device2 nvmlDevice, P2pIndex GpuP2PCapsIndex, P2pStatus *GpuP2PStatus) Return { + cDevice1, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&Device1)), cgoAllocsUnknown + cDevice2, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&Device2)), cgoAllocsUnknown + cP2pIndex, _ := (C.nvmlGpuP2PCapsIndex_t)(P2pIndex), cgoAllocsUnknown + cP2pStatus, _ := (*C.nvmlGpuP2PStatus_t)(unsafe.Pointer(P2pStatus)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetP2PStatus(cDevice1, cDevice2, cP2pIndex, cP2pStatus) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetUUID function as declared in nvml/nvml.h +func nvmlDeviceGetUUID(nvmlDevice nvmlDevice, Uuid *byte, Length uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cUuid, _ := (*C.char)(unsafe.Pointer(Uuid)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlDeviceGetUUID(cnvmlDevice, cUuid, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMinorNumber function as declared in nvml/nvml.h +func nvmlDeviceGetMinorNumber(nvmlDevice nvmlDevice, MinorNumber *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMinorNumber, _ := (*C.uint)(unsafe.Pointer(MinorNumber)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMinorNumber(cnvmlDevice, cMinorNumber) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetBoardPartNumber function as declared in nvml/nvml.h +func nvmlDeviceGetBoardPartNumber(nvmlDevice nvmlDevice, PartNumber *byte, Length uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPartNumber, _ := (*C.char)(unsafe.Pointer(PartNumber)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlDeviceGetBoardPartNumber(cnvmlDevice, cPartNumber, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetInforomVersion function as declared in nvml/nvml.h +func nvmlDeviceGetInforomVersion(nvmlDevice nvmlDevice, Object InforomObject, Version *byte, Length uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cObject, _ := (C.nvmlInforomObject_t)(Object), cgoAllocsUnknown + cVersion, _ := (*C.char)(unsafe.Pointer(Version)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlDeviceGetInforomVersion(cnvmlDevice, cObject, cVersion, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetInforomImageVersion function as declared in nvml/nvml.h +func nvmlDeviceGetInforomImageVersion(nvmlDevice nvmlDevice, Version *byte, Length uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVersion, _ := (*C.char)(unsafe.Pointer(Version)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlDeviceGetInforomImageVersion(cnvmlDevice, cVersion, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetInforomConfigurationChecksum function as declared in nvml/nvml.h +func nvmlDeviceGetInforomConfigurationChecksum(nvmlDevice nvmlDevice, Checksum *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cChecksum, _ := (*C.uint)(unsafe.Pointer(Checksum)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetInforomConfigurationChecksum(cnvmlDevice, cChecksum) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceValidateInforom function as declared in nvml/nvml.h +func nvmlDeviceValidateInforom(nvmlDevice nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceValidateInforom(cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetLastBBXFlushTime function as declared in nvml/nvml.h +func nvmlDeviceGetLastBBXFlushTime(nvmlDevice nvmlDevice, Timestamp *uint64, DurationUs *uint) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cTimestamp, _ := (*C.ulonglong)(unsafe.Pointer(Timestamp)), cgoAllocsUnknown + cDurationUs, _ := (*C.ulong)(unsafe.Pointer(DurationUs)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetLastBBXFlushTime(cnvmlDevice, cTimestamp, cDurationUs) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDisplayMode function as declared in nvml/nvml.h +func nvmlDeviceGetDisplayMode(nvmlDevice nvmlDevice, Display *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cDisplay, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(Display)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDisplayMode(cnvmlDevice, cDisplay) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDisplayActive function as declared in nvml/nvml.h +func nvmlDeviceGetDisplayActive(nvmlDevice nvmlDevice, IsActive *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIsActive, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(IsActive)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDisplayActive(cnvmlDevice, cIsActive) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPersistenceMode function as declared in nvml/nvml.h +func nvmlDeviceGetPersistenceMode(nvmlDevice nvmlDevice, Mode *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(Mode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPersistenceMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPciInfoExt function as declared in nvml/nvml.h +func nvmlDeviceGetPciInfoExt(nvmlDevice nvmlDevice, Pci *PciInfoExt) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPci, _ := (*C.nvmlPciInfoExt_t)(unsafe.Pointer(Pci)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPciInfoExt(cnvmlDevice, cPci) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPciInfo_v3 function as declared in nvml/nvml.h +func nvmlDeviceGetPciInfo_v3(nvmlDevice nvmlDevice, Pci *PciInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPci, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(Pci)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPciInfo_v3(cnvmlDevice, cPci) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMaxPcieLinkGeneration function as declared in nvml/nvml.h +func nvmlDeviceGetMaxPcieLinkGeneration(nvmlDevice nvmlDevice, MaxLinkGen *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMaxLinkGen, _ := (*C.uint)(unsafe.Pointer(MaxLinkGen)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMaxPcieLinkGeneration(cnvmlDevice, cMaxLinkGen) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuMaxPcieLinkGeneration function as declared in nvml/nvml.h +func nvmlDeviceGetGpuMaxPcieLinkGeneration(nvmlDevice nvmlDevice, MaxLinkGenDevice *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMaxLinkGenDevice, _ := (*C.uint)(unsafe.Pointer(MaxLinkGenDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuMaxPcieLinkGeneration(cnvmlDevice, cMaxLinkGenDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMaxPcieLinkWidth function as declared in nvml/nvml.h +func nvmlDeviceGetMaxPcieLinkWidth(nvmlDevice nvmlDevice, MaxLinkWidth *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMaxLinkWidth, _ := (*C.uint)(unsafe.Pointer(MaxLinkWidth)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMaxPcieLinkWidth(cnvmlDevice, cMaxLinkWidth) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCurrPcieLinkGeneration function as declared in nvml/nvml.h +func nvmlDeviceGetCurrPcieLinkGeneration(nvmlDevice nvmlDevice, CurrLinkGen *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrLinkGen, _ := (*C.uint)(unsafe.Pointer(CurrLinkGen)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCurrPcieLinkGeneration(cnvmlDevice, cCurrLinkGen) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCurrPcieLinkWidth function as declared in nvml/nvml.h +func nvmlDeviceGetCurrPcieLinkWidth(nvmlDevice nvmlDevice, CurrLinkWidth *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrLinkWidth, _ := (*C.uint)(unsafe.Pointer(CurrLinkWidth)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCurrPcieLinkWidth(cnvmlDevice, cCurrLinkWidth) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPcieThroughput function as declared in nvml/nvml.h +func nvmlDeviceGetPcieThroughput(nvmlDevice nvmlDevice, Counter PcieUtilCounter, Value *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCounter, _ := (C.nvmlPcieUtilCounter_t)(Counter), cgoAllocsUnknown + cValue, _ := (*C.uint)(unsafe.Pointer(Value)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPcieThroughput(cnvmlDevice, cCounter, cValue) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPcieReplayCounter function as declared in nvml/nvml.h +func nvmlDeviceGetPcieReplayCounter(nvmlDevice nvmlDevice, Value *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cValue, _ := (*C.uint)(unsafe.Pointer(Value)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPcieReplayCounter(cnvmlDevice, cValue) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetClockInfo function as declared in nvml/nvml.h +func nvmlDeviceGetClockInfo(nvmlDevice nvmlDevice, _type ClockType, Clock *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + c_type, _ := (C.nvmlClockType_t)(_type), cgoAllocsUnknown + cClock, _ := (*C.uint)(unsafe.Pointer(Clock)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetClockInfo(cnvmlDevice, c_type, cClock) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMaxClockInfo function as declared in nvml/nvml.h +func nvmlDeviceGetMaxClockInfo(nvmlDevice nvmlDevice, _type ClockType, Clock *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + c_type, _ := (C.nvmlClockType_t)(_type), cgoAllocsUnknown + cClock, _ := (*C.uint)(unsafe.Pointer(Clock)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMaxClockInfo(cnvmlDevice, c_type, cClock) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpcClkVfOffset function as declared in nvml/nvml.h +func nvmlDeviceGetGpcClkVfOffset(nvmlDevice nvmlDevice, Offset *int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cOffset, _ := (*C.int)(unsafe.Pointer(Offset)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpcClkVfOffset(cnvmlDevice, cOffset) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetApplicationsClock function as declared in nvml/nvml.h +func nvmlDeviceGetApplicationsClock(nvmlDevice nvmlDevice, ClockType ClockType, ClockMHz *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cClockType, _ := (C.nvmlClockType_t)(ClockType), cgoAllocsUnknown + cClockMHz, _ := (*C.uint)(unsafe.Pointer(ClockMHz)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetApplicationsClock(cnvmlDevice, cClockType, cClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDefaultApplicationsClock function as declared in nvml/nvml.h +func nvmlDeviceGetDefaultApplicationsClock(nvmlDevice nvmlDevice, ClockType ClockType, ClockMHz *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cClockType, _ := (C.nvmlClockType_t)(ClockType), cgoAllocsUnknown + cClockMHz, _ := (*C.uint)(unsafe.Pointer(ClockMHz)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDefaultApplicationsClock(cnvmlDevice, cClockType, cClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetClock function as declared in nvml/nvml.h +func nvmlDeviceGetClock(nvmlDevice nvmlDevice, ClockType ClockType, ClockId ClockId, ClockMHz *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cClockType, _ := (C.nvmlClockType_t)(ClockType), cgoAllocsUnknown + cClockId, _ := (C.nvmlClockId_t)(ClockId), cgoAllocsUnknown + cClockMHz, _ := (*C.uint)(unsafe.Pointer(ClockMHz)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetClock(cnvmlDevice, cClockType, cClockId, cClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMaxCustomerBoostClock function as declared in nvml/nvml.h +func nvmlDeviceGetMaxCustomerBoostClock(nvmlDevice nvmlDevice, ClockType ClockType, ClockMHz *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cClockType, _ := (C.nvmlClockType_t)(ClockType), cgoAllocsUnknown + cClockMHz, _ := (*C.uint)(unsafe.Pointer(ClockMHz)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMaxCustomerBoostClock(cnvmlDevice, cClockType, cClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSupportedMemoryClocks function as declared in nvml/nvml.h +func nvmlDeviceGetSupportedMemoryClocks(nvmlDevice nvmlDevice, Count *uint32, ClocksMHz *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + cClocksMHz, _ := (*C.uint)(unsafe.Pointer(ClocksMHz)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSupportedMemoryClocks(cnvmlDevice, cCount, cClocksMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSupportedGraphicsClocks function as declared in nvml/nvml.h +func nvmlDeviceGetSupportedGraphicsClocks(nvmlDevice nvmlDevice, MemoryClockMHz uint32, Count *uint32, ClocksMHz *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMemoryClockMHz, _ := (C.uint)(MemoryClockMHz), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + cClocksMHz, _ := (*C.uint)(unsafe.Pointer(ClocksMHz)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSupportedGraphicsClocks(cnvmlDevice, cMemoryClockMHz, cCount, cClocksMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAutoBoostedClocksEnabled function as declared in nvml/nvml.h +func nvmlDeviceGetAutoBoostedClocksEnabled(nvmlDevice nvmlDevice, IsEnabled *EnableState, DefaultIsEnabled *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIsEnabled, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(IsEnabled)), cgoAllocsUnknown + cDefaultIsEnabled, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(DefaultIsEnabled)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAutoBoostedClocksEnabled(cnvmlDevice, cIsEnabled, cDefaultIsEnabled) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetFanSpeed function as declared in nvml/nvml.h +func nvmlDeviceGetFanSpeed(nvmlDevice nvmlDevice, Speed *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSpeed, _ := (*C.uint)(unsafe.Pointer(Speed)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetFanSpeed(cnvmlDevice, cSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetFanSpeed_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetFanSpeed_v2(nvmlDevice nvmlDevice, Fan uint32, Speed *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFan, _ := (C.uint)(Fan), cgoAllocsUnknown + cSpeed, _ := (*C.uint)(unsafe.Pointer(Speed)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetFanSpeed_v2(cnvmlDevice, cFan, cSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetFanSpeedRPM function as declared in nvml/nvml.h +func nvmlDeviceGetFanSpeedRPM(nvmlDevice nvmlDevice, FanSpeed *FanSpeedInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFanSpeed, _ := (*C.nvmlFanSpeedInfo_t)(unsafe.Pointer(FanSpeed)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetFanSpeedRPM(cnvmlDevice, cFanSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTargetFanSpeed function as declared in nvml/nvml.h +func nvmlDeviceGetTargetFanSpeed(nvmlDevice nvmlDevice, Fan uint32, TargetSpeed *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFan, _ := (C.uint)(Fan), cgoAllocsUnknown + cTargetSpeed, _ := (*C.uint)(unsafe.Pointer(TargetSpeed)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTargetFanSpeed(cnvmlDevice, cFan, cTargetSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMinMaxFanSpeed function as declared in nvml/nvml.h +func nvmlDeviceGetMinMaxFanSpeed(nvmlDevice nvmlDevice, MinSpeed *uint32, MaxSpeed *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMinSpeed, _ := (*C.uint)(unsafe.Pointer(MinSpeed)), cgoAllocsUnknown + cMaxSpeed, _ := (*C.uint)(unsafe.Pointer(MaxSpeed)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMinMaxFanSpeed(cnvmlDevice, cMinSpeed, cMaxSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetFanControlPolicy_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetFanControlPolicy_v2(nvmlDevice nvmlDevice, Fan uint32, Policy *FanControlPolicy) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFan, _ := (C.uint)(Fan), cgoAllocsUnknown + cPolicy, _ := (*C.nvmlFanControlPolicy_t)(unsafe.Pointer(Policy)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetFanControlPolicy_v2(cnvmlDevice, cFan, cPolicy) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNumFans function as declared in nvml/nvml.h +func nvmlDeviceGetNumFans(nvmlDevice nvmlDevice, NumFans *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cNumFans, _ := (*C.uint)(unsafe.Pointer(NumFans)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNumFans(cnvmlDevice, cNumFans) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTemperature function as declared in nvml/nvml.h +func nvmlDeviceGetTemperature(nvmlDevice nvmlDevice, SensorType TemperatureSensors, Temp *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSensorType, _ := (C.nvmlTemperatureSensors_t)(SensorType), cgoAllocsUnknown + cTemp, _ := (*C.uint)(unsafe.Pointer(Temp)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTemperature(cnvmlDevice, cSensorType, cTemp) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCoolerInfo function as declared in nvml/nvml.h +func nvmlDeviceGetCoolerInfo(nvmlDevice nvmlDevice, CoolerInfo *CoolerInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCoolerInfo, _ := (*C.nvmlCoolerInfo_t)(unsafe.Pointer(CoolerInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCoolerInfo(cnvmlDevice, cCoolerInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTemperatureV function as declared in nvml/nvml.h +func nvmlDeviceGetTemperatureV(nvmlDevice nvmlDevice, Temperature *Temperature) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cTemperature, _ := (*C.nvmlTemperature_t)(unsafe.Pointer(Temperature)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTemperatureV(cnvmlDevice, cTemperature) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTemperatureThreshold function as declared in nvml/nvml.h +func nvmlDeviceGetTemperatureThreshold(nvmlDevice nvmlDevice, ThresholdType TemperatureThresholds, Temp *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cThresholdType, _ := (C.nvmlTemperatureThresholds_t)(ThresholdType), cgoAllocsUnknown + cTemp, _ := (*C.uint)(unsafe.Pointer(Temp)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTemperatureThreshold(cnvmlDevice, cThresholdType, cTemp) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMarginTemperature function as declared in nvml/nvml.h +func nvmlDeviceGetMarginTemperature(nvmlDevice nvmlDevice, MarginTempInfo *MarginTemperature) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMarginTempInfo, _ := (*C.nvmlMarginTemperature_t)(unsafe.Pointer(MarginTempInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMarginTemperature(cnvmlDevice, cMarginTempInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetThermalSettings function as declared in nvml/nvml.h +func nvmlDeviceGetThermalSettings(nvmlDevice nvmlDevice, SensorIndex uint32, PThermalSettings *GpuThermalSettings) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSensorIndex, _ := (C.uint)(SensorIndex), cgoAllocsUnknown + cPThermalSettings, _ := (*C.nvmlGpuThermalSettings_t)(unsafe.Pointer(PThermalSettings)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetThermalSettings(cnvmlDevice, cSensorIndex, cPThermalSettings) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPerformanceState function as declared in nvml/nvml.h +func nvmlDeviceGetPerformanceState(nvmlDevice nvmlDevice, PState *Pstates) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPState, _ := (*C.nvmlPstates_t)(unsafe.Pointer(PState)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPerformanceState(cnvmlDevice, cPState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCurrentClocksEventReasons function as declared in nvml/nvml.h +func nvmlDeviceGetCurrentClocksEventReasons(nvmlDevice nvmlDevice, ClocksEventReasons *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cClocksEventReasons, _ := (*C.ulonglong)(unsafe.Pointer(ClocksEventReasons)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCurrentClocksEventReasons(cnvmlDevice, cClocksEventReasons) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCurrentClocksThrottleReasons function as declared in nvml/nvml.h +func nvmlDeviceGetCurrentClocksThrottleReasons(nvmlDevice nvmlDevice, ClocksThrottleReasons *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cClocksThrottleReasons, _ := (*C.ulonglong)(unsafe.Pointer(ClocksThrottleReasons)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCurrentClocksThrottleReasons(cnvmlDevice, cClocksThrottleReasons) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSupportedClocksEventReasons function as declared in nvml/nvml.h +func nvmlDeviceGetSupportedClocksEventReasons(nvmlDevice nvmlDevice, SupportedClocksEventReasons *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSupportedClocksEventReasons, _ := (*C.ulonglong)(unsafe.Pointer(SupportedClocksEventReasons)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSupportedClocksEventReasons(cnvmlDevice, cSupportedClocksEventReasons) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSupportedClocksThrottleReasons function as declared in nvml/nvml.h +func nvmlDeviceGetSupportedClocksThrottleReasons(nvmlDevice nvmlDevice, SupportedClocksThrottleReasons *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSupportedClocksThrottleReasons, _ := (*C.ulonglong)(unsafe.Pointer(SupportedClocksThrottleReasons)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSupportedClocksThrottleReasons(cnvmlDevice, cSupportedClocksThrottleReasons) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerState function as declared in nvml/nvml.h +func nvmlDeviceGetPowerState(nvmlDevice nvmlDevice, PState *Pstates) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPState, _ := (*C.nvmlPstates_t)(unsafe.Pointer(PState)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerState(cnvmlDevice, cPState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDynamicPstatesInfo function as declared in nvml/nvml.h +func nvmlDeviceGetDynamicPstatesInfo(nvmlDevice nvmlDevice, PDynamicPstatesInfo *GpuDynamicPstatesInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPDynamicPstatesInfo, _ := (*C.nvmlGpuDynamicPstatesInfo_t)(unsafe.Pointer(PDynamicPstatesInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDynamicPstatesInfo(cnvmlDevice, cPDynamicPstatesInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMemClkVfOffset function as declared in nvml/nvml.h +func nvmlDeviceGetMemClkVfOffset(nvmlDevice nvmlDevice, Offset *int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cOffset, _ := (*C.int)(unsafe.Pointer(Offset)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMemClkVfOffset(cnvmlDevice, cOffset) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMinMaxClockOfPState function as declared in nvml/nvml.h +func nvmlDeviceGetMinMaxClockOfPState(nvmlDevice nvmlDevice, _type ClockType, Pstate Pstates, MinClockMHz *uint32, MaxClockMHz *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + c_type, _ := (C.nvmlClockType_t)(_type), cgoAllocsUnknown + cPstate, _ := (C.nvmlPstates_t)(Pstate), cgoAllocsUnknown + cMinClockMHz, _ := (*C.uint)(unsafe.Pointer(MinClockMHz)), cgoAllocsUnknown + cMaxClockMHz, _ := (*C.uint)(unsafe.Pointer(MaxClockMHz)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMinMaxClockOfPState(cnvmlDevice, c_type, cPstate, cMinClockMHz, cMaxClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSupportedPerformanceStates function as declared in nvml/nvml.h +func nvmlDeviceGetSupportedPerformanceStates(nvmlDevice nvmlDevice, Pstates *Pstates, Size uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPstates, _ := (*C.nvmlPstates_t)(unsafe.Pointer(Pstates)), cgoAllocsUnknown + cSize, _ := (C.uint)(Size), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSupportedPerformanceStates(cnvmlDevice, cPstates, cSize) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpcClkMinMaxVfOffset function as declared in nvml/nvml.h +func nvmlDeviceGetGpcClkMinMaxVfOffset(nvmlDevice nvmlDevice, MinOffset *int32, MaxOffset *int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMinOffset, _ := (*C.int)(unsafe.Pointer(MinOffset)), cgoAllocsUnknown + cMaxOffset, _ := (*C.int)(unsafe.Pointer(MaxOffset)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpcClkMinMaxVfOffset(cnvmlDevice, cMinOffset, cMaxOffset) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMemClkMinMaxVfOffset function as declared in nvml/nvml.h +func nvmlDeviceGetMemClkMinMaxVfOffset(nvmlDevice nvmlDevice, MinOffset *int32, MaxOffset *int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMinOffset, _ := (*C.int)(unsafe.Pointer(MinOffset)), cgoAllocsUnknown + cMaxOffset, _ := (*C.int)(unsafe.Pointer(MaxOffset)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMemClkMinMaxVfOffset(cnvmlDevice, cMinOffset, cMaxOffset) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetClockOffsets function as declared in nvml/nvml.h +func nvmlDeviceGetClockOffsets(nvmlDevice nvmlDevice, Info *ClockOffset) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlClockOffset_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetClockOffsets(cnvmlDevice, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetClockOffsets function as declared in nvml/nvml.h +func nvmlDeviceSetClockOffsets(nvmlDevice nvmlDevice, Info *ClockOffset) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlClockOffset_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetClockOffsets(cnvmlDevice, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPerformanceModes function as declared in nvml/nvml.h +func nvmlDeviceGetPerformanceModes(nvmlDevice nvmlDevice, PerfModes *DevicePerfModes) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPerfModes, _ := (*C.nvmlDevicePerfModes_t)(unsafe.Pointer(PerfModes)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPerformanceModes(cnvmlDevice, cPerfModes) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCurrentClockFreqs function as declared in nvml/nvml.h +func nvmlDeviceGetCurrentClockFreqs(nvmlDevice nvmlDevice, CurrentClockFreqs *DeviceCurrentClockFreqs) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrentClockFreqs, _ := (*C.nvmlDeviceCurrentClockFreqs_t)(unsafe.Pointer(CurrentClockFreqs)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCurrentClockFreqs(cnvmlDevice, cCurrentClockFreqs) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerManagementMode function as declared in nvml/nvml.h +func nvmlDeviceGetPowerManagementMode(nvmlDevice nvmlDevice, Mode *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(Mode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerManagementMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerManagementLimit function as declared in nvml/nvml.h +func nvmlDeviceGetPowerManagementLimit(nvmlDevice nvmlDevice, Limit *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLimit, _ := (*C.uint)(unsafe.Pointer(Limit)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerManagementLimit(cnvmlDevice, cLimit) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerManagementLimitConstraints function as declared in nvml/nvml.h +func nvmlDeviceGetPowerManagementLimitConstraints(nvmlDevice nvmlDevice, MinLimit *uint32, MaxLimit *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMinLimit, _ := (*C.uint)(unsafe.Pointer(MinLimit)), cgoAllocsUnknown + cMaxLimit, _ := (*C.uint)(unsafe.Pointer(MaxLimit)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerManagementLimitConstraints(cnvmlDevice, cMinLimit, cMaxLimit) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerManagementDefaultLimit function as declared in nvml/nvml.h +func nvmlDeviceGetPowerManagementDefaultLimit(nvmlDevice nvmlDevice, DefaultLimit *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cDefaultLimit, _ := (*C.uint)(unsafe.Pointer(DefaultLimit)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerManagementDefaultLimit(cnvmlDevice, cDefaultLimit) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerUsage function as declared in nvml/nvml.h +func nvmlDeviceGetPowerUsage(nvmlDevice nvmlDevice, Power *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPower, _ := (*C.uint)(unsafe.Pointer(Power)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerUsage(cnvmlDevice, cPower) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerMizerMode_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetPowerMizerMode_v1(nvmlDevice nvmlDevice, PowerMizerMode *DevicePowerMizerModes_v1) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPowerMizerMode, _ := (*C.nvmlDevicePowerMizerModes_v1_t)(unsafe.Pointer(PowerMizerMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerMizerMode_v1(cnvmlDevice, cPowerMizerMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetPowerMizerMode_v1 function as declared in nvml/nvml.h +func nvmlDeviceSetPowerMizerMode_v1(nvmlDevice nvmlDevice, PowerMizerMode *DevicePowerMizerModes_v1) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPowerMizerMode, _ := (*C.nvmlDevicePowerMizerModes_v1_t)(unsafe.Pointer(PowerMizerMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetPowerMizerMode_v1(cnvmlDevice, cPowerMizerMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTotalEnergyConsumption function as declared in nvml/nvml.h +func nvmlDeviceGetTotalEnergyConsumption(nvmlDevice nvmlDevice, Energy *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cEnergy, _ := (*C.ulonglong)(unsafe.Pointer(Energy)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTotalEnergyConsumption(cnvmlDevice, cEnergy) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetEnforcedPowerLimit function as declared in nvml/nvml.h +func nvmlDeviceGetEnforcedPowerLimit(nvmlDevice nvmlDevice, Limit *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLimit, _ := (*C.uint)(unsafe.Pointer(Limit)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetEnforcedPowerLimit(cnvmlDevice, cLimit) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuOperationMode function as declared in nvml/nvml.h +func nvmlDeviceGetGpuOperationMode(nvmlDevice nvmlDevice, Current *GpuOperationMode, Pending *GpuOperationMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrent, _ := (*C.nvmlGpuOperationMode_t)(unsafe.Pointer(Current)), cgoAllocsUnknown + cPending, _ := (*C.nvmlGpuOperationMode_t)(unsafe.Pointer(Pending)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuOperationMode(cnvmlDevice, cCurrent, cPending) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMemoryInfo function as declared in nvml/nvml.h +func nvmlDeviceGetMemoryInfo(nvmlDevice nvmlDevice, Memory *Memory) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMemory, _ := (*C.nvmlMemory_t)(unsafe.Pointer(Memory)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMemoryInfo(cnvmlDevice, cMemory) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMemoryInfo_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetMemoryInfo_v2(nvmlDevice nvmlDevice, Memory *Memory_v2) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMemory, _ := (*C.nvmlMemory_v2_t)(unsafe.Pointer(Memory)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMemoryInfo_v2(cnvmlDevice, cMemory) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetComputeMode function as declared in nvml/nvml.h +func nvmlDeviceGetComputeMode(nvmlDevice nvmlDevice, Mode *ComputeMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (*C.nvmlComputeMode_t)(unsafe.Pointer(Mode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetComputeMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCudaComputeCapability function as declared in nvml/nvml.h +func nvmlDeviceGetCudaComputeCapability(nvmlDevice nvmlDevice, Major *int32, Minor *int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMajor, _ := (*C.int)(unsafe.Pointer(Major)), cgoAllocsUnknown + cMinor, _ := (*C.int)(unsafe.Pointer(Minor)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCudaComputeCapability(cnvmlDevice, cMajor, cMinor) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDramEncryptionMode function as declared in nvml/nvml.h +func nvmlDeviceGetDramEncryptionMode(nvmlDevice nvmlDevice, Current *DramEncryptionInfo, Pending *DramEncryptionInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrent, _ := (*C.nvmlDramEncryptionInfo_t)(unsafe.Pointer(Current)), cgoAllocsUnknown + cPending, _ := (*C.nvmlDramEncryptionInfo_t)(unsafe.Pointer(Pending)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDramEncryptionMode(cnvmlDevice, cCurrent, cPending) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetDramEncryptionMode function as declared in nvml/nvml.h +func nvmlDeviceSetDramEncryptionMode(nvmlDevice nvmlDevice, DramEncryption *DramEncryptionInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cDramEncryption, _ := (*C.nvmlDramEncryptionInfo_t)(unsafe.Pointer(DramEncryption)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetDramEncryptionMode(cnvmlDevice, cDramEncryption) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetEccMode function as declared in nvml/nvml.h +func nvmlDeviceGetEccMode(nvmlDevice nvmlDevice, Current *EnableState, Pending *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrent, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(Current)), cgoAllocsUnknown + cPending, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(Pending)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetEccMode(cnvmlDevice, cCurrent, cPending) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDefaultEccMode function as declared in nvml/nvml.h +func nvmlDeviceGetDefaultEccMode(nvmlDevice nvmlDevice, DefaultMode *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cDefaultMode, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(DefaultMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDefaultEccMode(cnvmlDevice, cDefaultMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetBoardId function as declared in nvml/nvml.h +func nvmlDeviceGetBoardId(nvmlDevice nvmlDevice, BoardId *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cBoardId, _ := (*C.uint)(unsafe.Pointer(BoardId)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetBoardId(cnvmlDevice, cBoardId) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMultiGpuBoard function as declared in nvml/nvml.h +func nvmlDeviceGetMultiGpuBoard(nvmlDevice nvmlDevice, MultiGpuBool *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMultiGpuBool, _ := (*C.uint)(unsafe.Pointer(MultiGpuBool)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMultiGpuBoard(cnvmlDevice, cMultiGpuBool) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetTotalEccErrors function as declared in nvml/nvml.h +func nvmlDeviceGetTotalEccErrors(nvmlDevice nvmlDevice, ErrorType MemoryErrorType, CounterType EccCounterType, EccCounts *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cErrorType, _ := (C.nvmlMemoryErrorType_t)(ErrorType), cgoAllocsUnknown + cCounterType, _ := (C.nvmlEccCounterType_t)(CounterType), cgoAllocsUnknown + cEccCounts, _ := (*C.ulonglong)(unsafe.Pointer(EccCounts)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetTotalEccErrors(cnvmlDevice, cErrorType, cCounterType, cEccCounts) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDetailedEccErrors function as declared in nvml/nvml.h +func nvmlDeviceGetDetailedEccErrors(nvmlDevice nvmlDevice, ErrorType MemoryErrorType, CounterType EccCounterType, EccCounts *EccErrorCounts) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cErrorType, _ := (C.nvmlMemoryErrorType_t)(ErrorType), cgoAllocsUnknown + cCounterType, _ := (C.nvmlEccCounterType_t)(CounterType), cgoAllocsUnknown + cEccCounts, _ := (*C.nvmlEccErrorCounts_t)(unsafe.Pointer(EccCounts)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDetailedEccErrors(cnvmlDevice, cErrorType, cCounterType, cEccCounts) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMemoryErrorCounter function as declared in nvml/nvml.h +func nvmlDeviceGetMemoryErrorCounter(nvmlDevice nvmlDevice, ErrorType MemoryErrorType, CounterType EccCounterType, LocationType MemoryLocation, Count *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cErrorType, _ := (C.nvmlMemoryErrorType_t)(ErrorType), cgoAllocsUnknown + cCounterType, _ := (C.nvmlEccCounterType_t)(CounterType), cgoAllocsUnknown + cLocationType, _ := (C.nvmlMemoryLocation_t)(LocationType), cgoAllocsUnknown + cCount, _ := (*C.ulonglong)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMemoryErrorCounter(cnvmlDevice, cErrorType, cCounterType, cLocationType, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetUtilizationRates function as declared in nvml/nvml.h +func nvmlDeviceGetUtilizationRates(nvmlDevice nvmlDevice, Utilization *Utilization) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cUtilization, _ := (*C.nvmlUtilization_t)(unsafe.Pointer(Utilization)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetUtilizationRates(cnvmlDevice, cUtilization) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetEncoderUtilization function as declared in nvml/nvml.h +func nvmlDeviceGetEncoderUtilization(nvmlDevice nvmlDevice, Utilization *uint32, SamplingPeriodUs *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cUtilization, _ := (*C.uint)(unsafe.Pointer(Utilization)), cgoAllocsUnknown + cSamplingPeriodUs, _ := (*C.uint)(unsafe.Pointer(SamplingPeriodUs)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetEncoderUtilization(cnvmlDevice, cUtilization, cSamplingPeriodUs) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetEncoderCapacity function as declared in nvml/nvml.h +func nvmlDeviceGetEncoderCapacity(nvmlDevice nvmlDevice, EncoderQueryType EncoderType, EncoderCapacity *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cEncoderQueryType, _ := (C.nvmlEncoderType_t)(EncoderQueryType), cgoAllocsUnknown + cEncoderCapacity, _ := (*C.uint)(unsafe.Pointer(EncoderCapacity)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetEncoderCapacity(cnvmlDevice, cEncoderQueryType, cEncoderCapacity) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetEncoderStats function as declared in nvml/nvml.h +func nvmlDeviceGetEncoderStats(nvmlDevice nvmlDevice, SessionCount *uint32, AverageFps *uint32, AverageLatency *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSessionCount, _ := (*C.uint)(unsafe.Pointer(SessionCount)), cgoAllocsUnknown + cAverageFps, _ := (*C.uint)(unsafe.Pointer(AverageFps)), cgoAllocsUnknown + cAverageLatency, _ := (*C.uint)(unsafe.Pointer(AverageLatency)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetEncoderStats(cnvmlDevice, cSessionCount, cAverageFps, cAverageLatency) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetEncoderSessions function as declared in nvml/nvml.h +func nvmlDeviceGetEncoderSessions(nvmlDevice nvmlDevice, SessionCount *uint32, SessionInfos *EncoderSessionInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSessionCount, _ := (*C.uint)(unsafe.Pointer(SessionCount)), cgoAllocsUnknown + cSessionInfos, _ := (*C.nvmlEncoderSessionInfo_t)(unsafe.Pointer(SessionInfos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetEncoderSessions(cnvmlDevice, cSessionCount, cSessionInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDecoderUtilization function as declared in nvml/nvml.h +func nvmlDeviceGetDecoderUtilization(nvmlDevice nvmlDevice, Utilization *uint32, SamplingPeriodUs *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cUtilization, _ := (*C.uint)(unsafe.Pointer(Utilization)), cgoAllocsUnknown + cSamplingPeriodUs, _ := (*C.uint)(unsafe.Pointer(SamplingPeriodUs)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDecoderUtilization(cnvmlDevice, cUtilization, cSamplingPeriodUs) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetJpgUtilization function as declared in nvml/nvml.h +func nvmlDeviceGetJpgUtilization(nvmlDevice nvmlDevice, Utilization *uint32, SamplingPeriodUs *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cUtilization, _ := (*C.uint)(unsafe.Pointer(Utilization)), cgoAllocsUnknown + cSamplingPeriodUs, _ := (*C.uint)(unsafe.Pointer(SamplingPeriodUs)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetJpgUtilization(cnvmlDevice, cUtilization, cSamplingPeriodUs) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetOfaUtilization function as declared in nvml/nvml.h +func nvmlDeviceGetOfaUtilization(nvmlDevice nvmlDevice, Utilization *uint32, SamplingPeriodUs *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cUtilization, _ := (*C.uint)(unsafe.Pointer(Utilization)), cgoAllocsUnknown + cSamplingPeriodUs, _ := (*C.uint)(unsafe.Pointer(SamplingPeriodUs)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetOfaUtilization(cnvmlDevice, cUtilization, cSamplingPeriodUs) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetFBCStats function as declared in nvml/nvml.h +func nvmlDeviceGetFBCStats(nvmlDevice nvmlDevice, FbcStats *FBCStats) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFbcStats, _ := (*C.nvmlFBCStats_t)(unsafe.Pointer(FbcStats)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetFBCStats(cnvmlDevice, cFbcStats) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetFBCSessions function as declared in nvml/nvml.h +func nvmlDeviceGetFBCSessions(nvmlDevice nvmlDevice, SessionCount *uint32, SessionInfo *FBCSessionInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSessionCount, _ := (*C.uint)(unsafe.Pointer(SessionCount)), cgoAllocsUnknown + cSessionInfo, _ := (*C.nvmlFBCSessionInfo_t)(unsafe.Pointer(SessionInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetFBCSessions(cnvmlDevice, cSessionCount, cSessionInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDriverModel_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetDriverModel_v2(nvmlDevice nvmlDevice, Current *DriverModel, Pending *DriverModel) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrent, _ := (*C.nvmlDriverModel_t)(unsafe.Pointer(Current)), cgoAllocsUnknown + cPending, _ := (*C.nvmlDriverModel_t)(unsafe.Pointer(Pending)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDriverModel_v2(cnvmlDevice, cCurrent, cPending) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVbiosVersion function as declared in nvml/nvml.h +func nvmlDeviceGetVbiosVersion(nvmlDevice nvmlDevice, Version *byte, Length uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVersion, _ := (*C.char)(unsafe.Pointer(Version)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVbiosVersion(cnvmlDevice, cVersion, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetBridgeChipInfo function as declared in nvml/nvml.h +func nvmlDeviceGetBridgeChipInfo(nvmlDevice nvmlDevice, BridgeHierarchy *BridgeChipHierarchy) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cBridgeHierarchy, _ := (*C.nvmlBridgeChipHierarchy_t)(unsafe.Pointer(BridgeHierarchy)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetBridgeChipInfo(cnvmlDevice, cBridgeHierarchy) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetComputeRunningProcesses_v3 function as declared in nvml/nvml.h +func nvmlDeviceGetComputeRunningProcesses_v3(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetComputeRunningProcesses_v3(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGraphicsRunningProcesses_v3 function as declared in nvml/nvml.h +func nvmlDeviceGetGraphicsRunningProcesses_v3(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGraphicsRunningProcesses_v3(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMPSComputeRunningProcesses_v3 function as declared in nvml/nvml.h +func nvmlDeviceGetMPSComputeRunningProcesses_v3(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMPSComputeRunningProcesses_v3(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetRunningProcessDetailList function as declared in nvml/nvml.h +func nvmlDeviceGetRunningProcessDetailList(nvmlDevice nvmlDevice, Plist *ProcessDetailList) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPlist, _ := (*C.nvmlProcessDetailList_t)(unsafe.Pointer(Plist)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetRunningProcessDetailList(cnvmlDevice, cPlist) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceOnSameBoard function as declared in nvml/nvml.h +func nvmlDeviceOnSameBoard(Device1 nvmlDevice, Device2 nvmlDevice, OnSameBoard *int32) Return { + cDevice1, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&Device1)), cgoAllocsUnknown + cDevice2, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&Device2)), cgoAllocsUnknown + cOnSameBoard, _ := (*C.int)(unsafe.Pointer(OnSameBoard)), cgoAllocsUnknown + __ret := C.nvmlDeviceOnSameBoard(cDevice1, cDevice2, cOnSameBoard) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAPIRestriction function as declared in nvml/nvml.h +func nvmlDeviceGetAPIRestriction(nvmlDevice nvmlDevice, ApiType RestrictedAPI, IsRestricted *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cApiType, _ := (C.nvmlRestrictedAPI_t)(ApiType), cgoAllocsUnknown + cIsRestricted, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(IsRestricted)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAPIRestriction(cnvmlDevice, cApiType, cIsRestricted) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSamples function as declared in nvml/nvml.h +func nvmlDeviceGetSamples(nvmlDevice nvmlDevice, _type SamplingType, LastSeenTimeStamp uint64, SampleValType *ValueType, SampleCount *uint32, Samples *Sample) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + c_type, _ := (C.nvmlSamplingType_t)(_type), cgoAllocsUnknown + cLastSeenTimeStamp, _ := (C.ulonglong)(LastSeenTimeStamp), cgoAllocsUnknown + cSampleValType, _ := (*C.nvmlValueType_t)(unsafe.Pointer(SampleValType)), cgoAllocsUnknown + cSampleCount, _ := (*C.uint)(unsafe.Pointer(SampleCount)), cgoAllocsUnknown + cSamples, _ := (*C.nvmlSample_t)(unsafe.Pointer(Samples)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSamples(cnvmlDevice, c_type, cLastSeenTimeStamp, cSampleValType, cSampleCount, cSamples) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetBAR1MemoryInfo function as declared in nvml/nvml.h +func nvmlDeviceGetBAR1MemoryInfo(nvmlDevice nvmlDevice, Bar1Memory *BAR1Memory) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cBar1Memory, _ := (*C.nvmlBAR1Memory_t)(unsafe.Pointer(Bar1Memory)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetBAR1MemoryInfo(cnvmlDevice, cBar1Memory) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetViolationStatus function as declared in nvml/nvml.h +func nvmlDeviceGetViolationStatus(nvmlDevice nvmlDevice, PerfPolicyType PerfPolicyType, ViolTime *ViolationTime) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPerfPolicyType, _ := (C.nvmlPerfPolicyType_t)(PerfPolicyType), cgoAllocsUnknown + cViolTime, _ := (*C.nvmlViolationTime_t)(unsafe.Pointer(ViolTime)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetViolationStatus(cnvmlDevice, cPerfPolicyType, cViolTime) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetIrqNum function as declared in nvml/nvml.h +func nvmlDeviceGetIrqNum(nvmlDevice nvmlDevice, IrqNum *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIrqNum, _ := (*C.uint)(unsafe.Pointer(IrqNum)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetIrqNum(cnvmlDevice, cIrqNum) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNumGpuCores function as declared in nvml/nvml.h +func nvmlDeviceGetNumGpuCores(nvmlDevice nvmlDevice, NumCores *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cNumCores, _ := (*C.uint)(unsafe.Pointer(NumCores)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNumGpuCores(cnvmlDevice, cNumCores) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPowerSource function as declared in nvml/nvml.h +func nvmlDeviceGetPowerSource(nvmlDevice nvmlDevice, PowerSource *PowerSource) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPowerSource, _ := (*C.nvmlPowerSource_t)(unsafe.Pointer(PowerSource)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPowerSource(cnvmlDevice, cPowerSource) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMemoryBusWidth function as declared in nvml/nvml.h +func nvmlDeviceGetMemoryBusWidth(nvmlDevice nvmlDevice, BusWidth *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cBusWidth, _ := (*C.uint)(unsafe.Pointer(BusWidth)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMemoryBusWidth(cnvmlDevice, cBusWidth) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPcieLinkMaxSpeed function as declared in nvml/nvml.h +func nvmlDeviceGetPcieLinkMaxSpeed(nvmlDevice nvmlDevice, MaxSpeed *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMaxSpeed, _ := (*C.uint)(unsafe.Pointer(MaxSpeed)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPcieLinkMaxSpeed(cnvmlDevice, cMaxSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPcieSpeed function as declared in nvml/nvml.h +func nvmlDeviceGetPcieSpeed(nvmlDevice nvmlDevice, PcieSpeed *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPcieSpeed, _ := (*C.uint)(unsafe.Pointer(PcieSpeed)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPcieSpeed(cnvmlDevice, cPcieSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAdaptiveClockInfoStatus function as declared in nvml/nvml.h +func nvmlDeviceGetAdaptiveClockInfoStatus(nvmlDevice nvmlDevice, AdaptiveClockStatus *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cAdaptiveClockStatus, _ := (*C.uint)(unsafe.Pointer(AdaptiveClockStatus)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAdaptiveClockInfoStatus(cnvmlDevice, cAdaptiveClockStatus) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetBusType function as declared in nvml/nvml.h +func nvmlDeviceGetBusType(nvmlDevice nvmlDevice, _type *BusType) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + c_type, _ := (*C.nvmlBusType_t)(unsafe.Pointer(_type)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetBusType(cnvmlDevice, c_type) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuFabricInfo function as declared in nvml/nvml.h +func nvmlDeviceGetGpuFabricInfo(nvmlDevice nvmlDevice, GpuFabricInfo *GpuFabricInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cGpuFabricInfo, _ := (*C.nvmlGpuFabricInfo_t)(unsafe.Pointer(GpuFabricInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuFabricInfo(cnvmlDevice, cGpuFabricInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuFabricInfoV function as declared in nvml/nvml.h +func nvmlDeviceGetGpuFabricInfoV(nvmlDevice nvmlDevice, GpuFabricInfo *GpuFabricInfoV) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cGpuFabricInfo, _ := (*C.nvmlGpuFabricInfoV_t)(unsafe.Pointer(GpuFabricInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuFabricInfoV(cnvmlDevice, cGpuFabricInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetConfComputeCapabilities function as declared in nvml/nvml.h +func nvmlSystemGetConfComputeCapabilities(Capabilities *ConfComputeSystemCaps) Return { + cCapabilities, _ := (*C.nvmlConfComputeSystemCaps_t)(unsafe.Pointer(Capabilities)), cgoAllocsUnknown + __ret := C.nvmlSystemGetConfComputeCapabilities(cCapabilities) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetConfComputeState function as declared in nvml/nvml.h +func nvmlSystemGetConfComputeState(State *ConfComputeSystemState) Return { + cState, _ := (*C.nvmlConfComputeSystemState_t)(unsafe.Pointer(State)), cgoAllocsUnknown + __ret := C.nvmlSystemGetConfComputeState(cState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetConfComputeMemSizeInfo function as declared in nvml/nvml.h +func nvmlDeviceGetConfComputeMemSizeInfo(nvmlDevice nvmlDevice, MemInfo *ConfComputeMemSizeInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMemInfo, _ := (*C.nvmlConfComputeMemSizeInfo_t)(unsafe.Pointer(MemInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetConfComputeMemSizeInfo(cnvmlDevice, cMemInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetConfComputeGpusReadyState function as declared in nvml/nvml.h +func nvmlSystemGetConfComputeGpusReadyState(IsAcceptingWork *uint32) Return { + cIsAcceptingWork, _ := (*C.uint)(unsafe.Pointer(IsAcceptingWork)), cgoAllocsUnknown + __ret := C.nvmlSystemGetConfComputeGpusReadyState(cIsAcceptingWork) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetConfComputeProtectedMemoryUsage function as declared in nvml/nvml.h +func nvmlDeviceGetConfComputeProtectedMemoryUsage(nvmlDevice nvmlDevice, Memory *Memory) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMemory, _ := (*C.nvmlMemory_t)(unsafe.Pointer(Memory)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetConfComputeProtectedMemoryUsage(cnvmlDevice, cMemory) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetConfComputeGpuCertificate function as declared in nvml/nvml.h +func nvmlDeviceGetConfComputeGpuCertificate(nvmlDevice nvmlDevice, GpuCert *ConfComputeGpuCertificate) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cGpuCert, _ := (*C.nvmlConfComputeGpuCertificate_t)(unsafe.Pointer(GpuCert)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetConfComputeGpuCertificate(cnvmlDevice, cGpuCert) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetConfComputeGpuAttestationReport function as declared in nvml/nvml.h +func nvmlDeviceGetConfComputeGpuAttestationReport(nvmlDevice nvmlDevice, GpuAtstReport *ConfComputeGpuAttestationReport) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cGpuAtstReport, _ := (*C.nvmlConfComputeGpuAttestationReport_t)(unsafe.Pointer(GpuAtstReport)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetConfComputeGpuAttestationReport(cnvmlDevice, cGpuAtstReport) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetConfComputeKeyRotationThresholdInfo function as declared in nvml/nvml.h +func nvmlSystemGetConfComputeKeyRotationThresholdInfo(PKeyRotationThrInfo *ConfComputeGetKeyRotationThresholdInfo) Return { + cPKeyRotationThrInfo, _ := (*C.nvmlConfComputeGetKeyRotationThresholdInfo_t)(unsafe.Pointer(PKeyRotationThrInfo)), cgoAllocsUnknown + __ret := C.nvmlSystemGetConfComputeKeyRotationThresholdInfo(cPKeyRotationThrInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetConfComputeUnprotectedMemSize function as declared in nvml/nvml.h +func nvmlDeviceSetConfComputeUnprotectedMemSize(nvmlDevice nvmlDevice, SizeKiB uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSizeKiB, _ := (C.ulonglong)(SizeKiB), cgoAllocsUnknown + __ret := C.nvmlDeviceSetConfComputeUnprotectedMemSize(cnvmlDevice, cSizeKiB) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemSetConfComputeGpusReadyState function as declared in nvml/nvml.h +func nvmlSystemSetConfComputeGpusReadyState(IsAcceptingWork uint32) Return { + cIsAcceptingWork, _ := (C.uint)(IsAcceptingWork), cgoAllocsUnknown + __ret := C.nvmlSystemSetConfComputeGpusReadyState(cIsAcceptingWork) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemSetConfComputeKeyRotationThresholdInfo function as declared in nvml/nvml.h +func nvmlSystemSetConfComputeKeyRotationThresholdInfo(PKeyRotationThrInfo *ConfComputeSetKeyRotationThresholdInfo) Return { + cPKeyRotationThrInfo, _ := (*C.nvmlConfComputeSetKeyRotationThresholdInfo_t)(unsafe.Pointer(PKeyRotationThrInfo)), cgoAllocsUnknown + __ret := C.nvmlSystemSetConfComputeKeyRotationThresholdInfo(cPKeyRotationThrInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetConfComputeSettings function as declared in nvml/nvml.h +func nvmlSystemGetConfComputeSettings(Settings *SystemConfComputeSettings) Return { + cSettings, _ := (*C.nvmlSystemConfComputeSettings_t)(unsafe.Pointer(Settings)), cgoAllocsUnknown + __ret := C.nvmlSystemGetConfComputeSettings(cSettings) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGspFirmwareVersion function as declared in nvml/nvml.h +func nvmlDeviceGetGspFirmwareVersion(nvmlDevice nvmlDevice, Version *byte) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVersion, _ := (*C.char)(unsafe.Pointer(Version)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGspFirmwareVersion(cnvmlDevice, cVersion) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGspFirmwareMode function as declared in nvml/nvml.h +func nvmlDeviceGetGspFirmwareMode(nvmlDevice nvmlDevice, IsEnabled *uint32, DefaultMode *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIsEnabled, _ := (*C.uint)(unsafe.Pointer(IsEnabled)), cgoAllocsUnknown + cDefaultMode, _ := (*C.uint)(unsafe.Pointer(DefaultMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGspFirmwareMode(cnvmlDevice, cIsEnabled, cDefaultMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSramEccErrorStatus function as declared in nvml/nvml.h +func nvmlDeviceGetSramEccErrorStatus(nvmlDevice nvmlDevice, Status *EccSramErrorStatus) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cStatus, _ := (*C.nvmlEccSramErrorStatus_t)(unsafe.Pointer(Status)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSramEccErrorStatus(cnvmlDevice, cStatus) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetPowerManagementLimit_v2 function as declared in nvml/nvml.h +func nvmlDeviceSetPowerManagementLimit_v2(nvmlDevice nvmlDevice, PowerValue *PowerValue_v2) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPowerValue, _ := (*C.nvmlPowerValue_v2_t)(unsafe.Pointer(PowerValue)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetPowerManagementLimit_v2(cnvmlDevice, cPowerValue) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAccountingMode function as declared in nvml/nvml.h +func nvmlDeviceGetAccountingMode(nvmlDevice nvmlDevice, Mode *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(Mode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAccountingMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAccountingStats function as declared in nvml/nvml.h +func nvmlDeviceGetAccountingStats(nvmlDevice nvmlDevice, Pid uint32, Stats *AccountingStats) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPid, _ := (C.uint)(Pid), cgoAllocsUnknown + cStats, _ := (*C.nvmlAccountingStats_t)(unsafe.Pointer(Stats)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAccountingStats(cnvmlDevice, cPid, cStats) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAccountingPids function as declared in nvml/nvml.h +func nvmlDeviceGetAccountingPids(nvmlDevice nvmlDevice, Count *uint32, Pids *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + cPids, _ := (*C.uint)(unsafe.Pointer(Pids)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAccountingPids(cnvmlDevice, cCount, cPids) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAccountingBufferSize function as declared in nvml/nvml.h +func nvmlDeviceGetAccountingBufferSize(nvmlDevice nvmlDevice, BufferSize *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cBufferSize, _ := (*C.uint)(unsafe.Pointer(BufferSize)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAccountingBufferSize(cnvmlDevice, cBufferSize) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetRetiredPages function as declared in nvml/nvml.h +func nvmlDeviceGetRetiredPages(nvmlDevice nvmlDevice, Cause PageRetirementCause, PageCount *uint32, Addresses *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCause, _ := (C.nvmlPageRetirementCause_t)(Cause), cgoAllocsUnknown + cPageCount, _ := (*C.uint)(unsafe.Pointer(PageCount)), cgoAllocsUnknown + cAddresses, _ := (*C.ulonglong)(unsafe.Pointer(Addresses)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetRetiredPages(cnvmlDevice, cCause, cPageCount, cAddresses) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetRetiredPages_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetRetiredPages_v2(nvmlDevice nvmlDevice, Cause PageRetirementCause, PageCount *uint32, Addresses *uint64, Timestamps *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCause, _ := (C.nvmlPageRetirementCause_t)(Cause), cgoAllocsUnknown + cPageCount, _ := (*C.uint)(unsafe.Pointer(PageCount)), cgoAllocsUnknown + cAddresses, _ := (*C.ulonglong)(unsafe.Pointer(Addresses)), cgoAllocsUnknown + cTimestamps, _ := (*C.ulonglong)(unsafe.Pointer(Timestamps)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetRetiredPages_v2(cnvmlDevice, cCause, cPageCount, cAddresses, cTimestamps) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetRetiredPagesPendingStatus function as declared in nvml/nvml.h +func nvmlDeviceGetRetiredPagesPendingStatus(nvmlDevice nvmlDevice, IsPending *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIsPending, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(IsPending)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetRetiredPagesPendingStatus(cnvmlDevice, cIsPending) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetRemappedRows function as declared in nvml/nvml.h +func nvmlDeviceGetRemappedRows(nvmlDevice nvmlDevice, CorrRows *uint32, UncRows *uint32, IsPending *uint32, FailureOccurred *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCorrRows, _ := (*C.uint)(unsafe.Pointer(CorrRows)), cgoAllocsUnknown + cUncRows, _ := (*C.uint)(unsafe.Pointer(UncRows)), cgoAllocsUnknown + cIsPending, _ := (*C.uint)(unsafe.Pointer(IsPending)), cgoAllocsUnknown + cFailureOccurred, _ := (*C.uint)(unsafe.Pointer(FailureOccurred)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetRemappedRows(cnvmlDevice, cCorrRows, cUncRows, cIsPending, cFailureOccurred) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetRowRemapperHistogram function as declared in nvml/nvml.h +func nvmlDeviceGetRowRemapperHistogram(nvmlDevice nvmlDevice, Values *RowRemapperHistogramValues) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cValues, _ := (*C.nvmlRowRemapperHistogramValues_t)(unsafe.Pointer(Values)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetRowRemapperHistogram(cnvmlDevice, cValues) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetArchitecture function as declared in nvml/nvml.h +func nvmlDeviceGetArchitecture(nvmlDevice nvmlDevice, Arch *DeviceArchitecture) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cArch, _ := (*C.nvmlDeviceArchitecture_t)(unsafe.Pointer(Arch)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetArchitecture(cnvmlDevice, cArch) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetClkMonStatus function as declared in nvml/nvml.h +func nvmlDeviceGetClkMonStatus(nvmlDevice nvmlDevice, Status *ClkMonStatus) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cStatus, _ := (*C.nvmlClkMonStatus_t)(unsafe.Pointer(Status)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetClkMonStatus(cnvmlDevice, cStatus) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetProcessUtilization function as declared in nvml/nvml.h +func nvmlDeviceGetProcessUtilization(nvmlDevice nvmlDevice, Utilization *ProcessUtilizationSample, ProcessSamplesCount *uint32, LastSeenTimeStamp uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cUtilization, _ := (*C.nvmlProcessUtilizationSample_t)(unsafe.Pointer(Utilization)), cgoAllocsUnknown + cProcessSamplesCount, _ := (*C.uint)(unsafe.Pointer(ProcessSamplesCount)), cgoAllocsUnknown + cLastSeenTimeStamp, _ := (C.ulonglong)(LastSeenTimeStamp), cgoAllocsUnknown + __ret := C.nvmlDeviceGetProcessUtilization(cnvmlDevice, cUtilization, cProcessSamplesCount, cLastSeenTimeStamp) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetProcessesUtilizationInfo function as declared in nvml/nvml.h +func nvmlDeviceGetProcessesUtilizationInfo(nvmlDevice nvmlDevice, ProcesesUtilInfo *ProcessesUtilizationInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProcesesUtilInfo, _ := (*C.nvmlProcessesUtilizationInfo_t)(unsafe.Pointer(ProcesesUtilInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetProcessesUtilizationInfo(cnvmlDevice, cProcesesUtilInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPlatformInfo function as declared in nvml/nvml.h +func nvmlDeviceGetPlatformInfo(nvmlDevice nvmlDevice, PlatformInfo *PlatformInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPlatformInfo, _ := (*C.nvmlPlatformInfo_t)(unsafe.Pointer(PlatformInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPlatformInfo(cnvmlDevice, cPlatformInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPdi function as declared in nvml/nvml.h +func nvmlDeviceGetPdi(nvmlDevice nvmlDevice, Pdi *Pdi) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPdi, _ := (*C.nvmlPdi_t)(unsafe.Pointer(Pdi)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPdi(cnvmlDevice, cPdi) + __v := (Return)(__ret) + return __v +} + +// nvmlUnitSetLedState function as declared in nvml/nvml.h +func nvmlUnitSetLedState(nvmlUnit nvmlUnit, Color LedColor) Return { + cnvmlUnit, _ := *(*C.nvmlUnit_t)(unsafe.Pointer(&nvmlUnit)), cgoAllocsUnknown + cColor, _ := (C.nvmlLedColor_t)(Color), cgoAllocsUnknown + __ret := C.nvmlUnitSetLedState(cnvmlUnit, cColor) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetPersistenceMode function as declared in nvml/nvml.h +func nvmlDeviceSetPersistenceMode(nvmlDevice nvmlDevice, Mode EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (C.nvmlEnableState_t)(Mode), cgoAllocsUnknown + __ret := C.nvmlDeviceSetPersistenceMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetComputeMode function as declared in nvml/nvml.h +func nvmlDeviceSetComputeMode(nvmlDevice nvmlDevice, Mode ComputeMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (C.nvmlComputeMode_t)(Mode), cgoAllocsUnknown + __ret := C.nvmlDeviceSetComputeMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetEccMode function as declared in nvml/nvml.h +func nvmlDeviceSetEccMode(nvmlDevice nvmlDevice, Ecc EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cEcc, _ := (C.nvmlEnableState_t)(Ecc), cgoAllocsUnknown + __ret := C.nvmlDeviceSetEccMode(cnvmlDevice, cEcc) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceClearEccErrorCounts function as declared in nvml/nvml.h +func nvmlDeviceClearEccErrorCounts(nvmlDevice nvmlDevice, CounterType EccCounterType) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCounterType, _ := (C.nvmlEccCounterType_t)(CounterType), cgoAllocsUnknown + __ret := C.nvmlDeviceClearEccErrorCounts(cnvmlDevice, cCounterType) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetDriverModel function as declared in nvml/nvml.h +func nvmlDeviceSetDriverModel(nvmlDevice nvmlDevice, DriverModel DriverModel, Flags uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cDriverModel, _ := (C.nvmlDriverModel_t)(DriverModel), cgoAllocsUnknown + cFlags, _ := (C.uint)(Flags), cgoAllocsUnknown + __ret := C.nvmlDeviceSetDriverModel(cnvmlDevice, cDriverModel, cFlags) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetGpuLockedClocks function as declared in nvml/nvml.h +func nvmlDeviceSetGpuLockedClocks(nvmlDevice nvmlDevice, MinGpuClockMHz uint32, MaxGpuClockMHz uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMinGpuClockMHz, _ := (C.uint)(MinGpuClockMHz), cgoAllocsUnknown + cMaxGpuClockMHz, _ := (C.uint)(MaxGpuClockMHz), cgoAllocsUnknown + __ret := C.nvmlDeviceSetGpuLockedClocks(cnvmlDevice, cMinGpuClockMHz, cMaxGpuClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceResetGpuLockedClocks function as declared in nvml/nvml.h +func nvmlDeviceResetGpuLockedClocks(nvmlDevice nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceResetGpuLockedClocks(cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetMemoryLockedClocks function as declared in nvml/nvml.h +func nvmlDeviceSetMemoryLockedClocks(nvmlDevice nvmlDevice, MinMemClockMHz uint32, MaxMemClockMHz uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMinMemClockMHz, _ := (C.uint)(MinMemClockMHz), cgoAllocsUnknown + cMaxMemClockMHz, _ := (C.uint)(MaxMemClockMHz), cgoAllocsUnknown + __ret := C.nvmlDeviceSetMemoryLockedClocks(cnvmlDevice, cMinMemClockMHz, cMaxMemClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceResetMemoryLockedClocks function as declared in nvml/nvml.h +func nvmlDeviceResetMemoryLockedClocks(nvmlDevice nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceResetMemoryLockedClocks(cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetApplicationsClocks function as declared in nvml/nvml.h +func nvmlDeviceSetApplicationsClocks(nvmlDevice nvmlDevice, MemClockMHz uint32, GraphicsClockMHz uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMemClockMHz, _ := (C.uint)(MemClockMHz), cgoAllocsUnknown + cGraphicsClockMHz, _ := (C.uint)(GraphicsClockMHz), cgoAllocsUnknown + __ret := C.nvmlDeviceSetApplicationsClocks(cnvmlDevice, cMemClockMHz, cGraphicsClockMHz) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceResetApplicationsClocks function as declared in nvml/nvml.h +func nvmlDeviceResetApplicationsClocks(nvmlDevice nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceResetApplicationsClocks(cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetAutoBoostedClocksEnabled function as declared in nvml/nvml.h +func nvmlDeviceSetAutoBoostedClocksEnabled(nvmlDevice nvmlDevice, Enabled EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cEnabled, _ := (C.nvmlEnableState_t)(Enabled), cgoAllocsUnknown + __ret := C.nvmlDeviceSetAutoBoostedClocksEnabled(cnvmlDevice, cEnabled) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetDefaultAutoBoostedClocksEnabled function as declared in nvml/nvml.h +func nvmlDeviceSetDefaultAutoBoostedClocksEnabled(nvmlDevice nvmlDevice, Enabled EnableState, Flags uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cEnabled, _ := (C.nvmlEnableState_t)(Enabled), cgoAllocsUnknown + cFlags, _ := (C.uint)(Flags), cgoAllocsUnknown + __ret := C.nvmlDeviceSetDefaultAutoBoostedClocksEnabled(cnvmlDevice, cEnabled, cFlags) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetDefaultFanSpeed_v2 function as declared in nvml/nvml.h +func nvmlDeviceSetDefaultFanSpeed_v2(nvmlDevice nvmlDevice, Fan uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFan, _ := (C.uint)(Fan), cgoAllocsUnknown + __ret := C.nvmlDeviceSetDefaultFanSpeed_v2(cnvmlDevice, cFan) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetFanControlPolicy function as declared in nvml/nvml.h +func nvmlDeviceSetFanControlPolicy(nvmlDevice nvmlDevice, Fan uint32, Policy FanControlPolicy) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFan, _ := (C.uint)(Fan), cgoAllocsUnknown + cPolicy, _ := (C.nvmlFanControlPolicy_t)(Policy), cgoAllocsUnknown + __ret := C.nvmlDeviceSetFanControlPolicy(cnvmlDevice, cFan, cPolicy) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetTemperatureThreshold function as declared in nvml/nvml.h +func nvmlDeviceSetTemperatureThreshold(nvmlDevice nvmlDevice, ThresholdType TemperatureThresholds, Temp *int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cThresholdType, _ := (C.nvmlTemperatureThresholds_t)(ThresholdType), cgoAllocsUnknown + cTemp, _ := (*C.int)(unsafe.Pointer(Temp)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetTemperatureThreshold(cnvmlDevice, cThresholdType, cTemp) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetPowerManagementLimit function as declared in nvml/nvml.h +func nvmlDeviceSetPowerManagementLimit(nvmlDevice nvmlDevice, Limit uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLimit, _ := (C.uint)(Limit), cgoAllocsUnknown + __ret := C.nvmlDeviceSetPowerManagementLimit(cnvmlDevice, cLimit) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetGpuOperationMode function as declared in nvml/nvml.h +func nvmlDeviceSetGpuOperationMode(nvmlDevice nvmlDevice, Mode GpuOperationMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (C.nvmlGpuOperationMode_t)(Mode), cgoAllocsUnknown + __ret := C.nvmlDeviceSetGpuOperationMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetAPIRestriction function as declared in nvml/nvml.h +func nvmlDeviceSetAPIRestriction(nvmlDevice nvmlDevice, ApiType RestrictedAPI, IsRestricted EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cApiType, _ := (C.nvmlRestrictedAPI_t)(ApiType), cgoAllocsUnknown + cIsRestricted, _ := (C.nvmlEnableState_t)(IsRestricted), cgoAllocsUnknown + __ret := C.nvmlDeviceSetAPIRestriction(cnvmlDevice, cApiType, cIsRestricted) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetFanSpeed_v2 function as declared in nvml/nvml.h +func nvmlDeviceSetFanSpeed_v2(nvmlDevice nvmlDevice, Fan uint32, Speed uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cFan, _ := (C.uint)(Fan), cgoAllocsUnknown + cSpeed, _ := (C.uint)(Speed), cgoAllocsUnknown + __ret := C.nvmlDeviceSetFanSpeed_v2(cnvmlDevice, cFan, cSpeed) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetGpcClkVfOffset function as declared in nvml/nvml.h +func nvmlDeviceSetGpcClkVfOffset(nvmlDevice nvmlDevice, Offset int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cOffset, _ := (C.int)(Offset), cgoAllocsUnknown + __ret := C.nvmlDeviceSetGpcClkVfOffset(cnvmlDevice, cOffset) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetMemClkVfOffset function as declared in nvml/nvml.h +func nvmlDeviceSetMemClkVfOffset(nvmlDevice nvmlDevice, Offset int32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cOffset, _ := (C.int)(Offset), cgoAllocsUnknown + __ret := C.nvmlDeviceSetMemClkVfOffset(cnvmlDevice, cOffset) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetAccountingMode function as declared in nvml/nvml.h +func nvmlDeviceSetAccountingMode(nvmlDevice nvmlDevice, Mode EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (C.nvmlEnableState_t)(Mode), cgoAllocsUnknown + __ret := C.nvmlDeviceSetAccountingMode(cnvmlDevice, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceClearAccountingPids function as declared in nvml/nvml.h +func nvmlDeviceClearAccountingPids(nvmlDevice nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceClearAccountingPids(cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkState function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkState(nvmlDevice nvmlDevice, Link uint32, IsActive *EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cIsActive, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(IsActive)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkState(cnvmlDevice, cLink, cIsActive) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkVersion function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkVersion(nvmlDevice nvmlDevice, Link uint32, Version *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cVersion, _ := (*C.uint)(unsafe.Pointer(Version)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkVersion(cnvmlDevice, cLink, cVersion) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkCapability function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkCapability(nvmlDevice nvmlDevice, Link uint32, Capability NvLinkCapability, CapResult *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cCapability, _ := (C.nvmlNvLinkCapability_t)(Capability), cgoAllocsUnknown + cCapResult, _ := (*C.uint)(unsafe.Pointer(CapResult)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkCapability(cnvmlDevice, cLink, cCapability, cCapResult) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkRemotePciInfo_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkRemotePciInfo_v2(nvmlDevice nvmlDevice, Link uint32, Pci *PciInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cPci, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(Pci)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkRemotePciInfo_v2(cnvmlDevice, cLink, cPci) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkErrorCounter function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkErrorCounter(nvmlDevice nvmlDevice, Link uint32, Counter NvLinkErrorCounter, CounterValue *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cCounter, _ := (C.nvmlNvLinkErrorCounter_t)(Counter), cgoAllocsUnknown + cCounterValue, _ := (*C.ulonglong)(unsafe.Pointer(CounterValue)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkErrorCounter(cnvmlDevice, cLink, cCounter, cCounterValue) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceResetNvLinkErrorCounters function as declared in nvml/nvml.h +func nvmlDeviceResetNvLinkErrorCounters(nvmlDevice nvmlDevice, Link uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + __ret := C.nvmlDeviceResetNvLinkErrorCounters(cnvmlDevice, cLink) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetNvLinkUtilizationControl function as declared in nvml/nvml.h +func nvmlDeviceSetNvLinkUtilizationControl(nvmlDevice nvmlDevice, Link uint32, Counter uint32, Control *NvLinkUtilizationControl, Reset uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cCounter, _ := (C.uint)(Counter), cgoAllocsUnknown + cControl, _ := (*C.nvmlNvLinkUtilizationControl_t)(unsafe.Pointer(Control)), cgoAllocsUnknown + cReset, _ := (C.uint)(Reset), cgoAllocsUnknown + __ret := C.nvmlDeviceSetNvLinkUtilizationControl(cnvmlDevice, cLink, cCounter, cControl, cReset) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkUtilizationControl function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkUtilizationControl(nvmlDevice nvmlDevice, Link uint32, Counter uint32, Control *NvLinkUtilizationControl) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cCounter, _ := (C.uint)(Counter), cgoAllocsUnknown + cControl, _ := (*C.nvmlNvLinkUtilizationControl_t)(unsafe.Pointer(Control)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkUtilizationControl(cnvmlDevice, cLink, cCounter, cControl) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkUtilizationCounter function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkUtilizationCounter(nvmlDevice nvmlDevice, Link uint32, Counter uint32, Rxcounter *uint64, Txcounter *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cCounter, _ := (C.uint)(Counter), cgoAllocsUnknown + cRxcounter, _ := (*C.ulonglong)(unsafe.Pointer(Rxcounter)), cgoAllocsUnknown + cTxcounter, _ := (*C.ulonglong)(unsafe.Pointer(Txcounter)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkUtilizationCounter(cnvmlDevice, cLink, cCounter, cRxcounter, cTxcounter) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceFreezeNvLinkUtilizationCounter function as declared in nvml/nvml.h +func nvmlDeviceFreezeNvLinkUtilizationCounter(nvmlDevice nvmlDevice, Link uint32, Counter uint32, Freeze EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cCounter, _ := (C.uint)(Counter), cgoAllocsUnknown + cFreeze, _ := (C.nvmlEnableState_t)(Freeze), cgoAllocsUnknown + __ret := C.nvmlDeviceFreezeNvLinkUtilizationCounter(cnvmlDevice, cLink, cCounter, cFreeze) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceResetNvLinkUtilizationCounter function as declared in nvml/nvml.h +func nvmlDeviceResetNvLinkUtilizationCounter(nvmlDevice nvmlDevice, Link uint32, Counter uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cCounter, _ := (C.uint)(Counter), cgoAllocsUnknown + __ret := C.nvmlDeviceResetNvLinkUtilizationCounter(cnvmlDevice, cLink, cCounter) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkRemoteDeviceType function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkRemoteDeviceType(nvmlDevice nvmlDevice, Link uint32, PNvLinkDeviceType *IntNvLinkDeviceType) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cPNvLinkDeviceType, _ := (*C.nvmlIntNvLinkDeviceType_t)(unsafe.Pointer(PNvLinkDeviceType)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkRemoteDeviceType(cnvmlDevice, cLink, cPNvLinkDeviceType) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetNvLinkDeviceLowPowerThreshold function as declared in nvml/nvml.h +func nvmlDeviceSetNvLinkDeviceLowPowerThreshold(nvmlDevice nvmlDevice, Info *NvLinkPowerThres) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlNvLinkPowerThres_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetNvLinkDeviceLowPowerThreshold(cnvmlDevice, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemSetNvlinkBwMode function as declared in nvml/nvml.h +func nvmlSystemSetNvlinkBwMode(NvlinkBwMode uint32) Return { + cNvlinkBwMode, _ := (C.uint)(NvlinkBwMode), cgoAllocsUnknown + __ret := C.nvmlSystemSetNvlinkBwMode(cNvlinkBwMode) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemGetNvlinkBwMode function as declared in nvml/nvml.h +func nvmlSystemGetNvlinkBwMode(NvlinkBwMode *uint32) Return { + cNvlinkBwMode, _ := (*C.uint)(unsafe.Pointer(NvlinkBwMode)), cgoAllocsUnknown + __ret := C.nvmlSystemGetNvlinkBwMode(cNvlinkBwMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvlinkSupportedBwModes function as declared in nvml/nvml.h +func nvmlDeviceGetNvlinkSupportedBwModes(nvmlDevice nvmlDevice, SupportedBwMode *NvlinkSupportedBwModes) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSupportedBwMode, _ := (*C.nvmlNvlinkSupportedBwModes_t)(unsafe.Pointer(SupportedBwMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvlinkSupportedBwModes(cnvmlDevice, cSupportedBwMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvlinkBwMode function as declared in nvml/nvml.h +func nvmlDeviceGetNvlinkBwMode(nvmlDevice nvmlDevice, GetBwMode *NvlinkGetBwMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cGetBwMode, _ := (*C.nvmlNvlinkGetBwMode_t)(unsafe.Pointer(GetBwMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvlinkBwMode(cnvmlDevice, cGetBwMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetNvlinkBwMode function as declared in nvml/nvml.h +func nvmlDeviceSetNvlinkBwMode(nvmlDevice nvmlDevice, SetBwMode *NvlinkSetBwMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cSetBwMode, _ := (*C.nvmlNvlinkSetBwMode_t)(unsafe.Pointer(SetBwMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetNvlinkBwMode(cnvmlDevice, cSetBwMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkInfo function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkInfo(nvmlDevice nvmlDevice, Info *NvLinkInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlNvLinkInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkInfo(cnvmlDevice, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlEventSetCreate function as declared in nvml/nvml.h +func nvmlEventSetCreate(Set *nvmlEventSet) Return { + cSet, _ := (*C.nvmlEventSet_t)(unsafe.Pointer(Set)), cgoAllocsUnknown + __ret := C.nvmlEventSetCreate(cSet) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceRegisterEvents function as declared in nvml/nvml.h +func nvmlDeviceRegisterEvents(nvmlDevice nvmlDevice, EventTypes uint64, Set nvmlEventSet) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cEventTypes, _ := (C.ulonglong)(EventTypes), cgoAllocsUnknown + cSet, _ := *(*C.nvmlEventSet_t)(unsafe.Pointer(&Set)), cgoAllocsUnknown + __ret := C.nvmlDeviceRegisterEvents(cnvmlDevice, cEventTypes, cSet) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSupportedEventTypes function as declared in nvml/nvml.h +func nvmlDeviceGetSupportedEventTypes(nvmlDevice nvmlDevice, EventTypes *uint64) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cEventTypes, _ := (*C.ulonglong)(unsafe.Pointer(EventTypes)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSupportedEventTypes(cnvmlDevice, cEventTypes) + __v := (Return)(__ret) + return __v +} + +// nvmlEventSetWait_v2 function as declared in nvml/nvml.h +func nvmlEventSetWait_v2(Set nvmlEventSet, Data *nvmlEventData, Timeoutms uint32) Return { + cSet, _ := *(*C.nvmlEventSet_t)(unsafe.Pointer(&Set)), cgoAllocsUnknown + cData, _ := (*C.nvmlEventData_t)(unsafe.Pointer(Data)), cgoAllocsUnknown + cTimeoutms, _ := (C.uint)(Timeoutms), cgoAllocsUnknown + __ret := C.nvmlEventSetWait_v2(cSet, cData, cTimeoutms) + __v := (Return)(__ret) + return __v +} + +// nvmlEventSetFree function as declared in nvml/nvml.h +func nvmlEventSetFree(Set nvmlEventSet) Return { + cSet, _ := *(*C.nvmlEventSet_t)(unsafe.Pointer(&Set)), cgoAllocsUnknown + __ret := C.nvmlEventSetFree(cSet) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemEventSetCreate function as declared in nvml/nvml.h +func nvmlSystemEventSetCreate(Request *SystemEventSetCreateRequest) Return { + cRequest, _ := (*C.nvmlSystemEventSetCreateRequest_t)(unsafe.Pointer(Request)), cgoAllocsUnknown + __ret := C.nvmlSystemEventSetCreate(cRequest) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemEventSetFree function as declared in nvml/nvml.h +func nvmlSystemEventSetFree(Request *SystemEventSetFreeRequest) Return { + cRequest, _ := (*C.nvmlSystemEventSetFreeRequest_t)(unsafe.Pointer(Request)), cgoAllocsUnknown + __ret := C.nvmlSystemEventSetFree(cRequest) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemRegisterEvents function as declared in nvml/nvml.h +func nvmlSystemRegisterEvents(Request *SystemRegisterEventRequest) Return { + cRequest, _ := (*C.nvmlSystemRegisterEventRequest_t)(unsafe.Pointer(Request)), cgoAllocsUnknown + __ret := C.nvmlSystemRegisterEvents(cRequest) + __v := (Return)(__ret) + return __v +} + +// nvmlSystemEventSetWait function as declared in nvml/nvml.h +func nvmlSystemEventSetWait(Request *SystemEventSetWaitRequest) Return { + cRequest, _ := (*C.nvmlSystemEventSetWaitRequest_t)(unsafe.Pointer(Request)), cgoAllocsUnknown + __ret := C.nvmlSystemEventSetWait(cRequest) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceModifyDrainState function as declared in nvml/nvml.h +func nvmlDeviceModifyDrainState(PciInfo *PciInfo, NewState EnableState) Return { + cPciInfo, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(PciInfo)), cgoAllocsUnknown + cNewState, _ := (C.nvmlEnableState_t)(NewState), cgoAllocsUnknown + __ret := C.nvmlDeviceModifyDrainState(cPciInfo, cNewState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceQueryDrainState function as declared in nvml/nvml.h +func nvmlDeviceQueryDrainState(PciInfo *PciInfo, CurrentState *EnableState) Return { + cPciInfo, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(PciInfo)), cgoAllocsUnknown + cCurrentState, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(CurrentState)), cgoAllocsUnknown + __ret := C.nvmlDeviceQueryDrainState(cPciInfo, cCurrentState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceRemoveGpu_v2 function as declared in nvml/nvml.h +func nvmlDeviceRemoveGpu_v2(PciInfo *PciInfo, GpuState DetachGpuState, LinkState PcieLinkState) Return { + cPciInfo, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(PciInfo)), cgoAllocsUnknown + cGpuState, _ := (C.nvmlDetachGpuState_t)(GpuState), cgoAllocsUnknown + cLinkState, _ := (C.nvmlPcieLinkState_t)(LinkState), cgoAllocsUnknown + __ret := C.nvmlDeviceRemoveGpu_v2(cPciInfo, cGpuState, cLinkState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceDiscoverGpus function as declared in nvml/nvml.h +func nvmlDeviceDiscoverGpus(PciInfo *PciInfo) Return { + cPciInfo, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(PciInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceDiscoverGpus(cPciInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetFieldValues function as declared in nvml/nvml.h +func nvmlDeviceGetFieldValues(nvmlDevice nvmlDevice, ValuesCount int32, Values *FieldValue) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cValuesCount, _ := (C.int)(ValuesCount), cgoAllocsUnknown + cValues, _ := (*C.nvmlFieldValue_t)(unsafe.Pointer(Values)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetFieldValues(cnvmlDevice, cValuesCount, cValues) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceClearFieldValues function as declared in nvml/nvml.h +func nvmlDeviceClearFieldValues(nvmlDevice nvmlDevice, ValuesCount int32, Values *FieldValue) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cValuesCount, _ := (C.int)(ValuesCount), cgoAllocsUnknown + cValues, _ := (*C.nvmlFieldValue_t)(unsafe.Pointer(Values)), cgoAllocsUnknown + __ret := C.nvmlDeviceClearFieldValues(cnvmlDevice, cValuesCount, cValues) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVirtualizationMode function as declared in nvml/nvml.h +func nvmlDeviceGetVirtualizationMode(nvmlDevice nvmlDevice, PVirtualMode *GpuVirtualizationMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPVirtualMode, _ := (*C.nvmlGpuVirtualizationMode_t)(unsafe.Pointer(PVirtualMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVirtualizationMode(cnvmlDevice, cPVirtualMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHostVgpuMode function as declared in nvml/nvml.h +func nvmlDeviceGetHostVgpuMode(nvmlDevice nvmlDevice, PHostVgpuMode *HostVgpuMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPHostVgpuMode, _ := (*C.nvmlHostVgpuMode_t)(unsafe.Pointer(PHostVgpuMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHostVgpuMode(cnvmlDevice, cPHostVgpuMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetVirtualizationMode function as declared in nvml/nvml.h +func nvmlDeviceSetVirtualizationMode(nvmlDevice nvmlDevice, VirtualMode GpuVirtualizationMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVirtualMode, _ := (C.nvmlGpuVirtualizationMode_t)(VirtualMode), cgoAllocsUnknown + __ret := C.nvmlDeviceSetVirtualizationMode(cnvmlDevice, cVirtualMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuHeterogeneousMode function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuHeterogeneousMode(nvmlDevice nvmlDevice, PHeterogeneousMode *VgpuHeterogeneousMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPHeterogeneousMode, _ := (*C.nvmlVgpuHeterogeneousMode_t)(unsafe.Pointer(PHeterogeneousMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuHeterogeneousMode(cnvmlDevice, cPHeterogeneousMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetVgpuHeterogeneousMode function as declared in nvml/nvml.h +func nvmlDeviceSetVgpuHeterogeneousMode(nvmlDevice nvmlDevice, PHeterogeneousMode *VgpuHeterogeneousMode) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPHeterogeneousMode, _ := (*C.nvmlVgpuHeterogeneousMode_t)(unsafe.Pointer(PHeterogeneousMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetVgpuHeterogeneousMode(cnvmlDevice, cPHeterogeneousMode) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetPlacementId function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetPlacementId(nvmlVgpuInstance nvmlVgpuInstance, PPlacement *VgpuPlacementId) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cPPlacement, _ := (*C.nvmlVgpuPlacementId_t)(unsafe.Pointer(PPlacement)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetPlacementId(cnvmlVgpuInstance, cPPlacement) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuTypeSupportedPlacements function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuTypeSupportedPlacements(nvmlDevice nvmlDevice, nvmlVgpuTypeId nvmlVgpuTypeId, PPlacementList *VgpuPlacementList) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cPPlacementList, _ := (*C.nvmlVgpuPlacementList_t)(unsafe.Pointer(PPlacementList)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuTypeSupportedPlacements(cnvmlDevice, cnvmlVgpuTypeId, cPPlacementList) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuTypeCreatablePlacements function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuTypeCreatablePlacements(nvmlDevice nvmlDevice, nvmlVgpuTypeId nvmlVgpuTypeId, PPlacementList *VgpuPlacementList) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cPPlacementList, _ := (*C.nvmlVgpuPlacementList_t)(unsafe.Pointer(PPlacementList)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuTypeCreatablePlacements(cnvmlDevice, cnvmlVgpuTypeId, cPPlacementList) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetGspHeapSize function as declared in nvml/nvml.h +func nvmlVgpuTypeGetGspHeapSize(nvmlVgpuTypeId nvmlVgpuTypeId, GspHeapSize *uint64) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cGspHeapSize, _ := (*C.ulonglong)(unsafe.Pointer(GspHeapSize)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetGspHeapSize(cnvmlVgpuTypeId, cGspHeapSize) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetFbReservation function as declared in nvml/nvml.h +func nvmlVgpuTypeGetFbReservation(nvmlVgpuTypeId nvmlVgpuTypeId, FbReservation *uint64) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cFbReservation, _ := (*C.ulonglong)(unsafe.Pointer(FbReservation)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetFbReservation(cnvmlVgpuTypeId, cFbReservation) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetRuntimeStateSize function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetRuntimeStateSize(nvmlVgpuInstance nvmlVgpuInstance, PState *VgpuRuntimeState) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cPState, _ := (*C.nvmlVgpuRuntimeState_t)(unsafe.Pointer(PState)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetRuntimeStateSize(cnvmlVgpuInstance, cPState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetVgpuCapabilities function as declared in nvml/nvml.h +func nvmlDeviceSetVgpuCapabilities(nvmlDevice nvmlDevice, Capability DeviceVgpuCapability, State EnableState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCapability, _ := (C.nvmlDeviceVgpuCapability_t)(Capability), cgoAllocsUnknown + cState, _ := (C.nvmlEnableState_t)(State), cgoAllocsUnknown + __ret := C.nvmlDeviceSetVgpuCapabilities(cnvmlDevice, cCapability, cState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGridLicensableFeatures_v4 function as declared in nvml/nvml.h +func nvmlDeviceGetGridLicensableFeatures_v4(nvmlDevice nvmlDevice, PGridLicensableFeatures *GridLicensableFeatures) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPGridLicensableFeatures, _ := (*C.nvmlGridLicensableFeatures_t)(unsafe.Pointer(PGridLicensableFeatures)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGridLicensableFeatures_v4(cnvmlDevice, cPGridLicensableFeatures) + __v := (Return)(__ret) + return __v +} + +// nvmlGetVgpuDriverCapabilities function as declared in nvml/nvml.h +func nvmlGetVgpuDriverCapabilities(Capability VgpuDriverCapability, CapResult *uint32) Return { + cCapability, _ := (C.nvmlVgpuDriverCapability_t)(Capability), cgoAllocsUnknown + cCapResult, _ := (*C.uint)(unsafe.Pointer(CapResult)), cgoAllocsUnknown + __ret := C.nvmlGetVgpuDriverCapabilities(cCapability, cCapResult) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuCapabilities function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuCapabilities(nvmlDevice nvmlDevice, Capability DeviceVgpuCapability, CapResult *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCapability, _ := (C.nvmlDeviceVgpuCapability_t)(Capability), cgoAllocsUnknown + cCapResult, _ := (*C.uint)(unsafe.Pointer(CapResult)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuCapabilities(cnvmlDevice, cCapability, cCapResult) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSupportedVgpus function as declared in nvml/nvml.h +func nvmlDeviceGetSupportedVgpus(nvmlDevice nvmlDevice, VgpuCount *uint32, VgpuTypeIds *nvmlVgpuTypeId) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVgpuCount, _ := (*C.uint)(unsafe.Pointer(VgpuCount)), cgoAllocsUnknown + cVgpuTypeIds, _ := (*C.nvmlVgpuTypeId_t)(unsafe.Pointer(VgpuTypeIds)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSupportedVgpus(cnvmlDevice, cVgpuCount, cVgpuTypeIds) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCreatableVgpus function as declared in nvml/nvml.h +func nvmlDeviceGetCreatableVgpus(nvmlDevice nvmlDevice, VgpuCount *uint32, VgpuTypeIds *nvmlVgpuTypeId) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVgpuCount, _ := (*C.uint)(unsafe.Pointer(VgpuCount)), cgoAllocsUnknown + cVgpuTypeIds, _ := (*C.nvmlVgpuTypeId_t)(unsafe.Pointer(VgpuTypeIds)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCreatableVgpus(cnvmlDevice, cVgpuCount, cVgpuTypeIds) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetClass function as declared in nvml/nvml.h +func nvmlVgpuTypeGetClass(nvmlVgpuTypeId nvmlVgpuTypeId, VgpuTypeClass *byte, Size *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cVgpuTypeClass, _ := (*C.char)(unsafe.Pointer(VgpuTypeClass)), cgoAllocsUnknown + cSize, _ := (*C.uint)(unsafe.Pointer(Size)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetClass(cnvmlVgpuTypeId, cVgpuTypeClass, cSize) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetName function as declared in nvml/nvml.h +func nvmlVgpuTypeGetName(nvmlVgpuTypeId nvmlVgpuTypeId, VgpuTypeName *byte, Size *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cVgpuTypeName, _ := (*C.char)(unsafe.Pointer(VgpuTypeName)), cgoAllocsUnknown + cSize, _ := (*C.uint)(unsafe.Pointer(Size)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetName(cnvmlVgpuTypeId, cVgpuTypeName, cSize) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetGpuInstanceProfileId function as declared in nvml/nvml.h +func nvmlVgpuTypeGetGpuInstanceProfileId(nvmlVgpuTypeId nvmlVgpuTypeId, GpuInstanceProfileId *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cGpuInstanceProfileId, _ := (*C.uint)(unsafe.Pointer(GpuInstanceProfileId)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetGpuInstanceProfileId(cnvmlVgpuTypeId, cGpuInstanceProfileId) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetDeviceID function as declared in nvml/nvml.h +func nvmlVgpuTypeGetDeviceID(nvmlVgpuTypeId nvmlVgpuTypeId, DeviceID *uint64, SubsystemID *uint64) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cDeviceID, _ := (*C.ulonglong)(unsafe.Pointer(DeviceID)), cgoAllocsUnknown + cSubsystemID, _ := (*C.ulonglong)(unsafe.Pointer(SubsystemID)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetDeviceID(cnvmlVgpuTypeId, cDeviceID, cSubsystemID) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetFramebufferSize function as declared in nvml/nvml.h +func nvmlVgpuTypeGetFramebufferSize(nvmlVgpuTypeId nvmlVgpuTypeId, FbSize *uint64) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cFbSize, _ := (*C.ulonglong)(unsafe.Pointer(FbSize)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetFramebufferSize(cnvmlVgpuTypeId, cFbSize) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetNumDisplayHeads function as declared in nvml/nvml.h +func nvmlVgpuTypeGetNumDisplayHeads(nvmlVgpuTypeId nvmlVgpuTypeId, NumDisplayHeads *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cNumDisplayHeads, _ := (*C.uint)(unsafe.Pointer(NumDisplayHeads)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetNumDisplayHeads(cnvmlVgpuTypeId, cNumDisplayHeads) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetResolution function as declared in nvml/nvml.h +func nvmlVgpuTypeGetResolution(nvmlVgpuTypeId nvmlVgpuTypeId, DisplayIndex uint32, Xdim *uint32, Ydim *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cDisplayIndex, _ := (C.uint)(DisplayIndex), cgoAllocsUnknown + cXdim, _ := (*C.uint)(unsafe.Pointer(Xdim)), cgoAllocsUnknown + cYdim, _ := (*C.uint)(unsafe.Pointer(Ydim)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetResolution(cnvmlVgpuTypeId, cDisplayIndex, cXdim, cYdim) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetLicense function as declared in nvml/nvml.h +func nvmlVgpuTypeGetLicense(nvmlVgpuTypeId nvmlVgpuTypeId, VgpuTypeLicenseString *byte, Size uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cVgpuTypeLicenseString, _ := (*C.char)(unsafe.Pointer(VgpuTypeLicenseString)), cgoAllocsUnknown + cSize, _ := (C.uint)(Size), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetLicense(cnvmlVgpuTypeId, cVgpuTypeLicenseString, cSize) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetFrameRateLimit function as declared in nvml/nvml.h +func nvmlVgpuTypeGetFrameRateLimit(nvmlVgpuTypeId nvmlVgpuTypeId, FrameRateLimit *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cFrameRateLimit, _ := (*C.uint)(unsafe.Pointer(FrameRateLimit)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetFrameRateLimit(cnvmlVgpuTypeId, cFrameRateLimit) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetMaxInstances function as declared in nvml/nvml.h +func nvmlVgpuTypeGetMaxInstances(nvmlDevice nvmlDevice, nvmlVgpuTypeId nvmlVgpuTypeId, VgpuInstanceCount *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cVgpuInstanceCount, _ := (*C.uint)(unsafe.Pointer(VgpuInstanceCount)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetMaxInstances(cnvmlDevice, cnvmlVgpuTypeId, cVgpuInstanceCount) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetMaxInstancesPerVm function as declared in nvml/nvml.h +func nvmlVgpuTypeGetMaxInstancesPerVm(nvmlVgpuTypeId nvmlVgpuTypeId, VgpuInstanceCountPerVm *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cVgpuInstanceCountPerVm, _ := (*C.uint)(unsafe.Pointer(VgpuInstanceCountPerVm)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetMaxInstancesPerVm(cnvmlVgpuTypeId, cVgpuInstanceCountPerVm) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetBAR1Info function as declared in nvml/nvml.h +func nvmlVgpuTypeGetBAR1Info(nvmlVgpuTypeId nvmlVgpuTypeId, Bar1Info *VgpuTypeBar1Info) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cBar1Info, _ := (*C.nvmlVgpuTypeBar1Info_t)(unsafe.Pointer(Bar1Info)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetBAR1Info(cnvmlVgpuTypeId, cBar1Info) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetActiveVgpus function as declared in nvml/nvml.h +func nvmlDeviceGetActiveVgpus(nvmlDevice nvmlDevice, VgpuCount *uint32, VgpuInstances *nvmlVgpuInstance) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVgpuCount, _ := (*C.uint)(unsafe.Pointer(VgpuCount)), cgoAllocsUnknown + cVgpuInstances, _ := (*C.nvmlVgpuInstance_t)(unsafe.Pointer(VgpuInstances)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetActiveVgpus(cnvmlDevice, cVgpuCount, cVgpuInstances) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetVmID function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetVmID(nvmlVgpuInstance nvmlVgpuInstance, VmId *byte, Size uint32, VmIdType *VgpuVmIdType) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cVmId, _ := (*C.char)(unsafe.Pointer(VmId)), cgoAllocsUnknown + cSize, _ := (C.uint)(Size), cgoAllocsUnknown + cVmIdType, _ := (*C.nvmlVgpuVmIdType_t)(unsafe.Pointer(VmIdType)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetVmID(cnvmlVgpuInstance, cVmId, cSize, cVmIdType) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetUUID function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetUUID(nvmlVgpuInstance nvmlVgpuInstance, Uuid *byte, Size uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cUuid, _ := (*C.char)(unsafe.Pointer(Uuid)), cgoAllocsUnknown + cSize, _ := (C.uint)(Size), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetUUID(cnvmlVgpuInstance, cUuid, cSize) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetVmDriverVersion function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetVmDriverVersion(nvmlVgpuInstance nvmlVgpuInstance, Version *byte, Length uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cVersion, _ := (*C.char)(unsafe.Pointer(Version)), cgoAllocsUnknown + cLength, _ := (C.uint)(Length), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetVmDriverVersion(cnvmlVgpuInstance, cVersion, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetFbUsage function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetFbUsage(nvmlVgpuInstance nvmlVgpuInstance, FbUsage *uint64) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cFbUsage, _ := (*C.ulonglong)(unsafe.Pointer(FbUsage)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetFbUsage(cnvmlVgpuInstance, cFbUsage) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetLicenseStatus function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetLicenseStatus(nvmlVgpuInstance nvmlVgpuInstance, Licensed *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cLicensed, _ := (*C.uint)(unsafe.Pointer(Licensed)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetLicenseStatus(cnvmlVgpuInstance, cLicensed) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetType function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetType(nvmlVgpuInstance nvmlVgpuInstance, nvmlVgpuTypeId *nvmlVgpuTypeId) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cnvmlVgpuTypeId, _ := (*C.nvmlVgpuTypeId_t)(unsafe.Pointer(nvmlVgpuTypeId)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetType(cnvmlVgpuInstance, cnvmlVgpuTypeId) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetFrameRateLimit function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetFrameRateLimit(nvmlVgpuInstance nvmlVgpuInstance, FrameRateLimit *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cFrameRateLimit, _ := (*C.uint)(unsafe.Pointer(FrameRateLimit)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetFrameRateLimit(cnvmlVgpuInstance, cFrameRateLimit) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetEccMode function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetEccMode(nvmlVgpuInstance nvmlVgpuInstance, EccMode *EnableState) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cEccMode, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(EccMode)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetEccMode(cnvmlVgpuInstance, cEccMode) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetEncoderCapacity function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetEncoderCapacity(nvmlVgpuInstance nvmlVgpuInstance, EncoderCapacity *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cEncoderCapacity, _ := (*C.uint)(unsafe.Pointer(EncoderCapacity)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetEncoderCapacity(cnvmlVgpuInstance, cEncoderCapacity) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceSetEncoderCapacity function as declared in nvml/nvml.h +func nvmlVgpuInstanceSetEncoderCapacity(nvmlVgpuInstance nvmlVgpuInstance, EncoderCapacity uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cEncoderCapacity, _ := (C.uint)(EncoderCapacity), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceSetEncoderCapacity(cnvmlVgpuInstance, cEncoderCapacity) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetEncoderStats function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetEncoderStats(nvmlVgpuInstance nvmlVgpuInstance, SessionCount *uint32, AverageFps *uint32, AverageLatency *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cSessionCount, _ := (*C.uint)(unsafe.Pointer(SessionCount)), cgoAllocsUnknown + cAverageFps, _ := (*C.uint)(unsafe.Pointer(AverageFps)), cgoAllocsUnknown + cAverageLatency, _ := (*C.uint)(unsafe.Pointer(AverageLatency)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetEncoderStats(cnvmlVgpuInstance, cSessionCount, cAverageFps, cAverageLatency) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetEncoderSessions function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetEncoderSessions(nvmlVgpuInstance nvmlVgpuInstance, SessionCount *uint32, SessionInfo *EncoderSessionInfo) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cSessionCount, _ := (*C.uint)(unsafe.Pointer(SessionCount)), cgoAllocsUnknown + cSessionInfo, _ := (*C.nvmlEncoderSessionInfo_t)(unsafe.Pointer(SessionInfo)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetEncoderSessions(cnvmlVgpuInstance, cSessionCount, cSessionInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetFBCStats function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetFBCStats(nvmlVgpuInstance nvmlVgpuInstance, FbcStats *FBCStats) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cFbcStats, _ := (*C.nvmlFBCStats_t)(unsafe.Pointer(FbcStats)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetFBCStats(cnvmlVgpuInstance, cFbcStats) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetFBCSessions function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetFBCSessions(nvmlVgpuInstance nvmlVgpuInstance, SessionCount *uint32, SessionInfo *FBCSessionInfo) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cSessionCount, _ := (*C.uint)(unsafe.Pointer(SessionCount)), cgoAllocsUnknown + cSessionInfo, _ := (*C.nvmlFBCSessionInfo_t)(unsafe.Pointer(SessionInfo)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetFBCSessions(cnvmlVgpuInstance, cSessionCount, cSessionInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetGpuInstanceId function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetGpuInstanceId(nvmlVgpuInstance nvmlVgpuInstance, GpuInstanceId *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cGpuInstanceId, _ := (*C.uint)(unsafe.Pointer(GpuInstanceId)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetGpuInstanceId(cnvmlVgpuInstance, cGpuInstanceId) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetGpuPciId function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetGpuPciId(nvmlVgpuInstance nvmlVgpuInstance, VgpuPciId *byte, Length *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cVgpuPciId, _ := (*C.char)(unsafe.Pointer(VgpuPciId)), cgoAllocsUnknown + cLength, _ := (*C.uint)(unsafe.Pointer(Length)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetGpuPciId(cnvmlVgpuInstance, cVgpuPciId, cLength) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetCapabilities function as declared in nvml/nvml.h +func nvmlVgpuTypeGetCapabilities(nvmlVgpuTypeId nvmlVgpuTypeId, Capability VgpuCapability, CapResult *uint32) Return { + cnvmlVgpuTypeId, _ := (C.nvmlVgpuTypeId_t)(nvmlVgpuTypeId), cgoAllocsUnknown + cCapability, _ := (C.nvmlVgpuCapability_t)(Capability), cgoAllocsUnknown + cCapResult, _ := (*C.uint)(unsafe.Pointer(CapResult)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetCapabilities(cnvmlVgpuTypeId, cCapability, cCapResult) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetMdevUUID function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetMdevUUID(nvmlVgpuInstance nvmlVgpuInstance, MdevUuid *byte, Size uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cMdevUuid, _ := (*C.char)(unsafe.Pointer(MdevUuid)), cgoAllocsUnknown + cSize, _ := (C.uint)(Size), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetMdevUUID(cnvmlVgpuInstance, cMdevUuid, cSize) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetCreatableVgpus function as declared in nvml/nvml.h +func nvmlGpuInstanceGetCreatableVgpus(nvmlGpuInstance nvmlGpuInstance, PVgpus *VgpuTypeIdInfo) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPVgpus, _ := (*C.nvmlVgpuTypeIdInfo_t)(unsafe.Pointer(PVgpus)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetCreatableVgpus(cnvmlGpuInstance, cPVgpus) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuTypeGetMaxInstancesPerGpuInstance function as declared in nvml/nvml.h +func nvmlVgpuTypeGetMaxInstancesPerGpuInstance(PMaxInstance *VgpuTypeMaxInstance) Return { + cPMaxInstance, _ := (*C.nvmlVgpuTypeMaxInstance_t)(unsafe.Pointer(PMaxInstance)), cgoAllocsUnknown + __ret := C.nvmlVgpuTypeGetMaxInstancesPerGpuInstance(cPMaxInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetActiveVgpus function as declared in nvml/nvml.h +func nvmlGpuInstanceGetActiveVgpus(nvmlGpuInstance nvmlGpuInstance, PVgpuInstanceInfo *ActiveVgpuInstanceInfo) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPVgpuInstanceInfo, _ := (*C.nvmlActiveVgpuInstanceInfo_t)(unsafe.Pointer(PVgpuInstanceInfo)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetActiveVgpus(cnvmlGpuInstance, cPVgpuInstanceInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceSetVgpuSchedulerState function as declared in nvml/nvml.h +func nvmlGpuInstanceSetVgpuSchedulerState(nvmlGpuInstance nvmlGpuInstance, PScheduler *VgpuSchedulerState) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPScheduler, _ := (*C.nvmlVgpuSchedulerState_t)(unsafe.Pointer(PScheduler)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceSetVgpuSchedulerState(cnvmlGpuInstance, cPScheduler) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetVgpuSchedulerState function as declared in nvml/nvml.h +func nvmlGpuInstanceGetVgpuSchedulerState(nvmlGpuInstance nvmlGpuInstance, PSchedulerStateInfo *VgpuSchedulerStateInfo) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPSchedulerStateInfo, _ := (*C.nvmlVgpuSchedulerStateInfo_t)(unsafe.Pointer(PSchedulerStateInfo)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetVgpuSchedulerState(cnvmlGpuInstance, cPSchedulerStateInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetVgpuSchedulerLog function as declared in nvml/nvml.h +func nvmlGpuInstanceGetVgpuSchedulerLog(nvmlGpuInstance nvmlGpuInstance, PSchedulerLogInfo *VgpuSchedulerLogInfo) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPSchedulerLogInfo, _ := (*C.nvmlVgpuSchedulerLogInfo_t)(unsafe.Pointer(PSchedulerLogInfo)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetVgpuSchedulerLog(cnvmlGpuInstance, cPSchedulerLogInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetVgpuTypeCreatablePlacements function as declared in nvml/nvml.h +func nvmlGpuInstanceGetVgpuTypeCreatablePlacements(nvmlGpuInstance nvmlGpuInstance, PCreatablePlacementInfo *VgpuCreatablePlacementInfo) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPCreatablePlacementInfo, _ := (*C.nvmlVgpuCreatablePlacementInfo_t)(unsafe.Pointer(PCreatablePlacementInfo)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetVgpuTypeCreatablePlacements(cnvmlGpuInstance, cPCreatablePlacementInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetVgpuHeterogeneousMode function as declared in nvml/nvml.h +func nvmlGpuInstanceGetVgpuHeterogeneousMode(nvmlGpuInstance nvmlGpuInstance, PHeterogeneousMode *VgpuHeterogeneousMode) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPHeterogeneousMode, _ := (*C.nvmlVgpuHeterogeneousMode_t)(unsafe.Pointer(PHeterogeneousMode)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetVgpuHeterogeneousMode(cnvmlGpuInstance, cPHeterogeneousMode) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceSetVgpuHeterogeneousMode function as declared in nvml/nvml.h +func nvmlGpuInstanceSetVgpuHeterogeneousMode(nvmlGpuInstance nvmlGpuInstance, PHeterogeneousMode *VgpuHeterogeneousMode) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cPHeterogeneousMode, _ := (*C.nvmlVgpuHeterogeneousMode_t)(unsafe.Pointer(PHeterogeneousMode)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceSetVgpuHeterogeneousMode(cnvmlGpuInstance, cPHeterogeneousMode) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetMetadata function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetMetadata(nvmlVgpuInstance nvmlVgpuInstance, nvmlVgpuMetadata *nvmlVgpuMetadata, BufferSize *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cnvmlVgpuMetadata, _ := (*C.nvmlVgpuMetadata_t)(unsafe.Pointer(nvmlVgpuMetadata)), cgoAllocsUnknown + cBufferSize, _ := (*C.uint)(unsafe.Pointer(BufferSize)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetMetadata(cnvmlVgpuInstance, cnvmlVgpuMetadata, cBufferSize) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuMetadata function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuMetadata(nvmlDevice nvmlDevice, PgpuMetadata *nvmlVgpuPgpuMetadata, BufferSize *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPgpuMetadata, _ := (*C.nvmlVgpuPgpuMetadata_t)(unsafe.Pointer(PgpuMetadata)), cgoAllocsUnknown + cBufferSize, _ := (*C.uint)(unsafe.Pointer(BufferSize)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuMetadata(cnvmlDevice, cPgpuMetadata, cBufferSize) + __v := (Return)(__ret) + return __v +} + +// nvmlGetVgpuCompatibility function as declared in nvml/nvml.h +func nvmlGetVgpuCompatibility(nvmlVgpuMetadata *nvmlVgpuMetadata, PgpuMetadata *nvmlVgpuPgpuMetadata, CompatibilityInfo *VgpuPgpuCompatibility) Return { + cnvmlVgpuMetadata, _ := (*C.nvmlVgpuMetadata_t)(unsafe.Pointer(nvmlVgpuMetadata)), cgoAllocsUnknown + cPgpuMetadata, _ := (*C.nvmlVgpuPgpuMetadata_t)(unsafe.Pointer(PgpuMetadata)), cgoAllocsUnknown + cCompatibilityInfo, _ := (*C.nvmlVgpuPgpuCompatibility_t)(unsafe.Pointer(CompatibilityInfo)), cgoAllocsUnknown + __ret := C.nvmlGetVgpuCompatibility(cnvmlVgpuMetadata, cPgpuMetadata, cCompatibilityInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPgpuMetadataString function as declared in nvml/nvml.h +func nvmlDeviceGetPgpuMetadataString(nvmlDevice nvmlDevice, PgpuMetadata *byte, BufferSize *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPgpuMetadata, _ := (*C.char)(unsafe.Pointer(PgpuMetadata)), cgoAllocsUnknown + cBufferSize, _ := (*C.uint)(unsafe.Pointer(BufferSize)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPgpuMetadataString(cnvmlDevice, cPgpuMetadata, cBufferSize) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuSchedulerLog function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuSchedulerLog(nvmlDevice nvmlDevice, PSchedulerLog *VgpuSchedulerLog) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPSchedulerLog, _ := (*C.nvmlVgpuSchedulerLog_t)(unsafe.Pointer(PSchedulerLog)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuSchedulerLog(cnvmlDevice, cPSchedulerLog) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuSchedulerState function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuSchedulerState(nvmlDevice nvmlDevice, PSchedulerState *VgpuSchedulerGetState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPSchedulerState, _ := (*C.nvmlVgpuSchedulerGetState_t)(unsafe.Pointer(PSchedulerState)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuSchedulerState(cnvmlDevice, cPSchedulerState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuSchedulerCapabilities function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuSchedulerCapabilities(nvmlDevice nvmlDevice, PCapabilities *VgpuSchedulerCapabilities) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPCapabilities, _ := (*C.nvmlVgpuSchedulerCapabilities_t)(unsafe.Pointer(PCapabilities)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuSchedulerCapabilities(cnvmlDevice, cPCapabilities) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetVgpuSchedulerState function as declared in nvml/nvml.h +func nvmlDeviceSetVgpuSchedulerState(nvmlDevice nvmlDevice, PSchedulerState *VgpuSchedulerSetState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPSchedulerState, _ := (*C.nvmlVgpuSchedulerSetState_t)(unsafe.Pointer(PSchedulerState)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetVgpuSchedulerState(cnvmlDevice, cPSchedulerState) + __v := (Return)(__ret) + return __v +} + +// nvmlGetVgpuVersion function as declared in nvml/nvml.h +func nvmlGetVgpuVersion(Supported *VgpuVersion, Current *VgpuVersion) Return { + cSupported, _ := (*C.nvmlVgpuVersion_t)(unsafe.Pointer(Supported)), cgoAllocsUnknown + cCurrent, _ := (*C.nvmlVgpuVersion_t)(unsafe.Pointer(Current)), cgoAllocsUnknown + __ret := C.nvmlGetVgpuVersion(cSupported, cCurrent) + __v := (Return)(__ret) + return __v +} + +// nvmlSetVgpuVersion function as declared in nvml/nvml.h +func nvmlSetVgpuVersion(VgpuVersion *VgpuVersion) Return { + cVgpuVersion, _ := (*C.nvmlVgpuVersion_t)(unsafe.Pointer(VgpuVersion)), cgoAllocsUnknown + __ret := C.nvmlSetVgpuVersion(cVgpuVersion) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuUtilization function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuUtilization(nvmlDevice nvmlDevice, LastSeenTimeStamp uint64, SampleValType *ValueType, VgpuInstanceSamplesCount *uint32, UtilizationSamples *VgpuInstanceUtilizationSample) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLastSeenTimeStamp, _ := (C.ulonglong)(LastSeenTimeStamp), cgoAllocsUnknown + cSampleValType, _ := (*C.nvmlValueType_t)(unsafe.Pointer(SampleValType)), cgoAllocsUnknown + cVgpuInstanceSamplesCount, _ := (*C.uint)(unsafe.Pointer(VgpuInstanceSamplesCount)), cgoAllocsUnknown + cUtilizationSamples, _ := (*C.nvmlVgpuInstanceUtilizationSample_t)(unsafe.Pointer(UtilizationSamples)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuUtilization(cnvmlDevice, cLastSeenTimeStamp, cSampleValType, cVgpuInstanceSamplesCount, cUtilizationSamples) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuInstancesUtilizationInfo function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuInstancesUtilizationInfo(nvmlDevice nvmlDevice, VgpuUtilInfo *VgpuInstancesUtilizationInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVgpuUtilInfo, _ := (*C.nvmlVgpuInstancesUtilizationInfo_t)(unsafe.Pointer(VgpuUtilInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuInstancesUtilizationInfo(cnvmlDevice, cVgpuUtilInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuProcessUtilization function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuProcessUtilization(nvmlDevice nvmlDevice, LastSeenTimeStamp uint64, VgpuProcessSamplesCount *uint32, UtilizationSamples *VgpuProcessUtilizationSample) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLastSeenTimeStamp, _ := (C.ulonglong)(LastSeenTimeStamp), cgoAllocsUnknown + cVgpuProcessSamplesCount, _ := (*C.uint)(unsafe.Pointer(VgpuProcessSamplesCount)), cgoAllocsUnknown + cUtilizationSamples, _ := (*C.nvmlVgpuProcessUtilizationSample_t)(unsafe.Pointer(UtilizationSamples)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuProcessUtilization(cnvmlDevice, cLastSeenTimeStamp, cVgpuProcessSamplesCount, cUtilizationSamples) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetVgpuProcessesUtilizationInfo function as declared in nvml/nvml.h +func nvmlDeviceGetVgpuProcessesUtilizationInfo(nvmlDevice nvmlDevice, VgpuProcUtilInfo *VgpuProcessesUtilizationInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cVgpuProcUtilInfo, _ := (*C.nvmlVgpuProcessesUtilizationInfo_t)(unsafe.Pointer(VgpuProcUtilInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetVgpuProcessesUtilizationInfo(cnvmlDevice, cVgpuProcUtilInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetAccountingMode function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetAccountingMode(nvmlVgpuInstance nvmlVgpuInstance, Mode *EnableState) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cMode, _ := (*C.nvmlEnableState_t)(unsafe.Pointer(Mode)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetAccountingMode(cnvmlVgpuInstance, cMode) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetAccountingPids function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetAccountingPids(nvmlVgpuInstance nvmlVgpuInstance, Count *uint32, Pids *uint32) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + cPids, _ := (*C.uint)(unsafe.Pointer(Pids)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetAccountingPids(cnvmlVgpuInstance, cCount, cPids) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetAccountingStats function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetAccountingStats(nvmlVgpuInstance nvmlVgpuInstance, Pid uint32, Stats *AccountingStats) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cPid, _ := (C.uint)(Pid), cgoAllocsUnknown + cStats, _ := (*C.nvmlAccountingStats_t)(unsafe.Pointer(Stats)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetAccountingStats(cnvmlVgpuInstance, cPid, cStats) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceClearAccountingPids function as declared in nvml/nvml.h +func nvmlVgpuInstanceClearAccountingPids(nvmlVgpuInstance nvmlVgpuInstance) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceClearAccountingPids(cnvmlVgpuInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetLicenseInfo_v2 function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetLicenseInfo_v2(nvmlVgpuInstance nvmlVgpuInstance, LicenseInfo *VgpuLicenseInfo) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cLicenseInfo, _ := (*C.nvmlVgpuLicenseInfo_t)(unsafe.Pointer(LicenseInfo)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetLicenseInfo_v2(cnvmlVgpuInstance, cLicenseInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGetExcludedDeviceCount function as declared in nvml/nvml.h +func nvmlGetExcludedDeviceCount(DeviceCount *uint32) Return { + cDeviceCount, _ := (*C.uint)(unsafe.Pointer(DeviceCount)), cgoAllocsUnknown + __ret := C.nvmlGetExcludedDeviceCount(cDeviceCount) + __v := (Return)(__ret) + return __v +} + +// nvmlGetExcludedDeviceInfoByIndex function as declared in nvml/nvml.h +func nvmlGetExcludedDeviceInfoByIndex(Index uint32, Info *ExcludedDeviceInfo) Return { + cIndex, _ := (C.uint)(Index), cgoAllocsUnknown + cInfo, _ := (*C.nvmlExcludedDeviceInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlGetExcludedDeviceInfoByIndex(cIndex, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceReadWritePRM_v1 function as declared in nvml/nvml.h +func nvmlDeviceReadWritePRM_v1(nvmlDevice nvmlDevice, Buffer *PRMTLV_v1) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cBuffer, _ := (*C.nvmlPRMTLV_v1_t)(unsafe.Pointer(Buffer)), cgoAllocsUnknown + __ret := C.nvmlDeviceReadWritePRM_v1(cnvmlDevice, cBuffer) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceSetMigMode function as declared in nvml/nvml.h +func nvmlDeviceSetMigMode(nvmlDevice nvmlDevice, Mode uint32, ActivationStatus *Return) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cMode, _ := (C.uint)(Mode), cgoAllocsUnknown + cActivationStatus, _ := (*C.nvmlReturn_t)(unsafe.Pointer(ActivationStatus)), cgoAllocsUnknown + __ret := C.nvmlDeviceSetMigMode(cnvmlDevice, cMode, cActivationStatus) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMigMode function as declared in nvml/nvml.h +func nvmlDeviceGetMigMode(nvmlDevice nvmlDevice, CurrentMode *uint32, PendingMode *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrentMode, _ := (*C.uint)(unsafe.Pointer(CurrentMode)), cgoAllocsUnknown + cPendingMode, _ := (*C.uint)(unsafe.Pointer(PendingMode)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMigMode(cnvmlDevice, cCurrentMode, cPendingMode) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstanceProfileInfo function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstanceProfileInfo(nvmlDevice nvmlDevice, Profile uint32, Info *GpuInstanceProfileInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfile, _ := (C.uint)(Profile), cgoAllocsUnknown + cInfo, _ := (*C.nvmlGpuInstanceProfileInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstanceProfileInfo(cnvmlDevice, cProfile, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstanceProfileInfoV function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstanceProfileInfoV(nvmlDevice nvmlDevice, Profile uint32, Info *GpuInstanceProfileInfo_v2) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfile, _ := (C.uint)(Profile), cgoAllocsUnknown + cInfo, _ := (*C.nvmlGpuInstanceProfileInfo_v2_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstanceProfileInfoV(cnvmlDevice, cProfile, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstanceProfileInfoByIdV function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstanceProfileInfoByIdV(nvmlDevice nvmlDevice, ProfileId uint32, Info *GpuInstanceProfileInfo_v2) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cInfo, _ := (*C.nvmlGpuInstanceProfileInfo_v2_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstanceProfileInfoByIdV(cnvmlDevice, cProfileId, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstancePossiblePlacements_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstancePossiblePlacements_v2(nvmlDevice nvmlDevice, ProfileId uint32, Placements *GpuInstancePlacement, Count *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cPlacements, _ := (*C.nvmlGpuInstancePlacement_t)(unsafe.Pointer(Placements)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstancePossiblePlacements_v2(cnvmlDevice, cProfileId, cPlacements, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstanceRemainingCapacity function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstanceRemainingCapacity(nvmlDevice nvmlDevice, ProfileId uint32, Count *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstanceRemainingCapacity(cnvmlDevice, cProfileId, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceCreateGpuInstance function as declared in nvml/nvml.h +func nvmlDeviceCreateGpuInstance(nvmlDevice nvmlDevice, ProfileId uint32, nvmlGpuInstance *nvmlGpuInstance) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cnvmlGpuInstance, _ := (*C.nvmlGpuInstance_t)(unsafe.Pointer(nvmlGpuInstance)), cgoAllocsUnknown + __ret := C.nvmlDeviceCreateGpuInstance(cnvmlDevice, cProfileId, cnvmlGpuInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceCreateGpuInstanceWithPlacement function as declared in nvml/nvml.h +func nvmlDeviceCreateGpuInstanceWithPlacement(nvmlDevice nvmlDevice, ProfileId uint32, Placement *GpuInstancePlacement, nvmlGpuInstance *nvmlGpuInstance) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cPlacement, _ := (*C.nvmlGpuInstancePlacement_t)(unsafe.Pointer(Placement)), cgoAllocsUnknown + cnvmlGpuInstance, _ := (*C.nvmlGpuInstance_t)(unsafe.Pointer(nvmlGpuInstance)), cgoAllocsUnknown + __ret := C.nvmlDeviceCreateGpuInstanceWithPlacement(cnvmlDevice, cProfileId, cPlacement, cnvmlGpuInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceDestroy function as declared in nvml/nvml.h +func nvmlGpuInstanceDestroy(nvmlGpuInstance nvmlGpuInstance) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceDestroy(cnvmlGpuInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstances function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstances(nvmlDevice nvmlDevice, ProfileId uint32, GpuInstances *nvmlGpuInstance, Count *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cGpuInstances, _ := (*C.nvmlGpuInstance_t)(unsafe.Pointer(GpuInstances)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstances(cnvmlDevice, cProfileId, cGpuInstances, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstanceById function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstanceById(nvmlDevice nvmlDevice, Id uint32, nvmlGpuInstance *nvmlGpuInstance) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cId, _ := (C.uint)(Id), cgoAllocsUnknown + cnvmlGpuInstance, _ := (*C.nvmlGpuInstance_t)(unsafe.Pointer(nvmlGpuInstance)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstanceById(cnvmlDevice, cId, cnvmlGpuInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetInfo function as declared in nvml/nvml.h +func nvmlGpuInstanceGetInfo(nvmlGpuInstance nvmlGpuInstance, Info *nvmlGpuInstanceInfo) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlGpuInstanceInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetInfo(cnvmlGpuInstance, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetComputeInstanceProfileInfo function as declared in nvml/nvml.h +func nvmlGpuInstanceGetComputeInstanceProfileInfo(nvmlGpuInstance nvmlGpuInstance, Profile uint32, EngProfile uint32, Info *ComputeInstanceProfileInfo) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cProfile, _ := (C.uint)(Profile), cgoAllocsUnknown + cEngProfile, _ := (C.uint)(EngProfile), cgoAllocsUnknown + cInfo, _ := (*C.nvmlComputeInstanceProfileInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetComputeInstanceProfileInfo(cnvmlGpuInstance, cProfile, cEngProfile, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetComputeInstanceProfileInfoV function as declared in nvml/nvml.h +func nvmlGpuInstanceGetComputeInstanceProfileInfoV(nvmlGpuInstance nvmlGpuInstance, Profile uint32, EngProfile uint32, Info *ComputeInstanceProfileInfo_v2) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cProfile, _ := (C.uint)(Profile), cgoAllocsUnknown + cEngProfile, _ := (C.uint)(EngProfile), cgoAllocsUnknown + cInfo, _ := (*C.nvmlComputeInstanceProfileInfo_v2_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetComputeInstanceProfileInfoV(cnvmlGpuInstance, cProfile, cEngProfile, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetComputeInstanceRemainingCapacity function as declared in nvml/nvml.h +func nvmlGpuInstanceGetComputeInstanceRemainingCapacity(nvmlGpuInstance nvmlGpuInstance, ProfileId uint32, Count *uint32) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetComputeInstanceRemainingCapacity(cnvmlGpuInstance, cProfileId, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetComputeInstancePossiblePlacements function as declared in nvml/nvml.h +func nvmlGpuInstanceGetComputeInstancePossiblePlacements(nvmlGpuInstance nvmlGpuInstance, ProfileId uint32, Placements *ComputeInstancePlacement, Count *uint32) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cPlacements, _ := (*C.nvmlComputeInstancePlacement_t)(unsafe.Pointer(Placements)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetComputeInstancePossiblePlacements(cnvmlGpuInstance, cProfileId, cPlacements, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceCreateComputeInstance function as declared in nvml/nvml.h +func nvmlGpuInstanceCreateComputeInstance(nvmlGpuInstance nvmlGpuInstance, ProfileId uint32, nvmlComputeInstance *nvmlComputeInstance) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cnvmlComputeInstance, _ := (*C.nvmlComputeInstance_t)(unsafe.Pointer(nvmlComputeInstance)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceCreateComputeInstance(cnvmlGpuInstance, cProfileId, cnvmlComputeInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceCreateComputeInstanceWithPlacement function as declared in nvml/nvml.h +func nvmlGpuInstanceCreateComputeInstanceWithPlacement(nvmlGpuInstance nvmlGpuInstance, ProfileId uint32, Placement *ComputeInstancePlacement, nvmlComputeInstance *nvmlComputeInstance) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cPlacement, _ := (*C.nvmlComputeInstancePlacement_t)(unsafe.Pointer(Placement)), cgoAllocsUnknown + cnvmlComputeInstance, _ := (*C.nvmlComputeInstance_t)(unsafe.Pointer(nvmlComputeInstance)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceCreateComputeInstanceWithPlacement(cnvmlGpuInstance, cProfileId, cPlacement, cnvmlComputeInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlComputeInstanceDestroy function as declared in nvml/nvml.h +func nvmlComputeInstanceDestroy(nvmlComputeInstance nvmlComputeInstance) Return { + cnvmlComputeInstance, _ := *(*C.nvmlComputeInstance_t)(unsafe.Pointer(&nvmlComputeInstance)), cgoAllocsUnknown + __ret := C.nvmlComputeInstanceDestroy(cnvmlComputeInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetComputeInstances function as declared in nvml/nvml.h +func nvmlGpuInstanceGetComputeInstances(nvmlGpuInstance nvmlGpuInstance, ProfileId uint32, ComputeInstances *nvmlComputeInstance, Count *uint32) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cComputeInstances, _ := (*C.nvmlComputeInstance_t)(unsafe.Pointer(ComputeInstances)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetComputeInstances(cnvmlGpuInstance, cProfileId, cComputeInstances, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlGpuInstanceGetComputeInstanceById function as declared in nvml/nvml.h +func nvmlGpuInstanceGetComputeInstanceById(nvmlGpuInstance nvmlGpuInstance, Id uint32, nvmlComputeInstance *nvmlComputeInstance) Return { + cnvmlGpuInstance, _ := *(*C.nvmlGpuInstance_t)(unsafe.Pointer(&nvmlGpuInstance)), cgoAllocsUnknown + cId, _ := (C.uint)(Id), cgoAllocsUnknown + cnvmlComputeInstance, _ := (*C.nvmlComputeInstance_t)(unsafe.Pointer(nvmlComputeInstance)), cgoAllocsUnknown + __ret := C.nvmlGpuInstanceGetComputeInstanceById(cnvmlGpuInstance, cId, cnvmlComputeInstance) + __v := (Return)(__ret) + return __v +} + +// nvmlComputeInstanceGetInfo_v2 function as declared in nvml/nvml.h +func nvmlComputeInstanceGetInfo_v2(nvmlComputeInstance nvmlComputeInstance, Info *nvmlComputeInstanceInfo) Return { + cnvmlComputeInstance, _ := *(*C.nvmlComputeInstance_t)(unsafe.Pointer(&nvmlComputeInstance)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlComputeInstanceInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlComputeInstanceGetInfo_v2(cnvmlComputeInstance, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceIsMigDeviceHandle function as declared in nvml/nvml.h +func nvmlDeviceIsMigDeviceHandle(nvmlDevice nvmlDevice, IsMigDevice *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIsMigDevice, _ := (*C.uint)(unsafe.Pointer(IsMigDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceIsMigDeviceHandle(cnvmlDevice, cIsMigDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstanceId function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstanceId(nvmlDevice nvmlDevice, Id *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cId, _ := (*C.uint)(unsafe.Pointer(Id)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstanceId(cnvmlDevice, cId) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetComputeInstanceId function as declared in nvml/nvml.h +func nvmlDeviceGetComputeInstanceId(nvmlDevice nvmlDevice, Id *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cId, _ := (*C.uint)(unsafe.Pointer(Id)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetComputeInstanceId(cnvmlDevice, cId) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMaxMigDeviceCount function as declared in nvml/nvml.h +func nvmlDeviceGetMaxMigDeviceCount(nvmlDevice nvmlDevice, Count *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMaxMigDeviceCount(cnvmlDevice, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMigDeviceHandleByIndex function as declared in nvml/nvml.h +func nvmlDeviceGetMigDeviceHandleByIndex(nvmlDevice nvmlDevice, Index uint32, MigDevice *nvmlDevice) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cIndex, _ := (C.uint)(Index), cgoAllocsUnknown + cMigDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(MigDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMigDeviceHandleByIndex(cnvmlDevice, cIndex, cMigDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDeviceHandleFromMigDeviceHandle function as declared in nvml/nvml.h +func nvmlDeviceGetDeviceHandleFromMigDeviceHandle(MigDevice nvmlDevice, nvmlDevice *nvmlDevice) Return { + cMigDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&MigDevice)), cgoAllocsUnknown + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDeviceHandleFromMigDeviceHandle(cMigDevice, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmMetricsGet function as declared in nvml/nvml.h +func nvmlGpmMetricsGet(MetricsGet *nvmlGpmMetricsGetType) Return { + cMetricsGet, _ := (*C.nvmlGpmMetricsGet_t)(unsafe.Pointer(MetricsGet)), cgoAllocsUnknown + __ret := C.nvmlGpmMetricsGet(cMetricsGet) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmSampleFree function as declared in nvml/nvml.h +func nvmlGpmSampleFree(nvmlGpmSample nvmlGpmSample) Return { + cnvmlGpmSample, _ := *(*C.nvmlGpmSample_t)(unsafe.Pointer(&nvmlGpmSample)), cgoAllocsUnknown + __ret := C.nvmlGpmSampleFree(cnvmlGpmSample) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmSampleAlloc function as declared in nvml/nvml.h +func nvmlGpmSampleAlloc(nvmlGpmSample *nvmlGpmSample) Return { + cnvmlGpmSample, _ := (*C.nvmlGpmSample_t)(unsafe.Pointer(nvmlGpmSample)), cgoAllocsUnknown + __ret := C.nvmlGpmSampleAlloc(cnvmlGpmSample) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmSampleGet function as declared in nvml/nvml.h +func nvmlGpmSampleGet(nvmlDevice nvmlDevice, nvmlGpmSample nvmlGpmSample) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cnvmlGpmSample, _ := *(*C.nvmlGpmSample_t)(unsafe.Pointer(&nvmlGpmSample)), cgoAllocsUnknown + __ret := C.nvmlGpmSampleGet(cnvmlDevice, cnvmlGpmSample) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmMigSampleGet function as declared in nvml/nvml.h +func nvmlGpmMigSampleGet(nvmlDevice nvmlDevice, GpuInstanceId uint32, nvmlGpmSample nvmlGpmSample) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cGpuInstanceId, _ := (C.uint)(GpuInstanceId), cgoAllocsUnknown + cnvmlGpmSample, _ := *(*C.nvmlGpmSample_t)(unsafe.Pointer(&nvmlGpmSample)), cgoAllocsUnknown + __ret := C.nvmlGpmMigSampleGet(cnvmlDevice, cGpuInstanceId, cnvmlGpmSample) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmQueryDeviceSupport function as declared in nvml/nvml.h +func nvmlGpmQueryDeviceSupport(nvmlDevice nvmlDevice, GpmSupport *GpmSupport) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cGpmSupport, _ := (*C.nvmlGpmSupport_t)(unsafe.Pointer(GpmSupport)), cgoAllocsUnknown + __ret := C.nvmlGpmQueryDeviceSupport(cnvmlDevice, cGpmSupport) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmQueryIfStreamingEnabled function as declared in nvml/nvml.h +func nvmlGpmQueryIfStreamingEnabled(nvmlDevice nvmlDevice, State *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cState, _ := (*C.uint)(unsafe.Pointer(State)), cgoAllocsUnknown + __ret := C.nvmlGpmQueryIfStreamingEnabled(cnvmlDevice, cState) + __v := (Return)(__ret) + return __v +} + +// nvmlGpmSetStreamingEnabled function as declared in nvml/nvml.h +func nvmlGpmSetStreamingEnabled(nvmlDevice nvmlDevice, State uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cState, _ := (C.uint)(State), cgoAllocsUnknown + __ret := C.nvmlGpmSetStreamingEnabled(cnvmlDevice, cState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCapabilities function as declared in nvml/nvml.h +func nvmlDeviceGetCapabilities(nvmlDevice nvmlDevice, Caps *DeviceCapabilities) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCaps, _ := (*C.nvmlDeviceCapabilities_t)(unsafe.Pointer(Caps)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCapabilities(cnvmlDevice, cCaps) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceWorkloadPowerProfileGetProfilesInfo function as declared in nvml/nvml.h +func nvmlDeviceWorkloadPowerProfileGetProfilesInfo(nvmlDevice nvmlDevice, ProfilesInfo *WorkloadPowerProfileProfilesInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfilesInfo, _ := (*C.nvmlWorkloadPowerProfileProfilesInfo_t)(unsafe.Pointer(ProfilesInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceWorkloadPowerProfileGetProfilesInfo(cnvmlDevice, cProfilesInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceWorkloadPowerProfileGetCurrentProfiles function as declared in nvml/nvml.h +func nvmlDeviceWorkloadPowerProfileGetCurrentProfiles(nvmlDevice nvmlDevice, CurrentProfiles *WorkloadPowerProfileCurrentProfiles) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrentProfiles, _ := (*C.nvmlWorkloadPowerProfileCurrentProfiles_t)(unsafe.Pointer(CurrentProfiles)), cgoAllocsUnknown + __ret := C.nvmlDeviceWorkloadPowerProfileGetCurrentProfiles(cnvmlDevice, cCurrentProfiles) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceWorkloadPowerProfileSetRequestedProfiles function as declared in nvml/nvml.h +func nvmlDeviceWorkloadPowerProfileSetRequestedProfiles(nvmlDevice nvmlDevice, RequestedProfiles *WorkloadPowerProfileRequestedProfiles) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cRequestedProfiles, _ := (*C.nvmlWorkloadPowerProfileRequestedProfiles_t)(unsafe.Pointer(RequestedProfiles)), cgoAllocsUnknown + __ret := C.nvmlDeviceWorkloadPowerProfileSetRequestedProfiles(cnvmlDevice, cRequestedProfiles) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceWorkloadPowerProfileClearRequestedProfiles function as declared in nvml/nvml.h +func nvmlDeviceWorkloadPowerProfileClearRequestedProfiles(nvmlDevice nvmlDevice, RequestedProfiles *WorkloadPowerProfileRequestedProfiles) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cRequestedProfiles, _ := (*C.nvmlWorkloadPowerProfileRequestedProfiles_t)(unsafe.Pointer(RequestedProfiles)), cgoAllocsUnknown + __ret := C.nvmlDeviceWorkloadPowerProfileClearRequestedProfiles(cnvmlDevice, cRequestedProfiles) + __v := (Return)(__ret) + return __v +} + +// nvmlDevicePowerSmoothingActivatePresetProfile function as declared in nvml/nvml.h +func nvmlDevicePowerSmoothingActivatePresetProfile(nvmlDevice nvmlDevice, Profile *PowerSmoothingProfile) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfile, _ := (*C.nvmlPowerSmoothingProfile_t)(unsafe.Pointer(Profile)), cgoAllocsUnknown + __ret := C.nvmlDevicePowerSmoothingActivatePresetProfile(cnvmlDevice, cProfile) + __v := (Return)(__ret) + return __v +} + +// nvmlDevicePowerSmoothingUpdatePresetProfileParam function as declared in nvml/nvml.h +func nvmlDevicePowerSmoothingUpdatePresetProfileParam(nvmlDevice nvmlDevice, Profile *PowerSmoothingProfile) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfile, _ := (*C.nvmlPowerSmoothingProfile_t)(unsafe.Pointer(Profile)), cgoAllocsUnknown + __ret := C.nvmlDevicePowerSmoothingUpdatePresetProfileParam(cnvmlDevice, cProfile) + __v := (Return)(__ret) + return __v +} + +// nvmlDevicePowerSmoothingSetState function as declared in nvml/nvml.h +func nvmlDevicePowerSmoothingSetState(nvmlDevice nvmlDevice, State *PowerSmoothingState) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cState, _ := (*C.nvmlPowerSmoothingState_t)(unsafe.Pointer(State)), cgoAllocsUnknown + __ret := C.nvmlDevicePowerSmoothingSetState(cnvmlDevice, cState) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetSramUniqueUncorrectedEccErrorCounts function as declared in nvml/nvml.h +func nvmlDeviceGetSramUniqueUncorrectedEccErrorCounts(nvmlDevice nvmlDevice, ErrorCounts *EccSramUniqueUncorrectedErrorCounts) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cErrorCounts, _ := (*C.nvmlEccSramUniqueUncorrectedErrorCounts_t)(unsafe.Pointer(ErrorCounts)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetSramUniqueUncorrectedEccErrorCounts(cnvmlDevice, cErrorCounts) + __v := (Return)(__ret) + return __v +} + +// nvmlInit_v1 function as declared in nvml/nvml.h +func nvmlInit_v1() Return { + __ret := C.nvmlInit() + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetCount_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetCount_v1(DeviceCount *uint32) Return { + cDeviceCount, _ := (*C.uint)(unsafe.Pointer(DeviceCount)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetCount(cDeviceCount) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHandleByIndex_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetHandleByIndex_v1(Index uint32, nvmlDevice *nvmlDevice) Return { + cIndex, _ := (C.uint)(Index), cgoAllocsUnknown + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHandleByIndex(cIndex, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetHandleByPciBusId_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetHandleByPciBusId_v1(PciBusId string, nvmlDevice *nvmlDevice) Return { + cPciBusId, _ := unpackPCharString(PciBusId) + cnvmlDevice, _ := (*C.nvmlDevice_t)(unsafe.Pointer(nvmlDevice)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetHandleByPciBusId(cPciBusId, cnvmlDevice) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPciInfo_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetPciInfo_v1(nvmlDevice nvmlDevice, Pci *PciInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPci, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(Pci)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPciInfo(cnvmlDevice, cPci) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetPciInfo_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetPciInfo_v2(nvmlDevice nvmlDevice, Pci *PciInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPci, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(Pci)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetPciInfo_v2(cnvmlDevice, cPci) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetNvLinkRemotePciInfo_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetNvLinkRemotePciInfo_v1(nvmlDevice nvmlDevice, Link uint32, Pci *PciInfo) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cLink, _ := (C.uint)(Link), cgoAllocsUnknown + cPci, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(Pci)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetNvLinkRemotePciInfo(cnvmlDevice, cLink, cPci) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGridLicensableFeatures_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetGridLicensableFeatures_v1(nvmlDevice nvmlDevice, PGridLicensableFeatures *GridLicensableFeatures) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPGridLicensableFeatures, _ := (*C.nvmlGridLicensableFeatures_t)(unsafe.Pointer(PGridLicensableFeatures)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGridLicensableFeatures(cnvmlDevice, cPGridLicensableFeatures) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGridLicensableFeatures_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetGridLicensableFeatures_v2(nvmlDevice nvmlDevice, PGridLicensableFeatures *GridLicensableFeatures) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPGridLicensableFeatures, _ := (*C.nvmlGridLicensableFeatures_t)(unsafe.Pointer(PGridLicensableFeatures)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGridLicensableFeatures_v2(cnvmlDevice, cPGridLicensableFeatures) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGridLicensableFeatures_v3 function as declared in nvml/nvml.h +func nvmlDeviceGetGridLicensableFeatures_v3(nvmlDevice nvmlDevice, PGridLicensableFeatures *GridLicensableFeatures) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cPGridLicensableFeatures, _ := (*C.nvmlGridLicensableFeatures_t)(unsafe.Pointer(PGridLicensableFeatures)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGridLicensableFeatures_v3(cnvmlDevice, cPGridLicensableFeatures) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceRemoveGpu_v1 function as declared in nvml/nvml.h +func nvmlDeviceRemoveGpu_v1(PciInfo *PciInfo) Return { + cPciInfo, _ := (*C.nvmlPciInfo_t)(unsafe.Pointer(PciInfo)), cgoAllocsUnknown + __ret := C.nvmlDeviceRemoveGpu(cPciInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlEventSetWait_v1 function as declared in nvml/nvml.h +func nvmlEventSetWait_v1(Set nvmlEventSet, Data *nvmlEventData, Timeoutms uint32) Return { + cSet, _ := *(*C.nvmlEventSet_t)(unsafe.Pointer(&Set)), cgoAllocsUnknown + cData, _ := (*C.nvmlEventData_t)(unsafe.Pointer(Data)), cgoAllocsUnknown + cTimeoutms, _ := (C.uint)(Timeoutms), cgoAllocsUnknown + __ret := C.nvmlEventSetWait(cSet, cData, cTimeoutms) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetAttributes_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetAttributes_v1(nvmlDevice nvmlDevice, Attributes *DeviceAttributes) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cAttributes, _ := (*C.nvmlDeviceAttributes_t)(unsafe.Pointer(Attributes)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetAttributes(cnvmlDevice, cAttributes) + __v := (Return)(__ret) + return __v +} + +// nvmlComputeInstanceGetInfo_v1 function as declared in nvml/nvml.h +func nvmlComputeInstanceGetInfo_v1(nvmlComputeInstance nvmlComputeInstance, Info *nvmlComputeInstanceInfo) Return { + cnvmlComputeInstance, _ := *(*C.nvmlComputeInstance_t)(unsafe.Pointer(&nvmlComputeInstance)), cgoAllocsUnknown + cInfo, _ := (*C.nvmlComputeInstanceInfo_t)(unsafe.Pointer(Info)), cgoAllocsUnknown + __ret := C.nvmlComputeInstanceGetInfo(cnvmlComputeInstance, cInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetComputeRunningProcesses_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetComputeRunningProcesses_v1(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo_v1) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_v1_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetComputeRunningProcesses(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetComputeRunningProcesses_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetComputeRunningProcesses_v2(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo_v2) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_v2_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetComputeRunningProcesses_v2(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGraphicsRunningProcesses_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetGraphicsRunningProcesses_v1(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo_v1) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_v1_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGraphicsRunningProcesses(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGraphicsRunningProcesses_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetGraphicsRunningProcesses_v2(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo_v2) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_v2_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGraphicsRunningProcesses_v2(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMPSComputeRunningProcesses_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetMPSComputeRunningProcesses_v1(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo_v1) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_v1_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMPSComputeRunningProcesses(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetMPSComputeRunningProcesses_v2 function as declared in nvml/nvml.h +func nvmlDeviceGetMPSComputeRunningProcesses_v2(nvmlDevice nvmlDevice, InfoCount *uint32, Infos *ProcessInfo_v2) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cInfoCount, _ := (*C.uint)(unsafe.Pointer(InfoCount)), cgoAllocsUnknown + cInfos, _ := (*C.nvmlProcessInfo_v2_t)(unsafe.Pointer(Infos)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetMPSComputeRunningProcesses_v2(cnvmlDevice, cInfoCount, cInfos) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetGpuInstancePossiblePlacements_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetGpuInstancePossiblePlacements_v1(nvmlDevice nvmlDevice, ProfileId uint32, Placements *GpuInstancePlacement, Count *uint32) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cProfileId, _ := (C.uint)(ProfileId), cgoAllocsUnknown + cPlacements, _ := (*C.nvmlGpuInstancePlacement_t)(unsafe.Pointer(Placements)), cgoAllocsUnknown + cCount, _ := (*C.uint)(unsafe.Pointer(Count)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetGpuInstancePossiblePlacements(cnvmlDevice, cProfileId, cPlacements, cCount) + __v := (Return)(__ret) + return __v +} + +// nvmlVgpuInstanceGetLicenseInfo_v1 function as declared in nvml/nvml.h +func nvmlVgpuInstanceGetLicenseInfo_v1(nvmlVgpuInstance nvmlVgpuInstance, LicenseInfo *VgpuLicenseInfo) Return { + cnvmlVgpuInstance, _ := (C.nvmlVgpuInstance_t)(nvmlVgpuInstance), cgoAllocsUnknown + cLicenseInfo, _ := (*C.nvmlVgpuLicenseInfo_t)(unsafe.Pointer(LicenseInfo)), cgoAllocsUnknown + __ret := C.nvmlVgpuInstanceGetLicenseInfo(cnvmlVgpuInstance, cLicenseInfo) + __v := (Return)(__ret) + return __v +} + +// nvmlDeviceGetDriverModel_v1 function as declared in nvml/nvml.h +func nvmlDeviceGetDriverModel_v1(nvmlDevice nvmlDevice, Current *DriverModel, Pending *DriverModel) Return { + cnvmlDevice, _ := *(*C.nvmlDevice_t)(unsafe.Pointer(&nvmlDevice)), cgoAllocsUnknown + cCurrent, _ := (*C.nvmlDriverModel_t)(unsafe.Pointer(Current)), cgoAllocsUnknown + cPending, _ := (*C.nvmlDriverModel_t)(unsafe.Pointer(Pending)), cgoAllocsUnknown + __ret := C.nvmlDeviceGetDriverModel(cnvmlDevice, cCurrent, cPending) + __v := (Return)(__ret) + return __v +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.h b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.h new file mode 100644 index 00000000..917a8c93 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/nvml.h @@ -0,0 +1,13557 @@ +/*** NVML VERSION: 13.0.39 ***/ +/*** From https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvml_dev/linux-x86_64/cuda_nvml_dev-linux-x86_64-13.0.39-archive.tar.xz ***/ +/* + * Copyright 1993-2025 NVIDIA Corporation. All rights reserved. + * + * NOTICE TO USER: + * + * This source code is subject to NVIDIA ownership rights under U.S. and + * international Copyright laws. Users and possessors of this source code + * are hereby granted a nonexclusive, royalty-free license to use this code + * in individual and commercial software. + * + * NVIDIA MAKES NO REPRESENTATION ABOUT THE SUITABILITY OF THIS SOURCE + * CODE FOR ANY PURPOSE. IT IS PROVIDED "AS IS" WITHOUT EXPRESS OR + * IMPLIED WARRANTY OF ANY KIND. NVIDIA DISCLAIMS ALL WARRANTIES WITH + * REGARD TO THIS SOURCE CODE, INCLUDING ALL IMPLIED WARRANTIES OF + * MERCHANTABILITY, NONINFRINGEMENT, AND FITNESS FOR A PARTICULAR PURPOSE. + * IN NO EVENT SHALL NVIDIA BE LIABLE FOR ANY SPECIAL, INDIRECT, INCIDENTAL, + * OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS + * OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE + * OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE + * OR PERFORMANCE OF THIS SOURCE CODE. + * + * U.S. Government End Users. This source code is a "commercial item" as + * that term is defined at 48 C.F.R. 2.101 (OCT 1995), consisting of + * "commercial computer software" and "commercial computer software + * documentation" as such terms are used in 48 C.F.R. 12.212 (SEPT 1995) + * and is provided to the U.S. Government only as a commercial end item. + * Consistent with 48 C.F.R.12.212 and 48 C.F.R. 227.7202-1 through + * 227.7202-4 (JUNE 1995), all U.S. Government End Users acquire the + * source code with only those rights set forth herein. + * + * Any use of this source code in individual and commercial software must + * include, in the user documentation and internal comments to the code, + * the above Disclaimer and U.S. Government End Users Notice. + */ + +/* +NVML API Reference + +The NVIDIA Management Library (NVML) is a C-based programmatic interface for monitoring and +managing various states within NVIDIA Tesla &tm; GPUs. It is intended to be a platform for building +3rd party applications, and is also the underlying library for the NVIDIA-supported nvidia-smi +tool. NVML is thread-safe so it is safe to make simultaneous NVML calls from multiple threads. + +API Documentation + +Supported platforms: +- Windows: Windows Server 2008 R2 64bit, Windows Server 2012 R2 64bit, Windows 7 64bit, Windows 8 64bit, Windows 10 64bit +- Linux: 32-bit and 64-bit +- Hypervisors: Windows Server 2008R2/2012 Hyper-V 64bit, Citrix XenServer 6.2 SP1+, VMware ESX 5.1/5.5 + +Supported products: +- Full Support + - All Tesla products, starting with the Fermi architecture + - All Quadro products, starting with the Fermi architecture + - All vGPU Software products, starting with the Kepler architecture + - Selected GeForce Titan products +- Limited Support + - All Geforce products, starting with the Fermi architecture + +The NVML library can be found at \%ProgramW6432\%\\"NVIDIA Corporation"\\NVSMI\\ on Windows. It is +not be added to the system path by default. To dynamically link to NVML, add this path to the PATH +environmental variable. To dynamically load NVML, call LoadLibrary with this path. + +On Linux the NVML library will be found on the standard library path. For 64 bit Linux, both the 32 bit +and 64 bit NVML libraries will be installed. + +Online documentation for this library is available at http://docs.nvidia.com/deploy/nvml-api/index.html +*/ + +#ifndef __nvml_nvml_h__ +#define __nvml_nvml_h__ + +#ifdef __cplusplus +extern "C" { +#endif + +/* + * On Windows, set up methods for DLL export + * define NVML_STATIC_IMPORT when using nvml_loader library + */ +#if defined _WINDOWS + #if !defined NVML_STATIC_IMPORT + #if defined NVML_LIB_EXPORT + #define DECLDIR __declspec(dllexport) + #else + #define DECLDIR __declspec(dllimport) + #endif + #else + #define DECLDIR + #endif +#else + #define DECLDIR +#endif + +/* + * Deprecation definition. Starting CUDA 13.1 this will change to: + * #if defined _WINDOWS + * #define DEPRECATED(ver) __declspec(deprecated) + * #else + * #define DEPRECATED(ver) __attribute__((deprecated)) + * #endif + */ +#define DEPRECATED(ver) /* nop in CUDA 13.0, enabled in CUDA 13.1 */ + + #define NVML_MCDM_SUPPORT + +/** + * NVML API versioning support + */ +#define NVML_API_VERSION 13 +#define NVML_API_VERSION_STR "13" +/** + * Defining NVML_NO_UNVERSIONED_FUNC_DEFS will disable "auto upgrading" of APIs. + * e.g. the user will have to call nvmlInit_v2 instead of nvmlInit. Enable this + * guard if you need to support older versions of the API + */ +#ifndef NVML_NO_UNVERSIONED_FUNC_DEFS + #define nvmlInit nvmlInit_v2 + #define nvmlDeviceGetPciInfo nvmlDeviceGetPciInfo_v3 + #define nvmlDeviceGetCount nvmlDeviceGetCount_v2 + #define nvmlDeviceGetHandleByIndex nvmlDeviceGetHandleByIndex_v2 + #define nvmlDeviceGetHandleByPciBusId nvmlDeviceGetHandleByPciBusId_v2 + #define nvmlDeviceGetNvLinkRemotePciInfo nvmlDeviceGetNvLinkRemotePciInfo_v2 + #define nvmlDeviceRemoveGpu nvmlDeviceRemoveGpu_v2 + #define nvmlDeviceGetGridLicensableFeatures nvmlDeviceGetGridLicensableFeatures_v4 + #define nvmlEventSetWait nvmlEventSetWait_v2 + #define nvmlDeviceGetAttributes nvmlDeviceGetAttributes_v2 + #define nvmlComputeInstanceGetInfo nvmlComputeInstanceGetInfo_v2 + #define nvmlDeviceGetComputeRunningProcesses nvmlDeviceGetComputeRunningProcesses_v3 + #define nvmlDeviceGetGraphicsRunningProcesses nvmlDeviceGetGraphicsRunningProcesses_v3 + #define nvmlDeviceGetMPSComputeRunningProcesses nvmlDeviceGetMPSComputeRunningProcesses_v3 + #define nvmlBlacklistDeviceInfo_t nvmlExcludedDeviceInfo_t + #define nvmlGetBlacklistDeviceCount nvmlGetExcludedDeviceCount + #define nvmlGetBlacklistDeviceInfoByIndex nvmlGetExcludedDeviceInfoByIndex + #define nvmlDeviceGetGpuInstancePossiblePlacements nvmlDeviceGetGpuInstancePossiblePlacements_v2 + #define nvmlVgpuInstanceGetLicenseInfo nvmlVgpuInstanceGetLicenseInfo_v2 + #define nvmlDeviceGetDriverModel nvmlDeviceGetDriverModel_v2 +#endif // #ifndef NVML_NO_UNVERSIONED_FUNC_DEFS + +#define NVML_STRUCT_VERSION(data, ver) (unsigned int)(sizeof(nvml ## data ## _v ## ver ## _t) | \ + (ver << 24U)) + +/***************************************************************************************************/ +/** @defgroup nvmlDeviceStructs Device Structs + * @{ + */ +/***************************************************************************************************/ + +/** + * Special constant that some fields take when they are not available. + * Used when only part of the struct is not available. + * + * Each structure explicitly states when to check for this value. + */ +#define NVML_VALUE_NOT_AVAILABLE (-1) + +typedef struct +{ + struct nvmlDevice_st* handle; +} nvmlDevice_t; + +typedef struct +{ + struct nvmlGpuInstance_st* handle; +} nvmlGpuInstance_t; + +/** + * Buffer size guaranteed to be large enough for pci bus id + */ +#define NVML_DEVICE_PCI_BUS_ID_BUFFER_SIZE 32 + +/** + * Buffer size guaranteed to be large enough for pci bus id for \p busIdLegacy + */ +#define NVML_DEVICE_PCI_BUS_ID_BUFFER_V2_SIZE 16 + +/** + * PCI information about a GPU device. + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + unsigned int domain; //!< The PCI domain on which the device's bus resides, 0 to 0xffffffff + unsigned int bus; //!< The bus on which the device resides, 0 to 0xff + unsigned int device; //!< The device's id on the bus, 0 to 31 + + unsigned int pciDeviceId; //!< The combined 16-bit device id and 16-bit vendor id + unsigned int pciSubSystemId; //!< The 32-bit Sub System Device ID + + unsigned int baseClass; //!< The 8-bit PCI base class code + unsigned int subClass; //!< The 8-bit PCI sub class code + + char busId[NVML_DEVICE_PCI_BUS_ID_BUFFER_SIZE]; //!< The tuple domain:bus:device.function PCI identifier (& NULL terminator) +} nvmlPciInfoExt_v1_t; +typedef nvmlPciInfoExt_v1_t nvmlPciInfoExt_t; +#define nvmlPciInfoExt_v1 NVML_STRUCT_VERSION(PciInfoExt, 1) + +/** + * PCI information about a GPU device. + */ +typedef struct nvmlPciInfo_st +{ + char busIdLegacy[NVML_DEVICE_PCI_BUS_ID_BUFFER_V2_SIZE]; //!< The legacy tuple domain:bus:device.function PCI identifier (& NULL terminator) + unsigned int domain; //!< The PCI domain on which the device's bus resides, 0 to 0xffffffff + unsigned int bus; //!< The bus on which the device resides, 0 to 0xff + unsigned int device; //!< The device's id on the bus, 0 to 31 + unsigned int pciDeviceId; //!< The combined 16-bit device id and 16-bit vendor id + + // Added in NVML 2.285 API + unsigned int pciSubSystemId; //!< The 32-bit Sub System Device ID + + char busId[NVML_DEVICE_PCI_BUS_ID_BUFFER_SIZE]; //!< The tuple domain:bus:device.function PCI identifier (& NULL terminator) +} nvmlPciInfo_t; + +/** + * PCI format string for \p busIdLegacy + */ +#define NVML_DEVICE_PCI_BUS_ID_LEGACY_FMT "%04X:%02X:%02X.0" + +/** + * PCI format string for \p busId + */ +#define NVML_DEVICE_PCI_BUS_ID_FMT "%08X:%02X:%02X.0" + +/** + * Utility macro for filling the pci bus id format from a nvmlPciInfo_t + */ +#define NVML_DEVICE_PCI_BUS_ID_FMT_ARGS(pciInfo) (pciInfo)->domain, \ + (pciInfo)->bus, \ + (pciInfo)->device + +/** + * Detailed ECC error counts for a device. + * + * @deprecated Different GPU families can have different memory error counters + * See \ref nvmlDeviceGetMemoryErrorCounter + */ +typedef struct nvmlEccErrorCounts_st +{ + unsigned long long l1Cache; //!< L1 cache errors + unsigned long long l2Cache; //!< L2 cache errors + unsigned long long deviceMemory; //!< Device memory errors + unsigned long long registerFile; //!< Register file errors +} nvmlEccErrorCounts_t; + +/** + * Utilization information for a device. + * Each sample period may be between 1 second and 1/6 second, depending on the product being queried. + */ +typedef struct nvmlUtilization_st +{ + unsigned int gpu; //!< Percent of time over the past sample period during which one or more kernels was executing on the GPU + unsigned int memory; //!< Percent of time over the past sample period during which global (device) memory was being read or written +} nvmlUtilization_t; + +/** + * Memory allocation information for a device (v1). + * The total amount is equal to the sum of the amounts of free and used memory. + */ +typedef struct nvmlMemory_st +{ + unsigned long long total; //!< Total physical device memory (in bytes) + unsigned long long free; //!< Unallocated device memory (in bytes) + unsigned long long used; //!< Sum of Reserved and Allocated device memory (in bytes). + //!< Note that the driver/GPU always sets aside a small amount of memory for bookkeeping +} nvmlMemory_t; + +/** + * Memory allocation information for a device (v2). + * + * Version 2 adds versioning for the struct and the amount of system-reserved memory as an output. + */ +typedef struct nvmlMemory_v2_st +{ + unsigned int version; //!< Structure format version (must be 2) + unsigned long long total; //!< Total physical device memory (in bytes) + unsigned long long reserved; //!< Device memory (in bytes) reserved for system use (driver or firmware) + unsigned long long free; //!< Unallocated device memory (in bytes) + unsigned long long used; //!< Allocated device memory (in bytes). +} nvmlMemory_v2_t; + +#define nvmlMemory_v2 NVML_STRUCT_VERSION(Memory, 2) + +/** + * BAR1 Memory allocation Information for a device + */ +typedef struct nvmlBAR1Memory_st +{ + unsigned long long bar1Total; //!< Total BAR1 Memory (in bytes) + unsigned long long bar1Free; //!< Unallocated BAR1 Memory (in bytes) + unsigned long long bar1Used; //!< Allocated Used Memory (in bytes) +}nvmlBAR1Memory_t; + +/** + * Information about running compute processes on the GPU, legacy version + * for older versions of the API. + */ +typedef struct nvmlProcessInfo_v1_st +{ + unsigned int pid; //!< Process ID + unsigned long long usedGpuMemory; //!< Amount of used GPU memory in bytes. + //! Under WDDM, \ref NVML_VALUE_NOT_AVAILABLE is always reported + //! because Windows KMD manages all the memory and not the NVIDIA driver +} nvmlProcessInfo_v1_t; + +/** + * Information about running compute processes on the GPU + */ +typedef struct nvmlProcessInfo_v2_st +{ + unsigned int pid; //!< Process ID + unsigned long long usedGpuMemory; //!< Amount of used GPU memory in bytes. + //! Under WDDM, \ref NVML_VALUE_NOT_AVAILABLE is always reported + //! because Windows KMD manages all the memory and not the NVIDIA driver + unsigned int gpuInstanceId; //!< If MIG is enabled, stores a valid GPU instance ID. gpuInstanceId is set to + // 0xFFFFFFFF otherwise. + unsigned int computeInstanceId; //!< If MIG is enabled, stores a valid compute instance ID. computeInstanceId is set to + // 0xFFFFFFFF otherwise. +} nvmlProcessInfo_v2_t, nvmlProcessInfo_t; + +/** + * Information about running process on the GPU with protected memory + */ +typedef struct +{ + unsigned int pid; //!< Process ID + unsigned long long usedGpuMemory; //!< Amount of used GPU memory in bytes. + //! Under WDDM, \ref NVML_VALUE_NOT_AVAILABLE is always reported + //! because Windows KMD manages all the memory and not the NVIDIA driver + unsigned int gpuInstanceId; //!< If MIG is enabled, stores a valid GPU instance ID. gpuInstanceId is + // set to 0xFFFFFFFF otherwise. + unsigned int computeInstanceId; //!< If MIG is enabled, stores a valid compute instance ID. computeInstanceId + // is set to 0xFFFFFFFF otherwise. + unsigned long long usedGpuCcProtectedMemory; //!< Amount of used GPU conf compute protected memory in bytes. +} nvmlProcessDetail_v1_t; + +/** + * Information about all running processes on the GPU for the given mode + */ +typedef struct +{ + unsigned int version; //!< Struct version, MUST be nvmlProcessDetailList_v1 + unsigned int mode; //!< Process mode(Compute/Graphics/MPSCompute) + unsigned int numProcArrayEntries; //!< Number of process entries in procArray + nvmlProcessDetail_v1_t *procArray; //!< Process array +} nvmlProcessDetailList_v1_t; + +typedef nvmlProcessDetailList_v1_t nvmlProcessDetailList_t; + +/** + * nvmlProcessDetailList version + */ +#define nvmlProcessDetailList_v1 NVML_STRUCT_VERSION(ProcessDetailList, 1) + +typedef struct nvmlDeviceAttributes_st +{ + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count + unsigned int gpuInstanceSliceCount; //!< GPU instance slice count + unsigned int computeInstanceSliceCount; //!< Compute instance slice count + unsigned long long memorySizeMB; //!< Device memory size (in MiB) +} nvmlDeviceAttributes_t; + +/** + * C2C Mode information for a device + */ +typedef struct +{ + unsigned int isC2cEnabled; +} nvmlC2cModeInfo_v1_t; + +#define nvmlC2cModeInfo_v1 NVML_STRUCT_VERSION(C2cModeInfo, 1) + +/** + * Enum to represent device addressing mode values + */ +typedef enum +{ + NVML_DEVICE_ADDRESSING_MODE_NONE = 0, //!< No active mode + NVML_DEVICE_ADDRESSING_MODE_HMM = 1, //!< Heterogeneous Memory Management mode + NVML_DEVICE_ADDRESSING_MODE_ATS = 2, //!< Address Translation Services mode +} nvmlDeviceAddressingModeType_t; + +/** + * Struct to represent device addressing mode information + */ +typedef struct +{ + unsigned int version; //!< API version + unsigned int value; //!< One of \ref nvmlDeviceAddressingModeType_t +} nvmlDeviceAddressingMode_v1_t; +typedef nvmlDeviceAddressingMode_v1_t nvmlDeviceAddressingMode_t; + +#define nvmlDeviceAddressingMode_v1 NVML_STRUCT_VERSION(DeviceAddressingMode, 1) + +/** + * Struct to represent the NVML repair status + */ +typedef struct +{ + unsigned int version; //!< API version number + unsigned int bChannelRepairPending; //!< Reference to \a unsigned int + unsigned int bTpcRepairPending; //!< Reference to \a unsigned int +} nvmlRepairStatus_v1_t; +typedef nvmlRepairStatus_v1_t nvmlRepairStatus_t; + +#define nvmlRepairStatus_v1 NVML_STRUCT_VERSION(RepairStatus, 1) + +/** + * Possible values that classify the remap availability for each bank. The max + * field will contain the number of banks that have maximum remap availability + * (all reserved rows are available). None means that there are no reserved + * rows available. + */ +typedef struct nvmlRowRemapperHistogramValues_st +{ + unsigned int max; + unsigned int high; + unsigned int partial; + unsigned int low; + unsigned int none; +} nvmlRowRemapperHistogramValues_t; + +/** + * Enum to represent type of bridge chip + */ +typedef enum nvmlBridgeChipType_enum +{ + NVML_BRIDGE_CHIP_PLX = 0, + NVML_BRIDGE_CHIP_BRO4 = 1 +}nvmlBridgeChipType_t; + +/** + * Maximum number of NvLink links supported + */ +#define NVML_NVLINK_MAX_LINKS 18 + +/** + * Enum to represent the NvLink utilization counter packet units + */ +typedef enum nvmlNvLinkUtilizationCountUnits_enum +{ + NVML_NVLINK_COUNTER_UNIT_CYCLES = 0, // count by cycles + NVML_NVLINK_COUNTER_UNIT_PACKETS = 1, // count by packets + NVML_NVLINK_COUNTER_UNIT_BYTES = 2, // count by bytes + NVML_NVLINK_COUNTER_UNIT_RESERVED = 3, // count reserved for internal use + // this must be last + NVML_NVLINK_COUNTER_UNIT_COUNT +} nvmlNvLinkUtilizationCountUnits_t; + +/** + * Enum to represent the NvLink utilization counter packet types to count + * ** this is ONLY applicable with the units as packets or bytes + * ** as specified in \a nvmlNvLinkUtilizationCountUnits_t + * ** all packet filter descriptions are target GPU centric + * ** these can be "OR'd" together + */ +typedef enum nvmlNvLinkUtilizationCountPktTypes_enum +{ + NVML_NVLINK_COUNTER_PKTFILTER_NOP = 0x1, // no operation packets + NVML_NVLINK_COUNTER_PKTFILTER_READ = 0x2, // read packets + NVML_NVLINK_COUNTER_PKTFILTER_WRITE = 0x4, // write packets + NVML_NVLINK_COUNTER_PKTFILTER_RATOM = 0x8, // reduction atomic requests + NVML_NVLINK_COUNTER_PKTFILTER_NRATOM = 0x10, // non-reduction atomic requests + NVML_NVLINK_COUNTER_PKTFILTER_FLUSH = 0x20, // flush requests + NVML_NVLINK_COUNTER_PKTFILTER_RESPDATA = 0x40, // responses with data + NVML_NVLINK_COUNTER_PKTFILTER_RESPNODATA = 0x80, // responses without data + NVML_NVLINK_COUNTER_PKTFILTER_ALL = 0xFF // all packets +} nvmlNvLinkUtilizationCountPktTypes_t; + +/** + * Struct to define the NVLINK counter controls + */ +typedef struct nvmlNvLinkUtilizationControl_st +{ + nvmlNvLinkUtilizationCountUnits_t units; + nvmlNvLinkUtilizationCountPktTypes_t pktfilter; +} nvmlNvLinkUtilizationControl_t; + +/** + * Enum to represent NvLink queryable capabilities + */ +typedef enum nvmlNvLinkCapability_enum +{ + NVML_NVLINK_CAP_P2P_SUPPORTED = 0, // P2P over NVLink is supported + NVML_NVLINK_CAP_SYSMEM_ACCESS = 1, // Access to system memory is supported + NVML_NVLINK_CAP_P2P_ATOMICS = 2, // P2P atomics are supported + NVML_NVLINK_CAP_SYSMEM_ATOMICS= 3, // System memory atomics are supported + NVML_NVLINK_CAP_SLI_BRIDGE = 4, // SLI is supported over this link + NVML_NVLINK_CAP_VALID = 5, // Link is supported on this device + // should be last + NVML_NVLINK_CAP_COUNT +} nvmlNvLinkCapability_t; + +/** + * Enum to represent NvLink queryable error counters + */ +typedef enum nvmlNvLinkErrorCounter_enum +{ + NVML_NVLINK_ERROR_DL_REPLAY = 0, // Data link transmit replay error counter + NVML_NVLINK_ERROR_DL_RECOVERY = 1, // Data link transmit recovery error counter + NVML_NVLINK_ERROR_DL_CRC_FLIT = 2, // Data link receive flow control digit CRC error counter + NVML_NVLINK_ERROR_DL_CRC_DATA = 3, // Data link receive data CRC error counter + NVML_NVLINK_ERROR_DL_ECC_DATA = 4, // Data link receive data ECC error counter + + // this must be last + NVML_NVLINK_ERROR_COUNT +} nvmlNvLinkErrorCounter_t; + +/** + * Enum to represent NvLink's remote device type + */ +typedef enum nvmlIntNvLinkDeviceType_enum +{ + NVML_NVLINK_DEVICE_TYPE_GPU = 0x00, + NVML_NVLINK_DEVICE_TYPE_IBMNPU = 0x01, + NVML_NVLINK_DEVICE_TYPE_SWITCH = 0x02, + NVML_NVLINK_DEVICE_TYPE_UNKNOWN = 0xFF +} nvmlIntNvLinkDeviceType_t; + +/** + * Represents level relationships within a system between two GPUs + * The enums are spaced to allow for future relationships + */ +typedef enum nvmlGpuLevel_enum +{ + NVML_TOPOLOGY_INTERNAL = 0, // e.g. Tesla K80 + NVML_TOPOLOGY_SINGLE = 10, // all devices that only need traverse a single PCIe switch + NVML_TOPOLOGY_MULTIPLE = 20, // all devices that need not traverse a host bridge + NVML_TOPOLOGY_HOSTBRIDGE = 30, // all devices that are connected to the same host bridge + NVML_TOPOLOGY_NODE = 40, // all devices that are connected to the same NUMA node but possibly multiple host bridges + NVML_TOPOLOGY_SYSTEM = 50 // all devices in the system + + // there is purposefully no COUNT here because of the need for spacing above +} nvmlGpuTopologyLevel_t; + +/* Compatibility for CPU->NODE renaming */ +#define NVML_TOPOLOGY_CPU NVML_TOPOLOGY_NODE + +/* P2P Capability Index Status*/ +typedef enum nvmlGpuP2PStatus_enum +{ + NVML_P2P_STATUS_OK = 0, + NVML_P2P_STATUS_CHIPSET_NOT_SUPPORED, + NVML_P2P_STATUS_CHIPSET_NOT_SUPPORTED = NVML_P2P_STATUS_CHIPSET_NOT_SUPPORED, + NVML_P2P_STATUS_GPU_NOT_SUPPORTED, + NVML_P2P_STATUS_IOH_TOPOLOGY_NOT_SUPPORTED, + NVML_P2P_STATUS_DISABLED_BY_REGKEY, + NVML_P2P_STATUS_NOT_SUPPORTED, + NVML_P2P_STATUS_UNKNOWN + +} nvmlGpuP2PStatus_t; + +/* P2P Capability Index*/ +typedef enum nvmlGpuP2PCapsIndex_enum +{ + NVML_P2P_CAPS_INDEX_READ = 0, + NVML_P2P_CAPS_INDEX_WRITE = 1, + NVML_P2P_CAPS_INDEX_NVLINK = 2, + NVML_P2P_CAPS_INDEX_ATOMICS = 3, + NVML_P2P_CAPS_INDEX_PCI = 4, + /* + * DO NOT USE! NVML_P2P_CAPS_INDEX_PROP is deprecated. + * Use NVML_P2P_CAPS_INDEX_PCI instead. + */ + NVML_P2P_CAPS_INDEX_PROP = NVML_P2P_CAPS_INDEX_PCI, + NVML_P2P_CAPS_INDEX_UNKNOWN = 5, +}nvmlGpuP2PCapsIndex_t; + +/** + * Maximum limit on Physical Bridges per Board + */ +#define NVML_MAX_PHYSICAL_BRIDGE (128) + +/** + * Information about the Bridge Chip Firmware + */ +typedef struct nvmlBridgeChipInfo_st +{ + nvmlBridgeChipType_t type; //!< Type of Bridge Chip + unsigned int fwVersion; //!< Firmware Version. 0=Version is unavailable +}nvmlBridgeChipInfo_t; + +/** + * This structure stores the complete Hierarchy of the Bridge Chip within the board. The immediate + * bridge is stored at index 0 of bridgeInfoList, parent to immediate bridge is at index 1 and so forth. + */ +typedef struct nvmlBridgeChipHierarchy_st +{ + unsigned char bridgeCount; //!< Number of Bridge Chips on the Board + nvmlBridgeChipInfo_t bridgeChipInfo[NVML_MAX_PHYSICAL_BRIDGE]; //!< Hierarchy of Bridge Chips on the board +}nvmlBridgeChipHierarchy_t; + +/** + * Represents Type of Sampling Event + */ +typedef enum nvmlSamplingType_enum +{ + NVML_TOTAL_POWER_SAMPLES = 0, //!< To represent total power drawn by GPU + NVML_GPU_UTILIZATION_SAMPLES = 1, //!< To represent percent of time during which one or more kernels was executing on the GPU + NVML_MEMORY_UTILIZATION_SAMPLES = 2, //!< To represent percent of time during which global (device) memory was being read or written + NVML_ENC_UTILIZATION_SAMPLES = 3, //!< To represent percent of time during which NVENC remains busy + NVML_DEC_UTILIZATION_SAMPLES = 4, //!< To represent percent of time during which NVDEC remains busy + NVML_PROCESSOR_CLK_SAMPLES = 5, //!< To represent processor clock samples + NVML_MEMORY_CLK_SAMPLES = 6, //!< To represent memory clock samples + NVML_MODULE_POWER_SAMPLES = 7, //!< To represent module power samples for total module starting Grace Hopper + NVML_JPG_UTILIZATION_SAMPLES = 8, //!< To represent percent of time during which NVJPG remains busy + NVML_OFA_UTILIZATION_SAMPLES = 9, //!< To represent percent of time during which NVOFA remains busy + + // Keep this last + NVML_SAMPLINGTYPE_COUNT +}nvmlSamplingType_t; + +/** + * Represents the queryable PCIe utilization counters + */ +typedef enum nvmlPcieUtilCounter_enum +{ + NVML_PCIE_UTIL_TX_BYTES = 0, // 1KB granularity + NVML_PCIE_UTIL_RX_BYTES = 1, // 1KB granularity + + // Keep this last + NVML_PCIE_UTIL_COUNT +} nvmlPcieUtilCounter_t; + +/** + * Represents the type for sample value returned + */ +typedef enum nvmlValueType_enum +{ + NVML_VALUE_TYPE_DOUBLE = 0, + NVML_VALUE_TYPE_UNSIGNED_INT = 1, + NVML_VALUE_TYPE_UNSIGNED_LONG = 2, + NVML_VALUE_TYPE_UNSIGNED_LONG_LONG = 3, + NVML_VALUE_TYPE_SIGNED_LONG_LONG = 4, + NVML_VALUE_TYPE_SIGNED_INT = 5, + NVML_VALUE_TYPE_UNSIGNED_SHORT = 6, + + // Keep this last + NVML_VALUE_TYPE_COUNT +}nvmlValueType_t; + +/** + * Union to represent different types of Value + */ +typedef union nvmlValue_st +{ + double dVal; //!< If the value is double + int siVal; //!< If the value is signed int + unsigned int uiVal; //!< If the value is unsigned int + unsigned long ulVal; //!< If the value is unsigned long + unsigned long long ullVal; //!< If the value is unsigned long long + signed long long sllVal; //!< If the value is signed long long + unsigned short usVal; //!< If the value is unsigned short +}nvmlValue_t; + +/** + * Information for Sample + */ +typedef struct nvmlSample_st +{ + unsigned long long timeStamp; //!< CPU Timestamp in microseconds + nvmlValue_t sampleValue; //!< Sample Value +}nvmlSample_t; + +/** + * Represents type of perf policy for which violation times can be queried + */ +typedef enum nvmlPerfPolicyType_enum +{ + NVML_PERF_POLICY_POWER = 0, //!< How long did power violations cause the GPU to be below application clocks + NVML_PERF_POLICY_THERMAL = 1, //!< How long did thermal violations cause the GPU to be below application clocks + NVML_PERF_POLICY_SYNC_BOOST = 2, //!< How long did sync boost cause the GPU to be below application clocks + NVML_PERF_POLICY_BOARD_LIMIT = 3, //!< How long did the board limit cause the GPU to be below application clocks + NVML_PERF_POLICY_LOW_UTILIZATION = 4, //!< How long did low utilization cause the GPU to be below application clocks + NVML_PERF_POLICY_RELIABILITY = 5, //!< How long did the board reliability limit cause the GPU to be below application clocks + + NVML_PERF_POLICY_TOTAL_APP_CLOCKS = 10, //!< Total time the GPU was held below application clocks by any limiter (0 - 5 above) + NVML_PERF_POLICY_TOTAL_BASE_CLOCKS = 11, //!< Total time the GPU was held below base clocks + + // Keep this last + NVML_PERF_POLICY_COUNT +}nvmlPerfPolicyType_t; + +/** + * Struct to hold perf policy violation status data + */ +typedef struct nvmlViolationTime_st +{ + unsigned long long referenceTime; //!< referenceTime represents CPU timestamp in microseconds + unsigned long long violationTime; //!< violationTime in Nanoseconds +}nvmlViolationTime_t; + +#define NVML_MAX_THERMAL_SENSORS_PER_GPU 3 + +/** + * Represents the thermal sensor targets + */ +typedef enum +{ + NVML_THERMAL_TARGET_NONE = 0, + NVML_THERMAL_TARGET_GPU = 1, //!< GPU core temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_MEMORY = 2, //!< GPU memory temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_POWER_SUPPLY = 4, //!< GPU power supply temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_BOARD = 8, //!< GPU board ambient temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_VCD_BOARD = 9, //!< Visual Computing Device Board temperature requires NvVisualComputingDeviceHandle + NVML_THERMAL_TARGET_VCD_INLET = 10, //!< Visual Computing Device Inlet temperature requires NvVisualComputingDeviceHandle + NVML_THERMAL_TARGET_VCD_OUTLET = 11, //!< Visual Computing Device Outlet temperature requires NvVisualComputingDeviceHandle + + NVML_THERMAL_TARGET_ALL = 15, + NVML_THERMAL_TARGET_UNKNOWN = -1, +} nvmlThermalTarget_t; + +/** + * Represents the thermal sensor controllers + */ +typedef enum +{ + NVML_THERMAL_CONTROLLER_NONE = 0, + NVML_THERMAL_CONTROLLER_GPU_INTERNAL, + NVML_THERMAL_CONTROLLER_ADM1032, + NVML_THERMAL_CONTROLLER_ADT7461, + NVML_THERMAL_CONTROLLER_MAX6649, + NVML_THERMAL_CONTROLLER_MAX1617, + NVML_THERMAL_CONTROLLER_LM99, + NVML_THERMAL_CONTROLLER_LM89, + NVML_THERMAL_CONTROLLER_LM64, + NVML_THERMAL_CONTROLLER_G781, + NVML_THERMAL_CONTROLLER_ADT7473, + NVML_THERMAL_CONTROLLER_SBMAX6649, + NVML_THERMAL_CONTROLLER_VBIOSEVT, + NVML_THERMAL_CONTROLLER_OS, + NVML_THERMAL_CONTROLLER_NVSYSCON_CANOAS, + NVML_THERMAL_CONTROLLER_NVSYSCON_E551, + NVML_THERMAL_CONTROLLER_MAX6649R, + NVML_THERMAL_CONTROLLER_ADT7473S, + NVML_THERMAL_CONTROLLER_UNKNOWN = -1, +} nvmlThermalController_t; + +/** + * Struct to hold the thermal sensor settings + */ +typedef struct +{ + unsigned int count; + struct + { + nvmlThermalController_t controller; + int defaultMinTemp; + int defaultMaxTemp; + int currentTemp; + nvmlThermalTarget_t target; + } sensor[NVML_MAX_THERMAL_SENSORS_PER_GPU]; + +} nvmlGpuThermalSettings_t; + +/** + * Cooler control type + */ +typedef enum nvmlCoolerControl_enum +{ + NVML_THERMAL_COOLER_SIGNAL_NONE = 0, //!< This cooler has no control signal. + NVML_THERMAL_COOLER_SIGNAL_TOGGLE = 1, //!< This cooler can only be toggled either ON or OFF (eg a switch). + NVML_THERMAL_COOLER_SIGNAL_VARIABLE = 2, //!< This cooler's level can be adjusted from some minimum to some maximum (eg a knob). + + // Keep this last + NVML_THERMAL_COOLER_SIGNAL_COUNT +} nvmlCoolerControl_t; + +/** + * Cooler's target + */ +typedef enum nvmlCoolerTarget_enum +{ + NVML_THERMAL_COOLER_TARGET_NONE = 1 << 0, //!< This cooler cools nothing. + NVML_THERMAL_COOLER_TARGET_GPU = 1 << 1, //!< This cooler can cool the GPU. + NVML_THERMAL_COOLER_TARGET_MEMORY = 1 << 2, //!< This cooler can cool the memory. + NVML_THERMAL_COOLER_TARGET_POWER_SUPPLY = 1 << 3, //!< This cooler can cool the power supply. + NVML_THERMAL_COOLER_TARGET_GPU_RELATED = (NVML_THERMAL_COOLER_TARGET_GPU | NVML_THERMAL_COOLER_TARGET_MEMORY | NVML_THERMAL_COOLER_TARGET_POWER_SUPPLY) //!< This cooler cools all of the components related to its target gpu. GPU_RELATED = GPU | MEMORY | POWER_SUPPLY +} nvmlCoolerTarget_t; + +typedef struct +{ + unsigned int version; //!< the API version number + unsigned int index; //!< the cooler index + nvmlCoolerControl_t signalType; //!< OUT: the cooler's control signal characteristics + nvmlCoolerTarget_t target; //!< OUT: the target that cooler cools +} nvmlCoolerInfo_v1_t; +typedef nvmlCoolerInfo_v1_t nvmlCoolerInfo_t; + +#define nvmlCoolerInfo_v1 NVML_STRUCT_VERSION(CoolerInfo, 1) + +/** + * UUID length in ASCII format + */ +#define NVML_DEVICE_UUID_ASCII_LEN 41 + +/** + * UUID length in binary format + */ +#define NVML_DEVICE_UUID_BINARY_LEN 16 + +/** + * Enum to represent different UUID types + */ +typedef enum +{ + NVML_UUID_TYPE_NONE = 0, //!< Undefined type + NVML_UUID_TYPE_ASCII = 1, //!< ASCII format type + NVML_UUID_TYPE_BINARY = 2, //!< Binary format type +} nvmlUUIDType_t; + +/** + * Union to represent different UUID values + */ +typedef union +{ + char str[NVML_DEVICE_UUID_ASCII_LEN]; //!< ASCII format value + unsigned char bytes[NVML_DEVICE_UUID_BINARY_LEN]; //!< Binary format value +} nvmlUUIDValue_t; + +/** + * Struct to represent NVML UUID information + */ +typedef struct +{ + unsigned int version; //!< API version number + unsigned int type; //!< One of \p nvmlUUIDType_t + nvmlUUIDValue_t value; //!< One of \p nvmlUUIDValue_t, to be set based on the UUID format +} nvmlUUID_v1_t; +typedef nvmlUUID_v1_t nvmlUUID_t; + +#define nvmlUUID_v1 NVML_STRUCT_VERSION(UUID, 1) + +/** + * Struct to represent the NVML PDI information + */ +typedef struct +{ + unsigned int version; //!< API version number + unsigned long long value; //!< 64-bit PDI value +} nvmlPdi_v1_t; +typedef nvmlPdi_v1_t nvmlPdi_t; + +#define nvmlPdi_v1 NVML_STRUCT_VERSION(Pdi, 1) + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlDeviceEnumvs Device Enums + * @{ + */ +/***************************************************************************************************/ + +/** + * Generic enable/disable enum. + */ +typedef enum nvmlEnableState_enum +{ + NVML_FEATURE_DISABLED = 0, //!< Feature disabled + NVML_FEATURE_ENABLED = 1 //!< Feature enabled +} nvmlEnableState_t; + +//! Generic flag used to specify the default behavior of some functions. See description of particular functions for details. +#define nvmlFlagDefault 0x00 +//! Generic flag used to force some behavior. See description of particular functions for details. +#define nvmlFlagForce 0x01 + +/** + * DRAM Encryption Info + */ +typedef struct +{ + unsigned int version; //!< IN - the API version number + nvmlEnableState_t encryptionState; //!< IN/OUT - DRAM Encryption state +} nvmlDramEncryptionInfo_v1_t; +typedef nvmlDramEncryptionInfo_v1_t nvmlDramEncryptionInfo_t; + +#define nvmlDramEncryptionInfo_v1 NVML_STRUCT_VERSION(DramEncryptionInfo, 1) + +/** + * * The Brand of the GPU + * */ +typedef enum nvmlBrandType_enum +{ + NVML_BRAND_UNKNOWN = 0, + NVML_BRAND_QUADRO = 1, + NVML_BRAND_TESLA = 2, + NVML_BRAND_NVS = 3, + NVML_BRAND_GRID = 4, // Deprecated from API reporting. Keeping definition for backward compatibility. + NVML_BRAND_GEFORCE = 5, + NVML_BRAND_TITAN = 6, + NVML_BRAND_NVIDIA_VAPPS = 7, // NVIDIA Virtual Applications + NVML_BRAND_NVIDIA_VPC = 8, // NVIDIA Virtual PC + NVML_BRAND_NVIDIA_VCS = 9, // NVIDIA Virtual Compute Server + NVML_BRAND_NVIDIA_VWS = 10, // NVIDIA RTX Virtual Workstation + NVML_BRAND_NVIDIA_CLOUD_GAMING = 11, // NVIDIA Cloud Gaming + NVML_BRAND_NVIDIA_VGAMING = NVML_BRAND_NVIDIA_CLOUD_GAMING, // Deprecated from API reporting. Keeping definition for backward compatibility. + NVML_BRAND_QUADRO_RTX = 12, + NVML_BRAND_NVIDIA_RTX = 13, + NVML_BRAND_NVIDIA = 14, + NVML_BRAND_GEFORCE_RTX = 15, // Unused + NVML_BRAND_TITAN_RTX = 16, // Unused + // Keep this last + NVML_BRAND_COUNT = 18, +} nvmlBrandType_t; + +/** + * Temperature thresholds. + */ +typedef enum nvmlTemperatureThresholds_enum +{ + NVML_TEMPERATURE_THRESHOLD_SHUTDOWN = 0, // Temperature at which the GPU will + // shut down for HW protection + NVML_TEMPERATURE_THRESHOLD_SLOWDOWN = 1, // Temperature at which the GPU will + // begin HW slowdown + NVML_TEMPERATURE_THRESHOLD_MEM_MAX = 2, // Memory Temperature at which the GPU will + // begin SW slowdown + NVML_TEMPERATURE_THRESHOLD_GPU_MAX = 3, // GPU Temperature at which the GPU + // can be throttled below base clock + NVML_TEMPERATURE_THRESHOLD_ACOUSTIC_MIN = 4, // Minimum GPU Temperature that can be + // set as acoustic threshold + NVML_TEMPERATURE_THRESHOLD_ACOUSTIC_CURR = 5, // Current temperature that is set as + // acoustic threshold. + NVML_TEMPERATURE_THRESHOLD_ACOUSTIC_MAX = 6, // Maximum GPU temperature that can be + // set as acoustic threshold. + NVML_TEMPERATURE_THRESHOLD_GPS_CURR = 7, // Current temperature that is set as + // gps threshold. + // Keep this last + NVML_TEMPERATURE_THRESHOLD_COUNT +} nvmlTemperatureThresholds_t; + +/** + * Temperature sensors. + */ +typedef enum nvmlTemperatureSensors_enum +{ + NVML_TEMPERATURE_GPU = 0, //!< Temperature sensor for the GPU die + + // Keep this last + NVML_TEMPERATURE_COUNT +} nvmlTemperatureSensors_t; + +/** + * Margin temperature values + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + int marginTemperature; //!< The margin temperature value +} nvmlMarginTemperature_v1_t; + +typedef nvmlMarginTemperature_v1_t nvmlMarginTemperature_t; + +#define nvmlMarginTemperature_v1 NVML_STRUCT_VERSION(MarginTemperature, 1) + +/** + * Compute mode. + * + * NVML_COMPUTEMODE_EXCLUSIVE_PROCESS was added in CUDA 4.0. + * Earlier CUDA versions supported a single exclusive mode, + * which is equivalent to NVML_COMPUTEMODE_EXCLUSIVE_THREAD in CUDA 4.0 and beyond. + */ +typedef enum nvmlComputeMode_enum +{ + NVML_COMPUTEMODE_DEFAULT = 0, //!< Default compute mode -- multiple contexts per device + NVML_COMPUTEMODE_EXCLUSIVE_THREAD = 1, //!< Support Removed + NVML_COMPUTEMODE_PROHIBITED = 2, //!< Compute-prohibited mode -- no contexts per device + NVML_COMPUTEMODE_EXCLUSIVE_PROCESS = 3, //!< Compute-exclusive-process mode -- only one context per device, usable from multiple threads at a time + + // Keep this last + NVML_COMPUTEMODE_COUNT +} nvmlComputeMode_t; + +/** + * Max Clock Monitors available + */ +#define MAX_CLK_DOMAINS 32 + +/** + * Clock Monitor error types + */ +typedef struct nvmlClkMonFaultInfo_struct { + /** + * The Domain which faulted + */ + unsigned int clkApiDomain; + + /** + * Faults Information + */ + unsigned int clkDomainFaultMask; +} nvmlClkMonFaultInfo_t; + +/** + * Clock Monitor Status + */ +typedef struct nvmlClkMonStatus_status { + /** + * Fault status Indicator + */ + unsigned int bGlobalStatus; + + /** + * Total faulted domain numbers + */ + unsigned int clkMonListSize; + + /** + * The fault Information structure + */ + nvmlClkMonFaultInfo_t clkMonList[MAX_CLK_DOMAINS]; +} nvmlClkMonStatus_t; + +/** + * ECC bit types. + * + * @deprecated See \ref nvmlMemoryErrorType_t for a more flexible type + */ +#define nvmlEccBitType_t nvmlMemoryErrorType_t + +/** + * Single bit ECC errors + * + * @deprecated Mapped to \ref NVML_MEMORY_ERROR_TYPE_CORRECTED + */ +#define NVML_SINGLE_BIT_ECC NVML_MEMORY_ERROR_TYPE_CORRECTED + +/** + * Double bit ECC errors + * + * @deprecated Mapped to \ref NVML_MEMORY_ERROR_TYPE_UNCORRECTED + */ +#define NVML_DOUBLE_BIT_ECC NVML_MEMORY_ERROR_TYPE_UNCORRECTED + +/** + * Memory error types + */ +typedef enum nvmlMemoryErrorType_enum +{ + /** + * A memory error that was corrected + * + * For ECC errors, these are single bit errors + * For Texture memory, these are errors fixed by resend + */ + NVML_MEMORY_ERROR_TYPE_CORRECTED = 0, + /** + * A memory error that was not corrected + * + * For ECC errors, these are double bit errors + * For Texture memory, these are errors where the resend fails + */ + NVML_MEMORY_ERROR_TYPE_UNCORRECTED = 1, + + // Keep this last + NVML_MEMORY_ERROR_TYPE_COUNT //!< Count of memory error types + +} nvmlMemoryErrorType_t; + +/** + * Represents Nvlink Version + */ +typedef enum nvmlNvlinkVersion_enum +{ + NVML_NVLINK_VERSION_INVALID = 0, + NVML_NVLINK_VERSION_1_0 = 1, + NVML_NVLINK_VERSION_2_0 = 2, + NVML_NVLINK_VERSION_2_2 = 3, + NVML_NVLINK_VERSION_3_0 = 4, + NVML_NVLINK_VERSION_3_1 = 5, + NVML_NVLINK_VERSION_4_0 = 6, + NVML_NVLINK_VERSION_5_0 = 7, +}nvmlNvlinkVersion_t; + +/** + * ECC counter types. + * + * Note: Volatile counts are reset each time the driver loads. On Windows this is once per boot. On Linux this can be more frequent. + * On Linux the driver unloads when no active clients exist. If persistence mode is enabled or there is always a driver + * client active (e.g. X11), then Linux also sees per-boot behavior. If not, volatile counts are reset each time a compute app + * is run. + */ +typedef enum nvmlEccCounterType_enum +{ + NVML_VOLATILE_ECC = 0, //!< Volatile counts are reset each time the driver loads. + NVML_AGGREGATE_ECC = 1, //!< Aggregate counts persist across reboots (i.e. for the lifetime of the device) + + // Keep this last + NVML_ECC_COUNTER_TYPE_COUNT //!< Count of memory counter types +} nvmlEccCounterType_t; + +/** + * Clock types. + * + * All speeds are in Mhz. + */ +typedef enum nvmlClockType_enum +{ + NVML_CLOCK_GRAPHICS = 0, //!< Graphics clock domain + NVML_CLOCK_SM = 1, //!< SM clock domain + NVML_CLOCK_MEM = 2, //!< Memory clock domain + NVML_CLOCK_VIDEO = 3, //!< Video encoder/decoder clock domain + + // Keep this last + NVML_CLOCK_COUNT //!< Count of clock types +} nvmlClockType_t; + +/** + * Clock Ids. These are used in combination with nvmlClockType_t + * to specify a single clock value. + */ +typedef enum nvmlClockId_enum +{ + NVML_CLOCK_ID_CURRENT = 0, //!< Current actual clock value + NVML_CLOCK_ID_APP_CLOCK_TARGET = 1, //!< Target application clock. + //!< Deprecated, do not use. + NVML_CLOCK_ID_APP_CLOCK_DEFAULT = 2, //!< Default application clock target + //!< Deprecated, do not use. + NVML_CLOCK_ID_CUSTOMER_BOOST_MAX = 3, //!< OEM-defined maximum clock rate + + //Keep this last + NVML_CLOCK_ID_COUNT //!< Count of Clock Ids. +} nvmlClockId_t; + +/** + * Driver models. + * + * Windows only. + */ + +typedef enum nvmlDriverModel_enum +{ + NVML_DRIVER_WDDM = 0, //!< WDDM driver model -- GPU treated as a display device + NVML_DRIVER_WDM = 1, //!< WDM (TCC) model (deprecated) -- GPU treated as a generic compute device + NVML_DRIVER_MCDM = 2 //!< MCDM driver model -- GPU treated as a Microsoft compute device +} nvmlDriverModel_t; + +#define NVML_MAX_GPU_PERF_PSTATES 16 + +/** + * Allowed PStates. + */ +typedef enum nvmlPStates_enum +{ + NVML_PSTATE_0 = 0, //!< Performance state 0 -- Maximum Performance + NVML_PSTATE_1 = 1, //!< Performance state 1 + NVML_PSTATE_2 = 2, //!< Performance state 2 + NVML_PSTATE_3 = 3, //!< Performance state 3 + NVML_PSTATE_4 = 4, //!< Performance state 4 + NVML_PSTATE_5 = 5, //!< Performance state 5 + NVML_PSTATE_6 = 6, //!< Performance state 6 + NVML_PSTATE_7 = 7, //!< Performance state 7 + NVML_PSTATE_8 = 8, //!< Performance state 8 + NVML_PSTATE_9 = 9, //!< Performance state 9 + NVML_PSTATE_10 = 10, //!< Performance state 10 + NVML_PSTATE_11 = 11, //!< Performance state 11 + NVML_PSTATE_12 = 12, //!< Performance state 12 + NVML_PSTATE_13 = 13, //!< Performance state 13 + NVML_PSTATE_14 = 14, //!< Performance state 14 + NVML_PSTATE_15 = 15, //!< Performance state 15 -- Minimum Performance + NVML_PSTATE_UNKNOWN = 32 //!< Unknown performance state +} nvmlPstates_t; + +/** + * Clock offset info. + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + nvmlClockType_t type; + nvmlPstates_t pstate; + int clockOffsetMHz; + int minClockOffsetMHz; + int maxClockOffsetMHz; +} nvmlClockOffset_v1_t; + +typedef nvmlClockOffset_v1_t nvmlClockOffset_t; + +#define nvmlClockOffset_v1 NVML_STRUCT_VERSION(ClockOffset, 1) + +/** + * Fan speed info. + */ +typedef struct +{ + unsigned int version; //!< the API version number + unsigned int fan; //!< the fan index + unsigned int speed; //!< OUT: the fan speed in RPM +} nvmlFanSpeedInfo_v1_t; +typedef nvmlFanSpeedInfo_v1_t nvmlFanSpeedInfo_t; + +#define nvmlFanSpeedInfo_v1 NVML_STRUCT_VERSION(FanSpeedInfo, 1) + +#define NVML_PERF_MODES_BUFFER_SIZE 2048 + +/** + * Device performance modes string + */ +typedef struct +{ + unsigned int version; //!< the API version number + char str[NVML_PERF_MODES_BUFFER_SIZE]; //!< OUT: the performance modes string. +} nvmlDevicePerfModes_v1_t; +typedef nvmlDevicePerfModes_v1_t nvmlDevicePerfModes_t; + +#define nvmlDevicePerfModes_v1 NVML_STRUCT_VERSION(DevicePerfModes, 1) + +/** + * Device current clocks string + */ +typedef struct +{ + unsigned int version; //!< the API version number + char str[NVML_PERF_MODES_BUFFER_SIZE]; //!< OUT: the current clock frequency string. +} nvmlDeviceCurrentClockFreqs_v1_t; +typedef nvmlDeviceCurrentClockFreqs_v1_t nvmlDeviceCurrentClockFreqs_t; + +#define nvmlDeviceCurrentClockFreqs_v1 NVML_STRUCT_VERSION(DeviceCurrentClockFreqs, 1) + +/** + * Device powerMizer modes + */ +#define NVML_POWER_MIZER_MODE_ADAPTIVE 0 //!< adjust GPU clocks based on GPU utilization +#define NVML_POWER_MIZER_MODE_PREFER_MAXIMUM_PERFORMANCE 1 //!< raise GPU clocks to favor maximum performance, + //!< to the extent that thermal and other constraints allow +#define NVML_POWER_MIZER_MODE_AUTO 2 //!< PowerMizer mode is driver controlled. +#define NVML_POWER_MIZER_MODE_PREFER_CONSISTENT_PERFORMANCE 3 //!< lock to GPU base clocks + +typedef struct +{ + unsigned int currentMode; //!< OUT: the current powermizer mode + unsigned int mode; //!< IN: the powermizer mode to set + unsigned int supportedPowerMizerModes; //!< OUT: Bitmask of supported powermizer modes +} nvmlDevicePowerMizerModes_v1_t; + +/** + * GPU Operation Mode + * + * GOM allows to reduce power usage and optimize GPU throughput by disabling GPU features. + * + * Each GOM is designed to meet specific user needs. + */ +typedef enum nvmlGom_enum +{ + NVML_GOM_ALL_ON = 0, //!< Everything is enabled and running at full speed + + NVML_GOM_COMPUTE = 1, //!< Designed for running only compute tasks. Graphics operations + //!< are not allowed + + NVML_GOM_LOW_DP = 2 //!< Designed for running graphics applications that don't require + //!< high bandwidth double precision +} nvmlGpuOperationMode_t; + +/** + * Available infoROM objects. + */ +typedef enum nvmlInforomObject_enum +{ + NVML_INFOROM_OEM = 0, //!< An object defined by OEM + NVML_INFOROM_ECC = 1, //!< The ECC object determining the level of ECC support + NVML_INFOROM_POWER = 2, //!< The power management object + NVML_INFOROM_DEN = 3, //!< DRAM Encryption object + // Keep this last + NVML_INFOROM_COUNT //!< This counts the number of infoROM objects the driver knows about +} nvmlInforomObject_t; + +/** + * Return values for NVML API calls. + */ +typedef enum nvmlReturn_enum +{ + // cppcheck-suppress * + NVML_SUCCESS = 0, //!< The operation was successful + NVML_ERROR_UNINITIALIZED = 1, //!< NVML was not first initialized with nvmlInit() + NVML_ERROR_INVALID_ARGUMENT = 2, //!< A supplied argument is invalid + NVML_ERROR_NOT_SUPPORTED = 3, //!< The requested operation is not available on target device + NVML_ERROR_NO_PERMISSION = 4, //!< The current user does not have permission for operation + NVML_ERROR_ALREADY_INITIALIZED = 5, //!< Deprecated: Multiple initializations are now allowed through ref counting + NVML_ERROR_NOT_FOUND = 6, //!< A query to find an object was unsuccessful + NVML_ERROR_INSUFFICIENT_SIZE = 7, //!< An input argument is not large enough + NVML_ERROR_INSUFFICIENT_POWER = 8, //!< A device's external power cables are not properly attached + NVML_ERROR_DRIVER_NOT_LOADED = 9, //!< NVIDIA driver is not loaded + NVML_ERROR_TIMEOUT = 10, //!< User provided timeout passed + NVML_ERROR_IRQ_ISSUE = 11, //!< NVIDIA Kernel detected an interrupt issue with a GPU + NVML_ERROR_LIBRARY_NOT_FOUND = 12, //!< NVML Shared Library couldn't be found or loaded + NVML_ERROR_FUNCTION_NOT_FOUND = 13, //!< Local version of NVML doesn't implement this function + NVML_ERROR_CORRUPTED_INFOROM = 14, //!< infoROM is corrupted + NVML_ERROR_GPU_IS_LOST = 15, //!< The GPU has fallen off the bus or has otherwise become inaccessible + NVML_ERROR_RESET_REQUIRED = 16, //!< The GPU requires a reset before it can be used again + NVML_ERROR_OPERATING_SYSTEM = 17, //!< The GPU control device has been blocked by the operating system/cgroups + NVML_ERROR_LIB_RM_VERSION_MISMATCH = 18, //!< RM detects a driver/library version mismatch + NVML_ERROR_IN_USE = 19, //!< An operation cannot be performed because the GPU is currently in use + NVML_ERROR_MEMORY = 20, //!< Insufficient memory + NVML_ERROR_NO_DATA = 21, //!< No data + NVML_ERROR_VGPU_ECC_NOT_SUPPORTED = 22, //!< The requested vgpu operation is not available on target device, becasue ECC is enabled + NVML_ERROR_INSUFFICIENT_RESOURCES = 23, //!< Ran out of critical resources, other than memory + NVML_ERROR_FREQ_NOT_SUPPORTED = 24, //!< Ran out of critical resources, other than memory + NVML_ERROR_ARGUMENT_VERSION_MISMATCH = 25, //!< The provided version is invalid/unsupported + NVML_ERROR_DEPRECATED = 26, //!< The requested functionality has been deprecated + NVML_ERROR_NOT_READY = 27, //!< The system is not ready for the request + NVML_ERROR_GPU_NOT_FOUND = 28, //!< No GPUs were found + NVML_ERROR_INVALID_STATE = 29, //!< Resource not in correct state to perform requested operation + NVML_ERROR_RESET_TYPE_NOT_SUPPORTED = 30, //!< Reset not supported for given device/parameters + NVML_ERROR_UNKNOWN = 999 //!< An internal driver error occurred +} nvmlReturn_t; + +/** + * See \ref nvmlDeviceGetMemoryErrorCounter + */ +typedef enum nvmlMemoryLocation_enum +{ + NVML_MEMORY_LOCATION_L1_CACHE = 0, //!< GPU L1 Cache + NVML_MEMORY_LOCATION_L2_CACHE = 1, //!< GPU L2 Cache + NVML_MEMORY_LOCATION_DRAM = 2, //!< Turing+ DRAM + NVML_MEMORY_LOCATION_DEVICE_MEMORY = 2, //!< GPU Device Memory + NVML_MEMORY_LOCATION_REGISTER_FILE = 3, //!< GPU Register File + NVML_MEMORY_LOCATION_TEXTURE_MEMORY = 4, //!< GPU Texture Memory + NVML_MEMORY_LOCATION_TEXTURE_SHM = 5, //!< Shared memory + NVML_MEMORY_LOCATION_CBU = 6, //!< CBU + NVML_MEMORY_LOCATION_SRAM = 7, //!< Turing+ SRAM + // Keep this last + NVML_MEMORY_LOCATION_COUNT //!< This counts the number of memory locations the driver knows about +} nvmlMemoryLocation_t; + +/** + * Causes for page retirement + */ +typedef enum nvmlPageRetirementCause_enum +{ + NVML_PAGE_RETIREMENT_CAUSE_MULTIPLE_SINGLE_BIT_ECC_ERRORS = 0, //!< Page was retired due to multiple single bit ECC error + NVML_PAGE_RETIREMENT_CAUSE_DOUBLE_BIT_ECC_ERROR = 1, //!< Page was retired due to double bit ECC error + + // Keep this last + NVML_PAGE_RETIREMENT_CAUSE_COUNT +} nvmlPageRetirementCause_t; + +/** + * API types that allow changes to default permission restrictions + */ +typedef enum nvmlRestrictedAPI_enum +{ + NVML_RESTRICTED_API_SET_APPLICATION_CLOCKS = 0, //!< APIs that change application clocks, see nvmlDeviceSetApplicationsClocks + //!< and see nvmlDeviceResetApplicationsClocks. + //!< Deprecated, keeping definition for backward compatibility. + NVML_RESTRICTED_API_SET_AUTO_BOOSTED_CLOCKS = 1, //!< APIs that enable/disable Auto Boosted clocks + //!< see nvmlDeviceSetAutoBoostedClocksEnabled + // Keep this last + NVML_RESTRICTED_API_COUNT +} nvmlRestrictedAPI_t; + +/** + * Structure to store utilization value and process Id + */ +typedef struct nvmlProcessUtilizationSample_st +{ + unsigned int pid; //!< PID of process + unsigned long long timeStamp; //!< CPU Timestamp in microseconds + unsigned int smUtil; //!< SM (3D/Compute) Util Value + unsigned int memUtil; //!< Frame Buffer Memory Util Value + unsigned int encUtil; //!< Encoder Util Value + unsigned int decUtil; //!< Decoder Util Value +} nvmlProcessUtilizationSample_t; + +/** + * Structure to store utilization value and process Id -- version 1 + */ +typedef struct +{ + unsigned long long timeStamp; //!< CPU Timestamp in microseconds + unsigned int pid; //!< PID of process + unsigned int smUtil; //!< SM (3D/Compute) Util Value + unsigned int memUtil; //!< Frame Buffer Memory Util Value + unsigned int encUtil; //!< Encoder Util Value + unsigned int decUtil; //!< Decoder Util Value + unsigned int jpgUtil; //!< Jpeg Util Value + unsigned int ofaUtil; //!< Ofa Util Value +} nvmlProcessUtilizationInfo_v1_t; + +/** + * Structure to store utilization and process ID for each running process -- version 1 + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + unsigned int processSamplesCount; //!< Caller-supplied array size, and returns number of processes running + unsigned long long lastSeenTimeStamp; //!< Return only samples with timestamp greater than lastSeenTimeStamp + nvmlProcessUtilizationInfo_v1_t *procUtilArray; //!< The array (allocated by caller) of the utilization of GPU SM, framebuffer, video encoder, video decoder, JPEG, and OFA +} nvmlProcessesUtilizationInfo_v1_t; +typedef nvmlProcessesUtilizationInfo_v1_t nvmlProcessesUtilizationInfo_t; +#define nvmlProcessesUtilizationInfo_v1 NVML_STRUCT_VERSION(ProcessesUtilizationInfo, 1) + +/** + * Structure to store SRAM uncorrectable error counters + */ +typedef struct +{ + unsigned int version; //!< the API version number + unsigned long long aggregateUncParity; //!< aggregate uncorrectable parity error count + unsigned long long aggregateUncSecDed; //!< aggregate uncorrectable SEC-DED error count + unsigned long long aggregateCor; //!< aggregate correctable error count + unsigned long long volatileUncParity; //!< volatile uncorrectable parity error count + unsigned long long volatileUncSecDed; //!< volatile uncorrectable SEC-DED error count + unsigned long long volatileCor; //!< volatile correctable error count + unsigned long long aggregateUncBucketL2; //!< aggregate uncorrectable error count for L2 cache bucket + unsigned long long aggregateUncBucketSm; //!< aggregate uncorrectable error count for SM bucket + unsigned long long aggregateUncBucketPcie; //!< aggregate uncorrectable error count for PCIE bucket + unsigned long long aggregateUncBucketMcu; //!< aggregate uncorrectable error count for Microcontroller bucket + unsigned long long aggregateUncBucketOther; //!< aggregate uncorrectable error count for Other bucket + unsigned int bThresholdExceeded; //!< if the error threshold of field diag is exceeded +} nvmlEccSramErrorStatus_v1_t; + +typedef nvmlEccSramErrorStatus_v1_t nvmlEccSramErrorStatus_t; +#define nvmlEccSramErrorStatus_v1 NVML_STRUCT_VERSION(EccSramErrorStatus, 1) + +/** + * Structure to store platform information + * + * @deprecated The nvmlPlatformInfo_v1_t will be deprecated in the subsequent releases. + * Use nvmlPlatformInfo_v2_t + */ +typedef struct +{ + unsigned int version; //!< the API version number + unsigned char ibGuid[16]; //!< Infiniband GUID reported by platform (for Blackwell, ibGuid is 8 bytes so indices 8-15 are zero) + unsigned char rackGuid[16]; //!< GUID of the rack containing this GPU (for Blackwell rackGuid is 13 bytes so indices 13-15 are zero) + unsigned char chassisPhysicalSlotNumber; //!< The slot number in the rack containing this GPU (includes switches) + unsigned char computeSlotIndex; //!< The index within the compute slots in the rack containing this GPU (does not include switches) + unsigned char nodeIndex; //!< Index of the node within the slot containing this GPU + unsigned char peerType; //!< Platform indicated NVLink-peer type (e.g. switch present or not) + unsigned char moduleId; //!< ID of this GPU within the node +} nvmlPlatformInfo_v1_t; +#define nvmlPlatformInfo_v1 NVML_STRUCT_VERSION(PlatformInfo, 1) + +/** + * Structure to store platform information (v2) + */ +typedef struct +{ + unsigned int version; //!< the API version number + unsigned char ibGuid[16]; //!< Infiniband GUID reported by platform (for Blackwell, ibGuid is 8 bytes so indices 8-15 are zero) + unsigned char chassisSerialNumber[16]; //!< Serial number of the chassis containing this GPU (for Blackwell it is 13 bytes so indices 13-15 are zero) + unsigned char slotNumber; //!< The slot number in the chassis containing this GPU (includes switches) + unsigned char trayIndex; //!< The tray index within the compute slots in the chassis containing this GPU (does not include switches) + unsigned char hostId; //!< Index of the node within the slot containing this GPU + unsigned char peerType; //!< Platform indicated NVLink-peer type (e.g. switch present or not) + unsigned char moduleId; //!< ID of this GPU within the node +} nvmlPlatformInfo_v2_t; + +typedef nvmlPlatformInfo_v2_t nvmlPlatformInfo_t; +#define nvmlPlatformInfo_v2 NVML_STRUCT_VERSION(PlatformInfo, 2) + +typedef struct +{ + unsigned int unit; //!< the SRAM unit index + unsigned int location; //!< the error location within the SRAM unit + unsigned int sublocation; //!< the error sublocation within the SRAM unit + unsigned int extlocation; //!< the error extlocation within the SRAM unit + unsigned int address; //!< the error address within the SRAM unit + unsigned int isParity; //!< if the SRAM error is parity or not + unsigned int count; //!< the error count at the same SRAM address +} nvmlEccSramUniqueUncorrectedErrorEntry_v1_t; + +typedef struct +{ + unsigned int version; //!< the API version number + unsigned int entryCount; //!< the number of error count entries + nvmlEccSramUniqueUncorrectedErrorEntry_v1_t *entries; //!< pointer to caller-supplied buffer to return the SRAM unique uncorrected ECC error count entries +} nvmlEccSramUniqueUncorrectedErrorCounts_v1_t; + +typedef nvmlEccSramUniqueUncorrectedErrorCounts_v1_t nvmlEccSramUniqueUncorrectedErrorCounts_t; +#define nvmlEccSramUniqueUncorrectedErrorCounts_v1 NVML_STRUCT_VERSION(EccSramUniqueUncorrectedErrorCounts, 1) + +/** + * GSP firmware + */ +#define NVML_GSP_FIRMWARE_VERSION_BUF_SIZE 0x40 + +/** + * Simplified chip architecture + */ +#define NVML_DEVICE_ARCH_KEPLER 2 // Devices based on the NVIDIA Kepler architecture +#define NVML_DEVICE_ARCH_MAXWELL 3 // Devices based on the NVIDIA Maxwell architecture +#define NVML_DEVICE_ARCH_PASCAL 4 // Devices based on the NVIDIA Pascal architecture +#define NVML_DEVICE_ARCH_VOLTA 5 // Devices based on the NVIDIA Volta architecture +#define NVML_DEVICE_ARCH_TURING 6 // Devices based on the NVIDIA Turing architecture +#define NVML_DEVICE_ARCH_AMPERE 7 // Devices based on the NVIDIA Ampere architecture +#define NVML_DEVICE_ARCH_ADA 8 // Devices based on the NVIDIA Ada architecture +#define NVML_DEVICE_ARCH_HOPPER 9 // Devices based on the NVIDIA Hopper architecture + +#define NVML_DEVICE_ARCH_BLACKWELL 10 // Devices based on the NVIDIA Blackwell architecture + +#define NVML_DEVICE_ARCH_UNKNOWN 0xffffffff // Anything else, presumably something newer + +typedef unsigned int nvmlDeviceArchitecture_t; + +/** + * PCI bus types + */ +#define NVML_BUS_TYPE_UNKNOWN 0 +#define NVML_BUS_TYPE_PCI 1 +#define NVML_BUS_TYPE_PCIE 2 +#define NVML_BUS_TYPE_FPCI 3 +#define NVML_BUS_TYPE_AGP 4 + +typedef unsigned int nvmlBusType_t; + +/** + * Device Power Modes + */ + +/** + * Device Fan control policy + */ +#define NVML_FAN_POLICY_TEMPERATURE_CONTINOUS_SW 0 +#define NVML_FAN_POLICY_MANUAL 1 + +typedef unsigned int nvmlFanControlPolicy_t; + +/** + * Device Power Source + */ +#define NVML_POWER_SOURCE_AC 0x00000000 +#define NVML_POWER_SOURCE_BATTERY 0x00000001 +#define NVML_POWER_SOURCE_UNDERSIZED 0x00000002 + +typedef unsigned int nvmlPowerSource_t; + +/** + * Device PCIE link Max Speed + */ +#define NVML_PCIE_LINK_MAX_SPEED_INVALID 0x00000000 +#define NVML_PCIE_LINK_MAX_SPEED_2500MBPS 0x00000001 +#define NVML_PCIE_LINK_MAX_SPEED_5000MBPS 0x00000002 +#define NVML_PCIE_LINK_MAX_SPEED_8000MBPS 0x00000003 +#define NVML_PCIE_LINK_MAX_SPEED_16000MBPS 0x00000004 +#define NVML_PCIE_LINK_MAX_SPEED_32000MBPS 0x00000005 +#define NVML_PCIE_LINK_MAX_SPEED_64000MBPS 0x00000006 + +/** + * Adaptive clocking status + */ +#define NVML_ADAPTIVE_CLOCKING_INFO_STATUS_DISABLED 0x00000000 +#define NVML_ADAPTIVE_CLOCKING_INFO_STATUS_ENABLED 0x00000001 + +#define NVML_MAX_GPU_UTILIZATIONS 8 + +/** + * Represents the GPU utilization domains + */ +typedef enum nvmlGpuUtilizationDomainId_t +{ + NVML_GPU_UTILIZATION_DOMAIN_GPU = 0, //!< Graphics engine domain + NVML_GPU_UTILIZATION_DOMAIN_FB = 1, //!< Frame buffer domain + NVML_GPU_UTILIZATION_DOMAIN_VID = 2, //!< Video engine domain + NVML_GPU_UTILIZATION_DOMAIN_BUS = 3, //!< Bus interface domain +} nvmlGpuUtilizationDomainId_t; + +typedef struct nvmlGpuDynamicPstatesInfo_st +{ + unsigned int flags; //!< Reserved for future use + struct + { + unsigned int bIsPresent; //!< Set if this utilization domain is present on this GPU + unsigned int percentage; //!< Percentage of time where the domain is considered busy in the last 1-second interval + unsigned int incThreshold; //!< Utilization threshold that can trigger a perf-increasing P-State change when crossed + unsigned int decThreshold; //!< Utilization threshold that can trigger a perf-decreasing P-State change when crossed + } utilization[NVML_MAX_GPU_UTILIZATIONS]; +} nvmlGpuDynamicPstatesInfo_t; + +/* + * PCIe outbound/inbound atomic operations capability + */ +#define NVML_PCIE_ATOMICS_CAP_FETCHADD32 0x01 +#define NVML_PCIE_ATOMICS_CAP_FETCHADD64 0x02 +#define NVML_PCIE_ATOMICS_CAP_SWAP32 0x04 +#define NVML_PCIE_ATOMICS_CAP_SWAP64 0x08 +#define NVML_PCIE_ATOMICS_CAP_CAS32 0x10 +#define NVML_PCIE_ATOMICS_CAP_CAS64 0x20 +#define NVML_PCIE_ATOMICS_CAP_CAS128 0x40 +#define NVML_PCIE_ATOMICS_OPS_MAX 7 + +/** + * Device Scope - This is useful to retrieve the telemetry at GPU and module (e.g. GPU + CPU) level + */ +#define NVML_POWER_SCOPE_GPU 0U //!< Targets only GPU +#define NVML_POWER_SCOPE_MODULE 1U //!< Targets the whole module +#define NVML_POWER_SCOPE_MEMORY 2U //!< Targets the GPU Memory + +typedef unsigned char nvmlPowerScopeType_t; + +/** + * Contains the power management limit + */ +typedef struct +{ + unsigned int version; //!< Structure format version (must be 1) + nvmlPowerScopeType_t powerScope; //!< [in] Device type: GPU or Total Module + unsigned int powerValueMw; //!< [out] Power value to retrieve or set in milliwatts +} nvmlPowerValue_v2_t; + +#define nvmlPowerValue_v2 NVML_STRUCT_VERSION(PowerValue, 2) + +/** @} */ + +/***************************************************************************************************/ +/** @addtogroup virtualGPU vGPU Enums, Constants, Structs + * @{ + */ +/***************************************************************************************************/ +/** @defgroup nvmlVirtualGpuEnums vGPU Enums + * @{ + */ +/***************************************************************************************************/ + +/*! + * GPU virtualization mode types. + */ +typedef enum nvmlGpuVirtualizationMode { + NVML_GPU_VIRTUALIZATION_MODE_NONE = 0, //!< Represents Bare Metal GPU + NVML_GPU_VIRTUALIZATION_MODE_PASSTHROUGH = 1, //!< Device is associated with GPU-Passthorugh + NVML_GPU_VIRTUALIZATION_MODE_VGPU = 2, //!< Device is associated with vGPU inside virtual machine. + NVML_GPU_VIRTUALIZATION_MODE_HOST_VGPU = 3, //!< Device is associated with VGX hypervisor in vGPU mode + NVML_GPU_VIRTUALIZATION_MODE_HOST_VSGA = 4 //!< Device is associated with VGX hypervisor in vSGA mode +} nvmlGpuVirtualizationMode_t; + +/** + * Host vGPU modes + */ +typedef enum nvmlHostVgpuMode_enum +{ + NVML_HOST_VGPU_MODE_NON_SRIOV = 0, //!< Non SR-IOV mode + NVML_HOST_VGPU_MODE_SRIOV = 1 //!< SR-IOV mode +} nvmlHostVgpuMode_t; + +/*! + * Types of VM identifiers + */ +typedef enum nvmlVgpuVmIdType { + NVML_VGPU_VM_ID_DOMAIN_ID = 0, //!< VM ID represents DOMAIN ID + NVML_VGPU_VM_ID_UUID = 1 //!< VM ID represents UUID +} nvmlVgpuVmIdType_t; + +/** + * vGPU GUEST info state + */ +typedef enum nvmlVgpuGuestInfoState_enum +{ + NVML_VGPU_INSTANCE_GUEST_INFO_STATE_UNINITIALIZED = 0, //!< Guest-dependent fields uninitialized + NVML_VGPU_INSTANCE_GUEST_INFO_STATE_INITIALIZED = 1 //!< Guest-dependent fields initialized +} nvmlVgpuGuestInfoState_t; + +/** + * vGPU software licensable features + */ +typedef enum { + NVML_GRID_LICENSE_FEATURE_CODE_UNKNOWN = 0, //!< Unknown + NVML_GRID_LICENSE_FEATURE_CODE_VGPU = 1, //!< Virtual GPU + NVML_GRID_LICENSE_FEATURE_CODE_NVIDIA_RTX = 2, //!< Nvidia RTX + NVML_GRID_LICENSE_FEATURE_CODE_VWORKSTATION = NVML_GRID_LICENSE_FEATURE_CODE_NVIDIA_RTX, //!< Deprecated, do not use. + NVML_GRID_LICENSE_FEATURE_CODE_GAMING = 3, //!< Gaming + NVML_GRID_LICENSE_FEATURE_CODE_COMPUTE = 4 //!< Compute +} nvmlGridLicenseFeatureCode_t; + +/** + * Status codes for license expiry + */ +#define NVML_GRID_LICENSE_EXPIRY_NOT_AVAILABLE 0 //!< Expiry information not available +#define NVML_GRID_LICENSE_EXPIRY_INVALID 1 //!< Invalid expiry or error fetching expiry +#define NVML_GRID_LICENSE_EXPIRY_VALID 2 //!< Valid expiry +#define NVML_GRID_LICENSE_EXPIRY_NOT_APPLICABLE 3 //!< Expiry not applicable +#define NVML_GRID_LICENSE_EXPIRY_PERMANENT 4 //!< Permanent expiry + +/** + * vGPU queryable capabilities + */ +typedef enum nvmlVgpuCapability_enum +{ + NVML_VGPU_CAP_NVLINK_P2P = 0, //!< P2P over NVLink is supported + NVML_VGPU_CAP_GPUDIRECT = 1, //!< GPUDirect capability is supported + NVML_VGPU_CAP_MULTI_VGPU_EXCLUSIVE = 2, //!< vGPU profile cannot be mixed with other vGPU profiles in same VM + NVML_VGPU_CAP_EXCLUSIVE_TYPE = 3, //!< vGPU profile cannot run on a GPU alongside other profiles of different type + NVML_VGPU_CAP_EXCLUSIVE_SIZE = 4, //!< vGPU profile cannot run on a GPU alongside other profiles of different size + // Keep this last + NVML_VGPU_CAP_COUNT +} nvmlVgpuCapability_t; + +/** +* vGPU driver queryable capabilities +*/ +typedef enum nvmlVgpuDriverCapability_enum +{ + NVML_VGPU_DRIVER_CAP_HETEROGENEOUS_MULTI_VGPU = 0, //!< Supports mixing of different vGPU profiles within one guest VM + NVML_VGPU_DRIVER_CAP_WARM_UPDATE = 1, //!< Supports FSR and warm update of vGPU host driver without terminating the running guest VM + // Keep this last + NVML_VGPU_DRIVER_CAP_COUNT +} nvmlVgpuDriverCapability_t; + +/** +* Device vGPU queryable capabilities +*/ +typedef enum nvmlDeviceVgpuCapability_enum +{ + NVML_DEVICE_VGPU_CAP_FRACTIONAL_MULTI_VGPU = 0, //!< Query whether the fractional vGPU profiles on this GPU can be used in multi-vGPU configurations + NVML_DEVICE_VGPU_CAP_HETEROGENEOUS_TIMESLICE_PROFILES = 1, //!< Query whether the GPU support concurrent execution of timesliced vGPU profiles of differing types + NVML_DEVICE_VGPU_CAP_HETEROGENEOUS_TIMESLICE_SIZES = 2, //!< Query whether the GPU support concurrent execution of timesliced vGPU profiles of differing framebuffer sizes + NVML_DEVICE_VGPU_CAP_READ_DEVICE_BUFFER_BW = 3, //!< Query the GPU's read_device_buffer expected bandwidth capacity in megabytes per second + NVML_DEVICE_VGPU_CAP_WRITE_DEVICE_BUFFER_BW = 4, //!< Query the GPU's write_device_buffer expected bandwidth capacity in megabytes per second + NVML_DEVICE_VGPU_CAP_DEVICE_STREAMING = 5, //!< Query whether the vGPU profiles on the GPU supports migration data streaming + NVML_DEVICE_VGPU_CAP_MINI_QUARTER_GPU = 6, //!< Set/Get support for mini-quarter vGPU profiles + NVML_DEVICE_VGPU_CAP_COMPUTE_MEDIA_ENGINE_GPU = 7, //!< Set/Get support for compute media engine vGPU profiles + NVML_DEVICE_VGPU_CAP_WARM_UPDATE = 8, //!< Query whether the GPU supports FSR and warm update + NVML_DEVICE_VGPU_CAP_HOMOGENEOUS_PLACEMENTS = 9, //!< Query whether the GPU supports reporting of placements of timesliced vGPU profiles with identical framebuffer sizes + NVML_DEVICE_VGPU_CAP_MIG_TIMESLICING_SUPPORTED = 10, //!< Query whether the GPU supports timesliced vGPU on MIG + NVML_DEVICE_VGPU_CAP_MIG_TIMESLICING_ENABLED = 11, //!< Set/Get MIG timesliced mode reporting, without impacting the underlying functionality + // Keep this last + NVML_DEVICE_VGPU_CAP_COUNT +} nvmlDeviceVgpuCapability_t; + +/** @} */ + +/***************************************************************************************************/ + +/** @defgroup nvmlVgpuConstants vGPU Constants + * @{ + */ +/***************************************************************************************************/ + +/** + * Buffer size guaranteed to be large enough for \ref nvmlVgpuTypeGetLicense + */ +#define NVML_GRID_LICENSE_BUFFER_SIZE 128 + +#define NVML_VGPU_NAME_BUFFER_SIZE 64 + +#define NVML_GRID_LICENSE_FEATURE_MAX_COUNT 3 + +#define INVALID_GPU_INSTANCE_PROFILE_ID 0xFFFFFFFF + +#define INVALID_GPU_INSTANCE_ID 0xFFFFFFFF + +#define NVML_INVALID_VGPU_PLACEMENT_ID 0xFFFF + +/*! + * Macros for vGPU instance's virtualization capabilities bitfield. + */ +#define NVML_VGPU_VIRTUALIZATION_CAP_MIGRATION 0:0 +#define NVML_VGPU_VIRTUALIZATION_CAP_MIGRATION_NO 0x0 +#define NVML_VGPU_VIRTUALIZATION_CAP_MIGRATION_YES 0x1 + +/*! + * Macros for pGPU's virtualization capabilities bitfield. + */ +#define NVML_VGPU_PGPU_VIRTUALIZATION_CAP_MIGRATION 0:0 +#define NVML_VGPU_PGPU_VIRTUALIZATION_CAP_MIGRATION_NO 0x0 +#define NVML_VGPU_PGPU_VIRTUALIZATION_CAP_MIGRATION_YES 0x1 + +/** + * Macros to indicate the vGPU mode of the GPU. + */ +#define NVML_VGPU_PGPU_HETEROGENEOUS_MODE 0 +#define NVML_VGPU_PGPU_HOMOGENEOUS_MODE 1 + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlVgpuStructs vGPU Structs + * @{ + */ +/***************************************************************************************************/ + +typedef unsigned int nvmlVgpuTypeId_t; + +typedef unsigned int nvmlVgpuInstance_t; + +/** + * Structure to store the vGPU heterogeneous mode of device -- version 1 + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + unsigned int mode; //!< The vGPU heterogeneous mode +} nvmlVgpuHeterogeneousMode_v1_t; +typedef nvmlVgpuHeterogeneousMode_v1_t nvmlVgpuHeterogeneousMode_t; +#define nvmlVgpuHeterogeneousMode_v1 NVML_STRUCT_VERSION(VgpuHeterogeneousMode, 1) + +/** + * Structure to store the placement ID of vGPU instance -- version 1 + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + unsigned int placementId; //!< Placement ID of the active vGPU instance +} nvmlVgpuPlacementId_v1_t; +typedef nvmlVgpuPlacementId_v1_t nvmlVgpuPlacementId_t; +#define nvmlVgpuPlacementId_v1 NVML_STRUCT_VERSION(VgpuPlacementId, 1) + +/** + * Structure to store the list of vGPU placements -- version 1 + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + unsigned int placementSize; //!< The number of slots occupied by the vGPU type + unsigned int count; //!< Count of placement IDs fetched + unsigned int *placementIds; //!< Placement IDs for the vGPU type +} nvmlVgpuPlacementList_v1_t; +#define nvmlVgpuPlacementList_v1 NVML_STRUCT_VERSION(VgpuPlacementList, 1) + +/** + * Structure to store the list of vGPU placements -- version 2 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + unsigned int placementSize; //!< OUT: The number of slots occupied by the vGPU type + unsigned int count; //!< IN/OUT: Count of the placement IDs + unsigned int *placementIds; //!< IN/OUT: Placement IDs for the vGPU type + unsigned int mode; //!< IN: The vGPU mode. Either NVML_VGPU_PGPU_HETEROGENEOUS_MODE or NVML_VGPU_PGPU_HOMOGENEOUS_MODE +} nvmlVgpuPlacementList_v2_t; +typedef nvmlVgpuPlacementList_v2_t nvmlVgpuPlacementList_t; +#define nvmlVgpuPlacementList_v2 NVML_STRUCT_VERSION(VgpuPlacementList, 2) + +/** + * Structure to store BAR1 size information of vGPU type -- Version 1 + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + unsigned long long bar1Size; //!< BAR1 size in megabytes +} nvmlVgpuTypeBar1Info_v1_t; +typedef nvmlVgpuTypeBar1Info_v1_t nvmlVgpuTypeBar1Info_t; +#define nvmlVgpuTypeBar1Info_v1 NVML_STRUCT_VERSION(VgpuTypeBar1Info, 1) + +/** + * Structure to store Utilization Value and vgpuInstance + */ +typedef struct nvmlVgpuInstanceUtilizationSample_st +{ + nvmlVgpuInstance_t vgpuInstance; //!< vGPU Instance + unsigned long long timeStamp; //!< CPU Timestamp in microseconds + nvmlValue_t smUtil; //!< SM (3D/Compute) Util Value + nvmlValue_t memUtil; //!< Frame Buffer Memory Util Value + nvmlValue_t encUtil; //!< Encoder Util Value + nvmlValue_t decUtil; //!< Decoder Util Value +} nvmlVgpuInstanceUtilizationSample_t; + +/** + * Structure to store Utilization Value and vgpuInstance Info -- Version 1 + */ +typedef struct +{ + unsigned long long timeStamp; //!< CPU Timestamp in microseconds + nvmlVgpuInstance_t vgpuInstance; //!< vGPU Instance + nvmlValue_t smUtil; //!< SM (3D/Compute) Util Value + nvmlValue_t memUtil; //!< Frame Buffer Memory Util Value + nvmlValue_t encUtil; //!< Encoder Util Value + nvmlValue_t decUtil; //!< Decoder Util Value + nvmlValue_t jpgUtil; //!< Jpeg Util Value + nvmlValue_t ofaUtil; //!< Ofa Util Value +} nvmlVgpuInstanceUtilizationInfo_v1_t; + +/** + * Structure to store recent utilization for vGPU instances running on a device -- version 1 + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + nvmlValueType_t sampleValType; //!< Hold the type of returned sample values + unsigned int vgpuInstanceCount; //!< Hold the number of vGPU instances + unsigned long long lastSeenTimeStamp; //!< Return only samples with timestamp greater than lastSeenTimeStamp + nvmlVgpuInstanceUtilizationInfo_v1_t *vgpuUtilArray; //!< The array (allocated by caller) in which vGPU utilization are returned +} nvmlVgpuInstancesUtilizationInfo_v1_t; +typedef nvmlVgpuInstancesUtilizationInfo_v1_t nvmlVgpuInstancesUtilizationInfo_t; +#define nvmlVgpuInstancesUtilizationInfo_v1 NVML_STRUCT_VERSION(VgpuInstancesUtilizationInfo, 1) + +/** + * Structure to store Utilization Value, vgpuInstance and subprocess information + */ +typedef struct nvmlVgpuProcessUtilizationSample_st +{ + nvmlVgpuInstance_t vgpuInstance; //!< vGPU Instance + unsigned int pid; //!< PID of process running within the vGPU VM + char processName[NVML_VGPU_NAME_BUFFER_SIZE]; //!< Name of process running within the vGPU VM + unsigned long long timeStamp; //!< CPU Timestamp in microseconds + unsigned int smUtil; //!< SM (3D/Compute) Util Value + unsigned int memUtil; //!< Frame Buffer Memory Util Value + unsigned int encUtil; //!< Encoder Util Value + unsigned int decUtil; //!< Decoder Util Value +} nvmlVgpuProcessUtilizationSample_t; + +/** + * Structure to store Utilization Value, vgpuInstance and subprocess information for process running on vGPU instance -- version 1 + */ +typedef struct +{ + char processName[NVML_VGPU_NAME_BUFFER_SIZE]; //!< Name of process running within the vGPU VM + unsigned long long timeStamp; //!< CPU Timestamp in microseconds + nvmlVgpuInstance_t vgpuInstance; //!< vGPU Instance + unsigned int pid; //!< PID of process running within the vGPU VM + unsigned int smUtil; //!< SM (3D/Compute) Util Value + unsigned int memUtil; //!< Frame Buffer Memory Util Value + unsigned int encUtil; //!< Encoder Util Value + unsigned int decUtil; //!< Decoder Util Value + unsigned int jpgUtil; //!< Jpeg Util Value + unsigned int ofaUtil; //!< Ofa Util Value +} nvmlVgpuProcessUtilizationInfo_v1_t; + +/** + * Structure to store recent utilization, vgpuInstance and subprocess information for processes running on vGPU instances active on a device -- version 1 + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + unsigned int vgpuProcessCount; //!< Hold the number of processes running on vGPU instances + unsigned long long lastSeenTimeStamp; //!< Return only samples with timestamp greater than lastSeenTimeStamp + nvmlVgpuProcessUtilizationInfo_v1_t *vgpuProcUtilArray; //!< The array (allocated by caller) in which utilization of processes running on vGPU instances are returned +} nvmlVgpuProcessesUtilizationInfo_v1_t; +typedef nvmlVgpuProcessesUtilizationInfo_v1_t nvmlVgpuProcessesUtilizationInfo_t; +#define nvmlVgpuProcessesUtilizationInfo_v1 NVML_STRUCT_VERSION(VgpuProcessesUtilizationInfo, 1) + +/** + * Structure to store the information of vGPU runtime state -- version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + unsigned long long size; //!< OUT: The runtime state size of the vGPU instance +} nvmlVgpuRuntimeState_v1_t; +typedef nvmlVgpuRuntimeState_v1_t nvmlVgpuRuntimeState_t; +#define nvmlVgpuRuntimeState_v1 NVML_STRUCT_VERSION(VgpuRuntimeState, 1) + +/** + * vGPU scheduler policies + */ +#define NVML_VGPU_SCHEDULER_POLICY_UNKNOWN 0 +#define NVML_VGPU_SCHEDULER_POLICY_BEST_EFFORT 1 +#define NVML_VGPU_SCHEDULER_POLICY_EQUAL_SHARE 2 +#define NVML_VGPU_SCHEDULER_POLICY_FIXED_SHARE 3 + +#define NVML_SUPPORTED_VGPU_SCHEDULER_POLICY_COUNT 3 + +#define NVML_SCHEDULER_SW_MAX_LOG_ENTRIES 200 + +#define NVML_VGPU_SCHEDULER_ARR_DEFAULT 0 +#define NVML_VGPU_SCHEDULER_ARR_DISABLE 1 +#define NVML_VGPU_SCHEDULER_ARR_ENABLE 2 + +/** + * vGPU scheduler engine types + */ +#define NVML_VGPU_SCHEDULER_ENGINE_TYPE_GRAPHICS 1 + +/** + * Union to represent the vGPU Scheduler Parameters + */ +typedef union +{ + struct + { + unsigned int avgFactor; //!< Average factor in compensating the timeslice for Adaptive Round Robin mode + unsigned int timeslice; //!< The timeslice in ns for each software run list as configured, or the default value otherwise + } vgpuSchedDataWithARR; + + struct + { + unsigned int timeslice; //!< The timeslice in ns for each software run list as configured, or the default value otherwise + } vgpuSchedData; + +} nvmlVgpuSchedulerParams_t; + +/** + * Structure to store the state and logs of a software runlist + */ +typedef struct nvmlVgpuSchedulerLogEntries_st +{ + unsigned long long timestamp; //!< Timestamp in ns when this software runlist was preeempted + unsigned long long timeRunTotal; //!< Total time in ns this software runlist has run + unsigned long long timeRun; //!< Time in ns this software runlist ran before preemption + unsigned int swRunlistId; //!< Software runlist Id + unsigned long long targetTimeSlice; //!< The actual timeslice after deduction + unsigned long long cumulativePreemptionTime; //!< Preemption time in ns for this SW runlist +} nvmlVgpuSchedulerLogEntry_t; + +/** + * Structure to store a vGPU software scheduler log + */ +typedef struct nvmlVgpuSchedulerLog_st +{ + unsigned int engineId; //!< Engine whose software runlist log entries are fetched + unsigned int schedulerPolicy; //!< Scheduler policy + unsigned int arrMode; //!< Adaptive Round Robin scheduler mode. One of the NVML_VGPU_SCHEDULER_ARR_*. + nvmlVgpuSchedulerParams_t schedulerParams; + unsigned int entriesCount; //!< Count of log entries fetched + nvmlVgpuSchedulerLogEntry_t logEntries[NVML_SCHEDULER_SW_MAX_LOG_ENTRIES]; +} nvmlVgpuSchedulerLog_t; + +/** + * Structure to store the vGPU scheduler state + */ +typedef struct nvmlVgpuSchedulerGetState_st +{ + unsigned int schedulerPolicy; //!< Scheduler policy + unsigned int arrMode; //!< Adaptive Round Robin scheduler mode. One of the NVML_VGPU_SCHEDULER_ARR_*. + nvmlVgpuSchedulerParams_t schedulerParams; +} nvmlVgpuSchedulerGetState_t; + +/** + * Union to represent the vGPU Scheduler set Parameters + */ +typedef union +{ + struct + { + unsigned int avgFactor; //!< Average factor in compensating the timeslice for Adaptive Round Robin mode + unsigned int frequency; //!< Frequency for Adaptive Round Robin mode + } vgpuSchedDataWithARR; + + struct + { + unsigned int timeslice; //!< The timeslice in ns(Nanoseconds) for each software run list as configured, or the default value otherwise + } vgpuSchedData; + +} nvmlVgpuSchedulerSetParams_t; + +/** + * Structure to set the vGPU scheduler state + */ +typedef struct nvmlVgpuSchedulerSetState_st +{ + unsigned int schedulerPolicy; //!< Scheduler policy + unsigned int enableARRMode; //!< Adaptive Round Robin scheduler + nvmlVgpuSchedulerSetParams_t schedulerParams; +} nvmlVgpuSchedulerSetState_t; + +/** + * Structure to store the vGPU scheduler capabilities + */ +typedef struct nvmlVgpuSchedulerCapabilities_st +{ + unsigned int supportedSchedulers[NVML_SUPPORTED_VGPU_SCHEDULER_POLICY_COUNT]; //!< List the supported vGPU schedulers on the device + unsigned int maxTimeslice; //!< Maximum timeslice value in ns + unsigned int minTimeslice; //!< Minimum timeslice value in ns + unsigned int isArrModeSupported; //!< Flag to check Adaptive Round Robin mode enabled/disabled. + unsigned int maxFrequencyForARR; //!< Maximum frequency for Adaptive Round Robin mode + unsigned int minFrequencyForARR; //!< Minimum frequency for Adaptive Round Robin mode + unsigned int maxAvgFactorForARR; //!< Maximum averaging factor for Adaptive Round Robin mode + unsigned int minAvgFactorForARR; //!< Minimum averaging factor for Adaptive Round Robin mode +} nvmlVgpuSchedulerCapabilities_t; + +/** + * Structure to store the vGPU license expiry details + */ +typedef struct nvmlVgpuLicenseExpiry_st +{ + unsigned int year; //!< Year of license expiry + unsigned short month; //!< Month of license expiry + unsigned short day; //!< Day of license expiry + unsigned short hour; //!< Hour of license expiry + unsigned short min; //!< Minutes of license expiry + unsigned short sec; //!< Seconds of license expiry + unsigned char status; //!< License expiry status +} nvmlVgpuLicenseExpiry_t; + +/** + * vGPU license state + */ +#define NVML_GRID_LICENSE_STATE_UNKNOWN 0 //!< Unknown state +#define NVML_GRID_LICENSE_STATE_UNINITIALIZED 1 //!< Uninitialized state +#define NVML_GRID_LICENSE_STATE_UNLICENSED_UNRESTRICTED 2 //!< Unlicensed unrestricted state +#define NVML_GRID_LICENSE_STATE_UNLICENSED_RESTRICTED 3 //!< Unlicensed restricted state +#define NVML_GRID_LICENSE_STATE_UNLICENSED 4 //!< Unlicensed state +#define NVML_GRID_LICENSE_STATE_LICENSED 5 //!< Licensed state + +typedef struct nvmlVgpuLicenseInfo_st +{ + unsigned char isLicensed; //!< License status + nvmlVgpuLicenseExpiry_t licenseExpiry; //!< License expiry information + unsigned int currentState; //!< Current license state +} nvmlVgpuLicenseInfo_t; + +/** + * Structure to store license expiry date and time values + */ +typedef struct nvmlGridLicenseExpiry_st +{ + unsigned int year; //!< Year value of license expiry + unsigned short month; //!< Month value of license expiry + unsigned short day; //!< Day value of license expiry + unsigned short hour; //!< Hour value of license expiry + unsigned short min; //!< Minutes value of license expiry + unsigned short sec; //!< Seconds value of license expiry + unsigned char status; //!< License expiry status +} nvmlGridLicenseExpiry_t; + +/** + * Structure containing vGPU software licensable feature information + */ +typedef struct nvmlGridLicensableFeature_st +{ + nvmlGridLicenseFeatureCode_t featureCode; //!< Licensed feature code + unsigned int featureState; //!< Non-zero if feature is currently licensed, otherwise zero + char licenseInfo[NVML_GRID_LICENSE_BUFFER_SIZE]; //!< Deprecated. + char productName[NVML_GRID_LICENSE_BUFFER_SIZE]; //!< Product name of feature + unsigned int featureEnabled; //!< Non-zero if feature is enabled, otherwise zero + nvmlGridLicenseExpiry_t licenseExpiry; //!< License expiry structure containing date and time +} nvmlGridLicensableFeature_t; + +/** + * Structure to store vGPU software licensable features + */ +typedef struct nvmlGridLicensableFeatures_st +{ + int isGridLicenseSupported; //!< Non-zero if vGPU Software Licensing is supported on the system, otherwise zero + unsigned int licensableFeaturesCount; //!< Entries returned in \a gridLicensableFeatures array + nvmlGridLicensableFeature_t gridLicensableFeatures[NVML_GRID_LICENSE_FEATURE_MAX_COUNT]; //!< Array of vGPU software licensable features. +} nvmlGridLicensableFeatures_t; + +/** + * Enum describing the GPU Recovery Action + */ +typedef enum nvmlDeviceGpuRecoveryAction_s { + NVML_GPU_RECOVERY_ACTION_NONE = 0, + NVML_GPU_RECOVERY_ACTION_GPU_RESET = 1, + NVML_GPU_RECOVERY_ACTION_NODE_REBOOT = 2, + NVML_GPU_RECOVERY_ACTION_DRAIN_P2P = 3, + NVML_GPU_RECOVERY_ACTION_DRAIN_AND_RESET = 4, +} nvmlDeviceGpuRecoveryAction_t; + +/** + * Structure to store the vGPU type IDs -- version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + unsigned int vgpuCount; //!< IN/OUT: Number of vGPU types + nvmlVgpuTypeId_t *vgpuTypeIds; //!< OUT: List of vGPU type IDs +} nvmlVgpuTypeIdInfo_v1_t; +typedef nvmlVgpuTypeIdInfo_v1_t nvmlVgpuTypeIdInfo_t; +#define nvmlVgpuTypeIdInfo_v1 NVML_STRUCT_VERSION(VgpuTypeIdInfo, 1) + +/** + * Structure to store the maximum number of possible vGPU type IDs -- version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + nvmlVgpuTypeId_t vgpuTypeId; //!< IN: Handle to vGPU type + unsigned int maxInstancePerGI; //!< OUT: Maximum number of vGPU instances per GPU instance +} nvmlVgpuTypeMaxInstance_v1_t; +typedef nvmlVgpuTypeMaxInstance_v1_t nvmlVgpuTypeMaxInstance_t; +#define nvmlVgpuTypeMaxInstance_v1 NVML_STRUCT_VERSION(VgpuTypeMaxInstance, 1) + +/** + * Structure to store active vGPU instance information -- Version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + unsigned int vgpuCount; //!< IN/OUT: Count of the active vGPU instances + nvmlVgpuInstance_t *vgpuInstances; //!< IN/OUT: list of active vGPU instances +} nvmlActiveVgpuInstanceInfo_v1_t; +typedef nvmlActiveVgpuInstanceInfo_v1_t nvmlActiveVgpuInstanceInfo_t; +#define nvmlActiveVgpuInstanceInfo_v1 NVML_STRUCT_VERSION(ActiveVgpuInstanceInfo, 1) + +/** + * Structure to set vGPU scheduler state information -- version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + unsigned int engineId; //!< IN: One of NVML_VGPU_SCHEDULER_ENGINE_TYPE_*. + unsigned int schedulerPolicy; //!< IN: Scheduler policy + unsigned int enableARRMode; //!< IN: Adaptive Round Robin scheduler + nvmlVgpuSchedulerSetParams_t schedulerParams; //!< IN: vGPU Scheduler Parameters +} nvmlVgpuSchedulerState_v1_t; +typedef nvmlVgpuSchedulerState_v1_t nvmlVgpuSchedulerState_t; +#define nvmlVgpuSchedulerState_v1 NVML_STRUCT_VERSION(VgpuSchedulerState, 1) + +/** + * Structure to store vGPU scheduler state information -- Version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + unsigned int engineId; //!< IN: Engine whose software scheduler state info is fetched. One of NVML_VGPU_SCHEDULER_ENGINE_TYPE_*. + unsigned int schedulerPolicy; //!< OUT: Scheduler policy + unsigned int arrMode; //!< OUT: Adaptive Round Robin scheduler mode. One of the NVML_VGPU_SCHEDULER_ARR_*. + nvmlVgpuSchedulerParams_t schedulerParams; //!< OUT: vGPU Scheduler Parameters +} nvmlVgpuSchedulerStateInfo_v1_t; +typedef nvmlVgpuSchedulerStateInfo_v1_t nvmlVgpuSchedulerStateInfo_t; +#define nvmlVgpuSchedulerStateInfo_v1 NVML_STRUCT_VERSION(VgpuSchedulerStateInfo, 1) + +/** + * Structure to store vGPU scheduler log information -- Version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + unsigned int engineId; //!< IN: Engine whose software runlist log entries are fetched. One of One of NVML_VGPU_SCHEDULER_ENGINE_TYPE_*. + unsigned int schedulerPolicy; //!< OUT: Scheduler policy + unsigned int arrMode; //!< OUT: Adaptive Round Robin scheduler mode. One of the NVML_VGPU_SCHEDULER_ARR_*. + nvmlVgpuSchedulerParams_t schedulerParams; //!< OUT: vGPU Scheduler Parameters + unsigned int entriesCount; //!< OUT: Count of log entries fetched + nvmlVgpuSchedulerLogEntry_t logEntries[NVML_SCHEDULER_SW_MAX_LOG_ENTRIES]; //!< OUT: Structure to store the state and logs of a software runlist +} nvmlVgpuSchedulerLogInfo_v1_t; +typedef nvmlVgpuSchedulerLogInfo_v1_t nvmlVgpuSchedulerLogInfo_t; +#define nvmlVgpuSchedulerLogInfo_v1 NVML_STRUCT_VERSION(VgpuSchedulerLogInfo, 1) + +/** + * Structure to store creatable vGPU placement information -- version 1 + */ +typedef struct +{ + unsigned int version; //!< IN: The version number of this struct + nvmlVgpuTypeId_t vgpuTypeId; //!< IN: Handle to vGPU type + unsigned int count; //!< IN/OUT: Count of the placement IDs + unsigned int *placementIds; //!< IN/OUT: Placement IDs for the vGPU type + unsigned int placementSize; //!< OUT: The number of slots occupied by the vGPU type +} nvmlVgpuCreatablePlacementInfo_v1_t; +typedef nvmlVgpuCreatablePlacementInfo_v1_t nvmlVgpuCreatablePlacementInfo_t; +#define nvmlVgpuCreatablePlacementInfo_v1 NVML_STRUCT_VERSION(VgpuCreatablePlacementInfo, 1) + +/** @} */ +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlFieldValueEnums Field Value Enums + * @{ + */ +/***************************************************************************************************/ + +/** + * Field Identifiers. + * + * All Identifiers pertain to a device. Each ID is only used once and is guaranteed never to change. + */ +#define NVML_FI_DEV_ECC_CURRENT 1 //!< Current ECC mode. 1=Active. 0=Inactive +#define NVML_FI_DEV_ECC_PENDING 2 //!< Pending ECC mode. 1=Active. 0=Inactive +/* ECC Count Totals */ +#define NVML_FI_DEV_ECC_SBE_VOL_TOTAL 3 //!< Total single bit volatile ECC errors +#define NVML_FI_DEV_ECC_DBE_VOL_TOTAL 4 //!< Total double bit volatile ECC errors +#define NVML_FI_DEV_ECC_SBE_AGG_TOTAL 5 //!< Total single bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_DBE_AGG_TOTAL 6 //!< Total double bit aggregate (persistent) ECC errors +/* Individual ECC locations */ +#define NVML_FI_DEV_ECC_SBE_VOL_L1 7 //!< L1 cache single bit volatile ECC errors +#define NVML_FI_DEV_ECC_DBE_VOL_L1 8 //!< L1 cache double bit volatile ECC errors +#define NVML_FI_DEV_ECC_SBE_VOL_L2 9 //!< L2 cache single bit volatile ECC errors +#define NVML_FI_DEV_ECC_DBE_VOL_L2 10 //!< L2 cache double bit volatile ECC errors +#define NVML_FI_DEV_ECC_SBE_VOL_DEV 11 //!< Device memory single bit volatile ECC errors +#define NVML_FI_DEV_ECC_DBE_VOL_DEV 12 //!< Device memory double bit volatile ECC errors +#define NVML_FI_DEV_ECC_SBE_VOL_REG 13 //!< Register file single bit volatile ECC errors +#define NVML_FI_DEV_ECC_DBE_VOL_REG 14 //!< Register file double bit volatile ECC errors +#define NVML_FI_DEV_ECC_SBE_VOL_TEX 15 //!< Texture memory single bit volatile ECC errors +#define NVML_FI_DEV_ECC_DBE_VOL_TEX 16 //!< Texture memory double bit volatile ECC errors +#define NVML_FI_DEV_ECC_DBE_VOL_CBU 17 //!< CBU double bit volatile ECC errors +#define NVML_FI_DEV_ECC_SBE_AGG_L1 18 //!< L1 cache single bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_DBE_AGG_L1 19 //!< L1 cache double bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_SBE_AGG_L2 20 //!< L2 cache single bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_DBE_AGG_L2 21 //!< L2 cache double bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_SBE_AGG_DEV 22 //!< Device memory single bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_DBE_AGG_DEV 23 //!< Device memory double bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_SBE_AGG_REG 24 //!< Register File single bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_DBE_AGG_REG 25 //!< Register File double bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_SBE_AGG_TEX 26 //!< Texture memory single bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_DBE_AGG_TEX 27 //!< Texture memory double bit aggregate (persistent) ECC errors +#define NVML_FI_DEV_ECC_DBE_AGG_CBU 28 //!< CBU double bit aggregate ECC errors + +/* Page Retirement */ +#define NVML_FI_DEV_RETIRED_SBE 29 //!< Number of retired pages because of single bit errors +#define NVML_FI_DEV_RETIRED_DBE 30 //!< Number of retired pages because of double bit errors +#define NVML_FI_DEV_RETIRED_PENDING 31 //!< If any pages are pending retirement. 1=yes. 0=no. + +/** + * NVLink Flit Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L0 32 //!< NVLink flow control CRC Error Counter for Lane 0 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L1 33 //!< NVLink flow control CRC Error Counter for Lane 1 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L2 34 //!< NVLink flow control CRC Error Counter for Lane 2 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L3 35 //!< NVLink flow control CRC Error Counter for Lane 3 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L4 36 //!< NVLink flow control CRC Error Counter for Lane 4 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L5 37 //!< NVLink flow control CRC Error Counter for Lane 5 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_TOTAL 38 //!< NVLink flow control CRC Error Counter total for all Lanes + +/** + * NVLink CRC Data Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L0 39 //!< NVLink data CRC Error Counter for Lane 0 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L1 40 //!< NVLink data CRC Error Counter for Lane 1 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L2 41 //!< NVLink data CRC Error Counter for Lane 2 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L3 42 //!< NVLink data CRC Error Counter for Lane 3 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L4 43 //!< NVLink data CRC Error Counter for Lane 4 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L5 44 //!< NVLink data CRC Error Counter for Lane 5 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_TOTAL 45 //!< NvLink data CRC Error Counter total for all Lanes + +/** + * NVLink Replay Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L0 46 //!< NVLink Replay Error Counter for Lane 0 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L1 47 //!< NVLink Replay Error Counter for Lane 1 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L2 48 //!< NVLink Replay Error Counter for Lane 2 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L3 49 //!< NVLink Replay Error Counter for Lane 3 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L4 50 //!< NVLink Replay Error Counter for Lane 4 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L5 51 //!< NVLink Replay Error Counter for Lane 5 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_TOTAL 52 //!< NVLink Replay Error Counter total for all Lanes + +/** + * NVLink Recovery Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L0 53 //!< NVLink Recovery Error Counter for Lane 0 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L1 54 //!< NVLink Recovery Error Counter for Lane 1 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L2 55 //!< NVLink Recovery Error Counter for Lane 2 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L3 56 //!< NVLink Recovery Error Counter for Lane 3 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L4 57 //!< NVLink Recovery Error Counter for Lane 4 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L5 58 //!< NVLink Recovery Error Counter for Lane 5 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_TOTAL 59 //!< NVLink Recovery Error Counter total for all Lanes + +/* NvLink Bandwidth Counters */ +/* + * NVML_FI_DEV_NVLINK_BANDWIDTH_* field values are now deprecated. + * Please use the following field values instead: + * NVML_FI_DEV_NVLINK_THROUGHPUT_DATA_TX + * NVML_FI_DEV_NVLINK_THROUGHPUT_DATA_RX + * NVML_FI_DEV_NVLINK_THROUGHPUT_RAW_TX + * NVML_FI_DEV_NVLINK_THROUGHPUT_RAW_RX + */ +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L0 60 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 0 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L1 61 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 1 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L2 62 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 2 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L3 63 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 3 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L4 64 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 4 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L5 65 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 5 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_TOTAL 66 //!< NVLink Bandwidth Counter Total for Counter Set 0, All Lanes + +/* NvLink Bandwidth Counters */ +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L0 67 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 0 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L1 68 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 1 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L2 69 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 2 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L3 70 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 3 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L4 71 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 4 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L5 72 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 5 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_TOTAL 73 //!< NVLink Bandwidth Counter Total for Counter Set 1, All Lanes + +/* NVML Perf Policy Counters */ +#define NVML_FI_DEV_PERF_POLICY_POWER 74 //!< Perf Policy Counter for Power Policy +#define NVML_FI_DEV_PERF_POLICY_THERMAL 75 //!< Perf Policy Counter for Thermal Policy +#define NVML_FI_DEV_PERF_POLICY_SYNC_BOOST 76 //!< Perf Policy Counter for Sync boost Policy +#define NVML_FI_DEV_PERF_POLICY_BOARD_LIMIT 77 //!< Perf Policy Counter for Board Limit +#define NVML_FI_DEV_PERF_POLICY_LOW_UTILIZATION 78 //!< Perf Policy Counter for Low GPU Utilization Policy +#define NVML_FI_DEV_PERF_POLICY_RELIABILITY 79 //!< Perf Policy Counter for Reliability Policy +#define NVML_FI_DEV_PERF_POLICY_TOTAL_APP_CLOCKS 80 //!< Perf Policy Counter for Total App Clock Policy +#define NVML_FI_DEV_PERF_POLICY_TOTAL_BASE_CLOCKS 81 //!< Perf Policy Counter for Total Base Clocks Policy + +/* Memory temperatures */ +#define NVML_FI_DEV_MEMORY_TEMP 82 //!< Memory temperature for the device + +/* Energy Counter */ +#define NVML_FI_DEV_TOTAL_ENERGY_CONSUMPTION 83 //!< Total energy consumption for the GPU in mJ since the driver was last reloaded + +/** + * NVLink Speed + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L0 84 //!< NVLink Speed in MBps for Link 0 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L1 85 //!< NVLink Speed in MBps for Link 1 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L2 86 //!< NVLink Speed in MBps for Link 2 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L3 87 //!< NVLink Speed in MBps for Link 3 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L4 88 //!< NVLink Speed in MBps for Link 4 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L5 89 //!< NVLink Speed in MBps for Link 5 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_COMMON 90 //!< Common NVLink Speed in MBps for active links + +#define NVML_FI_DEV_NVLINK_LINK_COUNT 91 //!< Number of NVLinks present on the device + +#define NVML_FI_DEV_RETIRED_PENDING_SBE 92 //!< If any pages are pending retirement due to SBE. 1=yes. 0=no. +#define NVML_FI_DEV_RETIRED_PENDING_DBE 93 //!< If any pages are pending retirement due to DBE. 1=yes. 0=no. + +#define NVML_FI_DEV_PCIE_REPLAY_COUNTER 94 //!< PCIe replay counter +#define NVML_FI_DEV_PCIE_REPLAY_ROLLOVER_COUNTER 95 //!< PCIe replay rollover counter + +/** + * NVLink Flit Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L6 96 //!< NVLink flow control CRC Error Counter for Lane 6 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L7 97 //!< NVLink flow control CRC Error Counter for Lane 7 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L8 98 //!< NVLink flow control CRC Error Counter for Lane 8 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L9 99 //!< NVLink flow control CRC Error Counter for Lane 9 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L10 100 //!< NVLink flow control CRC Error Counter for Lane 10 +#define NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_L11 101 //!< NVLink flow control CRC Error Counter for Lane 11 + +/** + * NVLink CRC Data Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L6 102 //!< NVLink data CRC Error Counter for Lane 6 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L7 103 //!< NVLink data CRC Error Counter for Lane 7 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L8 104 //!< NVLink data CRC Error Counter for Lane 8 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L9 105 //!< NVLink data CRC Error Counter for Lane 9 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L10 106 //!< NVLink data CRC Error Counter for Lane 10 +#define NVML_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_L11 107 //!< NVLink data CRC Error Counter for Lane 11 + +/** + * NVLink Replay Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L6 108 //!< NVLink Replay Error Counter for Lane 6 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L7 109 //!< NVLink Replay Error Counter for Lane 7 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L8 110 //!< NVLink Replay Error Counter for Lane 8 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L9 111 //!< NVLink Replay Error Counter for Lane 9 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L10 112 //!< NVLink Replay Error Counter for Lane 10 +#define NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L11 113 //!< NVLink Replay Error Counter for Lane 11 + +/** + * NVLink Recovery Error Counters + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L6 114 //!< NVLink Recovery Error Counter for Lane 6 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L7 115 //!< NVLink Recovery Error Counter for Lane 7 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L8 116 //!< NVLink Recovery Error Counter for Lane 8 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L9 117 //!< NVLink Recovery Error Counter for Lane 9 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L10 118 //!< NVLink Recovery Error Counter for Lane 10 +#define NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_L11 119 //!< NVLink Recovery Error Counter for Lane 11 + +/* NvLink Bandwidth Counters */ +/* + * NVML_FI_DEV_NVLINK_BANDWIDTH_* field values are now deprecated. + * Please use the following field values instead: + * NVML_FI_DEV_NVLINK_THROUGHPUT_DATA_TX + * NVML_FI_DEV_NVLINK_THROUGHPUT_DATA_RX + * NVML_FI_DEV_NVLINK_THROUGHPUT_RAW_TX + * NVML_FI_DEV_NVLINK_THROUGHPUT_RAW_RX + */ +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L6 120 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 6 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L7 121 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 7 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L8 122 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 8 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L9 123 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 9 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L10 124 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 10 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C0_L11 125 //!< NVLink Bandwidth Counter for Counter Set 0, Lane 11 + +/* NvLink Bandwidth Counters */ +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L6 126 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 6 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L7 127 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 7 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L8 128 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 8 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L9 129 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 9 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L10 130 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 10 +#define NVML_FI_DEV_NVLINK_BANDWIDTH_C1_L11 131 //!< NVLink Bandwidth Counter for Counter Set 1, Lane 11 + +/** + * NVLink Speed + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L6 132 //!< NVLink Speed in MBps for Link 6 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L7 133 //!< NVLink Speed in MBps for Link 7 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L8 134 //!< NVLink Speed in MBps for Link 8 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L9 135 //!< NVLink Speed in MBps for Link 9 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L10 136 //!< NVLink Speed in MBps for Link 10 +#define NVML_FI_DEV_NVLINK_SPEED_MBPS_L11 137 //!< NVLink Speed in MBps for Link 11 + +/** + * NVLink throughput counters field values + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + * A scopeId of UINT_MAX returns aggregate value summed up across all links + * for the specified counter type in fieldId. + */ +#define NVML_FI_DEV_NVLINK_THROUGHPUT_DATA_TX 138 //!< NVLink TX Data throughput in KiB +#define NVML_FI_DEV_NVLINK_THROUGHPUT_DATA_RX 139 //!< NVLink RX Data throughput in KiB +#define NVML_FI_DEV_NVLINK_THROUGHPUT_RAW_TX 140 //!< NVLink TX Data + protocol overhead in KiB +#define NVML_FI_DEV_NVLINK_THROUGHPUT_RAW_RX 141 //!< NVLink RX Data + protocol overhead in KiB + +/* Row Remapper */ +#define NVML_FI_DEV_REMAPPED_COR 142 //!< Number of remapped rows due to correctable errors +#define NVML_FI_DEV_REMAPPED_UNC 143 //!< Number of remapped rows due to uncorrectable errors +#define NVML_FI_DEV_REMAPPED_PENDING 144 //!< If any rows are pending remapping. 1=yes 0=no +#define NVML_FI_DEV_REMAPPED_FAILURE 145 //!< If any rows failed to be remapped 1=yes 0=no + +/** + * Remote device NVLink ID + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_REMOTE_NVLINK_ID 146 //!< Remote device NVLink ID + +/** + * NVSwitch: connected NVLink count + */ +#define NVML_FI_DEV_NVSWITCH_CONNECTED_LINK_COUNT 147 //!< Number of NVLinks connected to NVSwitch + +/* NvLink ECC Data Error Counters + * + * Lane ID needs to be specified in the scopeId field in nvmlFieldValue_t. + * + */ +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L0 148 //!< NVLink data ECC Error Counter for Link 0 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L1 149 //!< NVLink data ECC Error Counter for Link 1 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L2 150 //!< NVLink data ECC Error Counter for Link 2 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L3 151 //!< NVLink data ECC Error Counter for Link 3 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L4 152 //!< NVLink data ECC Error Counter for Link 4 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L5 153 //!< NVLink data ECC Error Counter for Link 5 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L6 154 //!< NVLink data ECC Error Counter for Link 6 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L7 155 //!< NVLink data ECC Error Counter for Link 7 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L8 156 //!< NVLink data ECC Error Counter for Link 8 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L9 157 //!< NVLink data ECC Error Counter for Link 9 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L10 158 //!< NVLink data ECC Error Counter for Link 10 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_L11 159 //!< NVLink data ECC Error Counter for Link 11 +#define NVML_FI_DEV_NVLINK_ECC_DATA_ERROR_COUNT_TOTAL 160 //!< NVLink data ECC Error Counter total for all Links + +/** + * NVLink Error Replay + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_ERROR_DL_REPLAY 161 //!< NVLink Replay Error Counter + //!< This is unsupported for Blackwell+. + //!< Please use NVML_FI_DEV_NVLINK_COUNT_LINK_RECOVERY_* +/** + * NVLink Recovery Error Counter + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_ERROR_DL_RECOVERY 162 //!< NVLink Recovery Error Counter + //!< This is unsupported for Blackwell+ + //!< Please use NVML_FI_DEV_NVLINK_COUNT_LINK_RECOVERY_* + +/** + * NVLink Recovery Error CRC Counter + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_ERROR_DL_CRC 163 //!< NVLink CRC Error Counter + //!< This is unsupported for Blackwell+ + //!< Please use NVML_FI_DEV_NVLINK_COUNT_LINK_RECOVERY_* + +/** + * NVLink Speed, State and Version field id 164, 165, and 166 + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_GET_SPEED 164 //!< NVLink Speed in MBps +#define NVML_FI_DEV_NVLINK_GET_STATE 165 //!< NVLink State - Active,Inactive +#define NVML_FI_DEV_NVLINK_GET_VERSION 166 //!< NVLink Version + +#define NVML_FI_DEV_NVLINK_GET_POWER_STATE 167 //!< NVLink Power state. 0=HIGH_SPEED 1=LOW_SPEED +#define NVML_FI_DEV_NVLINK_GET_POWER_THRESHOLD 168 //!< NVLink length of idle period (units can be found from + //!< NVML_FI_DEV_NVLINK_GET_POWER_THRESHOLD_UNITS) before + //!< transitioning links to sleep state + +#define NVML_FI_DEV_PCIE_L0_TO_RECOVERY_COUNTER 169 //!< Device PEX error recovery counter + +#define NVML_FI_DEV_C2C_LINK_COUNT 170 //!< Number of C2C Links present on the device +#define NVML_FI_DEV_C2C_LINK_GET_STATUS 171 //!< C2C Link Status 0=INACTIVE 1=ACTIVE +#define NVML_FI_DEV_C2C_LINK_GET_MAX_BW 172 //!< C2C Link Speed in MBps for active links + +#define NVML_FI_DEV_PCIE_COUNT_CORRECTABLE_ERRORS 173 //!< PCIe Correctable Errors Counter +#define NVML_FI_DEV_PCIE_COUNT_NAKS_RECEIVED 174 //!< PCIe NAK Receive Counter +#define NVML_FI_DEV_PCIE_COUNT_RECEIVER_ERROR 175 //!< PCIe Receiver Error Counter +#define NVML_FI_DEV_PCIE_COUNT_BAD_TLP 176 //!< PCIe Bad TLP Counter +#define NVML_FI_DEV_PCIE_COUNT_NAKS_SENT 177 //!< PCIe NAK Send Counter +#define NVML_FI_DEV_PCIE_COUNT_BAD_DLLP 178 //!< PCIe Bad DLLP Counter +#define NVML_FI_DEV_PCIE_COUNT_NON_FATAL_ERROR 179 //!< PCIe Non Fatal Error Counter +#define NVML_FI_DEV_PCIE_COUNT_FATAL_ERROR 180 //!< PCIe Fatal Error Counter +#define NVML_FI_DEV_PCIE_COUNT_UNSUPPORTED_REQ 181 //!< PCIe Unsupported Request Counter +#define NVML_FI_DEV_PCIE_COUNT_LCRC_ERROR 182 //!< PCIe LCRC Error Counter +#define NVML_FI_DEV_PCIE_COUNT_LANE_ERROR 183 //!< PCIe Per Lane Error Counter. + +#define NVML_FI_DEV_IS_RESETLESS_MIG_SUPPORTED 184 //!< Device's Restless MIG Capability + +/** + * Retrieves power usage for this GPU in milliwatts. + * It is only available if power management mode is supported. See \ref nvmlDeviceGetPowerManagementMode and + * \ref nvmlDeviceGetPowerUsage. + * + * scopeId needs to be specified. It signifies: + * 0 - GPU Only Scope - Metrics for GPU are retrieved + * 1 - Module scope - Metrics for the module (e.g. CPU + GPU) are retrieved. + * Note: CPU here refers to NVIDIA CPU (e.g. Grace). x86 or non-NVIDIA ARM is not supported + */ +#define NVML_FI_DEV_POWER_AVERAGE 185 //!< GPU power averaged over 1 sec interval, supported on Ampere (except GA100) or newer architectures. +#define NVML_FI_DEV_POWER_INSTANT 186 //!< Current GPU power, supported on all architectures. +#define NVML_FI_DEV_POWER_MIN_LIMIT 187 //!< Minimum power limit in milliwatts. +#define NVML_FI_DEV_POWER_MAX_LIMIT 188 //!< Maximum power limit in milliwatts. +#define NVML_FI_DEV_POWER_DEFAULT_LIMIT 189 //!< Default power limit in milliwatts (limit which device boots with). +#define NVML_FI_DEV_POWER_CURRENT_LIMIT 190 //!< Limit currently enforced in milliwatts (This includes other limits set elsewhere. E.g. Out-of-band). +#define NVML_FI_DEV_ENERGY 191 //!< Total energy consumption (in mJ) since the driver was last reloaded. Same as \ref NVML_FI_DEV_TOTAL_ENERGY_CONSUMPTION for the GPU. +#define NVML_FI_DEV_POWER_REQUESTED_LIMIT 192 //!< Power limit requested by NVML or any other userspace client. + +/** + * GPU T.Limit temperature thresholds in degree Celsius + * + * These fields are supported on Ada and later architectures and supersedes \ref nvmlDeviceGetTemperatureThreshold. + */ +#define NVML_FI_DEV_TEMPERATURE_SHUTDOWN_TLIMIT 193 //!< T.Limit temperature after which GPU may shut down for HW protection +#define NVML_FI_DEV_TEMPERATURE_SLOWDOWN_TLIMIT 194 //!< T.Limit temperature after which GPU may begin HW slowdown +#define NVML_FI_DEV_TEMPERATURE_MEM_MAX_TLIMIT 195 //!< T.Limit temperature after which GPU may begin SW slowdown due to memory temperature +#define NVML_FI_DEV_TEMPERATURE_GPU_MAX_TLIMIT 196 //!< T.Limit temperature after which GPU may be throttled below base clock + +#define NVML_FI_DEV_PCIE_COUNT_TX_BYTES 197 //!< PCIe transmit bytes. Value can be wrapped. +#define NVML_FI_DEV_PCIE_COUNT_RX_BYTES 198 //!< PCIe receive bytes. Value can be wrapped. + +#define NVML_FI_DEV_IS_MIG_MODE_INDEPENDENT_MIG_QUERY_CAPABLE 199 //!< MIG mode independent, MIG query capable device. 1=yes. 0=no. + +#define NVML_FI_DEV_NVLINK_GET_POWER_THRESHOLD_MAX 200 //!< Max Nvlink Power Threshold. See NVML_FI_DEV_NVLINK_GET_POWER_THRESHOLD + +/** + * NVLink counter field id 201-225 + * + * Link ID needs to be specified in the scopeId field in nvmlFieldValue_t. + */ +#define NVML_FI_DEV_NVLINK_COUNT_XMIT_PACKETS 201 //!usedGpuMemory is not supported + + + unsigned long long time; //!< Amount of time in ms during which the compute context was active. The time is reported as 0 if + //!< the process is not terminated + + unsigned long long startTime; //!< CPU Timestamp in usec representing start time for the process + + unsigned int isRunning; //!< Flag to represent if the process is running (1 for running, 0 for terminated) + + unsigned int reserved[5]; //!< Reserved for future use +} nvmlAccountingStats_t; + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlEncoderStructs Encoder Structs + * @{ + */ +/***************************************************************************************************/ + +/** + * Represents type of encoder for capacity can be queried + */ +typedef enum nvmlEncoderQueryType_enum +{ + NVML_ENCODER_QUERY_H264 = 0x00, //!< H264 encoder + NVML_ENCODER_QUERY_HEVC = 0x01, //!< HEVC encoder + NVML_ENCODER_QUERY_AV1 = 0x02, //!< AV1 encoder + NVML_ENCODER_QUERY_UNKNOWN = 0xFF //!< Unknown encoder +}nvmlEncoderType_t; + +/** + * Structure to hold encoder session data + */ +typedef struct nvmlEncoderSessionInfo_st +{ + unsigned int sessionId; //!< Unique session ID + unsigned int pid; //!< Owning process ID + nvmlVgpuInstance_t vgpuInstance; //!< Owning vGPU instance ID (only valid on vGPU hosts, otherwise zero) + nvmlEncoderType_t codecType; //!< Video encoder type + unsigned int hResolution; //!< Current encode horizontal resolution + unsigned int vResolution; //!< Current encode vertical resolution + unsigned int averageFps; //!< Moving average encode frames per second + unsigned int averageLatency; //!< Moving average encode latency in microseconds +}nvmlEncoderSessionInfo_t; + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlFBCStructs Frame Buffer Capture Structures +* @{ +*/ +/***************************************************************************************************/ + +/** + * Represents frame buffer capture session type + */ +typedef enum nvmlFBCSessionType_enum +{ + NVML_FBC_SESSION_TYPE_UNKNOWN = 0, //!< Unknown + NVML_FBC_SESSION_TYPE_TOSYS, //!< ToSys + NVML_FBC_SESSION_TYPE_CUDA, //!< Cuda + NVML_FBC_SESSION_TYPE_VID, //!< Vid + NVML_FBC_SESSION_TYPE_HWENC //!< HEnc +} nvmlFBCSessionType_t; + +/** + * Structure to hold frame buffer capture sessions stats + */ +typedef struct nvmlFBCStats_st +{ + unsigned int sessionsCount; //!< Total no of sessions + unsigned int averageFPS; //!< Moving average new frames captured per second + unsigned int averageLatency; //!< Moving average new frame capture latency in microseconds +} nvmlFBCStats_t; + +#define NVML_NVFBC_SESSION_FLAG_DIFFMAP_ENABLED 0x00000001 //!< Bit specifying differential map state. +#define NVML_NVFBC_SESSION_FLAG_CLASSIFICATIONMAP_ENABLED 0x00000002 //!< Bit specifying classification map state. +#define NVML_NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_NO_WAIT 0x00000004 //!< Bit specifying if capture was requested as non-blocking call. +#define NVML_NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_INFINITE 0x00000008 //!< Bit specifying if capture was requested as blocking call. +#define NVML_NVFBC_SESSION_FLAG_CAPTURE_WITH_WAIT_TIMEOUT 0x00000010 //!< Bit specifying if capture was requested as blocking call with timeout period. + +/** + * Structure to hold FBC session data + */ +typedef struct nvmlFBCSessionInfo_st +{ + unsigned int sessionId; //!< Unique session ID + unsigned int pid; //!< Owning process ID + nvmlVgpuInstance_t vgpuInstance; //!< Owning vGPU instance ID (only valid on vGPU hosts, otherwise zero) + unsigned int displayOrdinal; //!< Display identifier + nvmlFBCSessionType_t sessionType; //!< Type of frame buffer capture session + unsigned int sessionFlags; //!< Session flags (one or more of NVML_NVFBC_SESSION_FLAG_XXX). + unsigned int hMaxResolution; //!< Max horizontal resolution supported by the capture session + unsigned int vMaxResolution; //!< Max vertical resolution supported by the capture session + unsigned int hResolution; //!< Horizontal resolution requested by caller in capture call + unsigned int vResolution; //!< Vertical resolution requested by caller in capture call + unsigned int averageFPS; //!< Moving average new frames captured per second + unsigned int averageLatency; //!< Moving average new frame capture latency in microseconds +} nvmlFBCSessionInfo_t; + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlDrainDefs Drain State definitions + * @{ + */ +/***************************************************************************************************/ + +/** + * Is the GPU device to be removed from the kernel by nvmlDeviceRemoveGpu() + */ +typedef enum nvmlDetachGpuState_enum +{ + NVML_DETACH_GPU_KEEP = 0, + NVML_DETACH_GPU_REMOVE +} nvmlDetachGpuState_t; + +/** + * Parent bridge PCIe link state requested by nvmlDeviceRemoveGpu() + */ +typedef enum nvmlPcieLinkState_enum +{ + NVML_PCIE_LINK_KEEP = 0, + NVML_PCIE_LINK_SHUT_DOWN +} nvmlPcieLinkState_t; + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlConfidentialComputingDefs Confidential Computing definitions + * @{ + */ +/***************************************************************************************************/ +/** + * Confidential Compute CPU Capabilities values + */ +#define NVML_CC_SYSTEM_CPU_CAPS_NONE 0 +#define NVML_CC_SYSTEM_CPU_CAPS_AMD_SEV 1 +#define NVML_CC_SYSTEM_CPU_CAPS_INTEL_TDX 2 +#define NVML_CC_SYSTEM_CPU_CAPS_AMD_SEV_SNP 3 +#define NVML_CC_SYSTEM_CPU_CAPS_AMD_SNP_VTOM 4 + +/** + * Confidenial Compute GPU Capabilities values + */ +#define NVML_CC_SYSTEM_GPUS_CC_NOT_CAPABLE 0 +#define NVML_CC_SYSTEM_GPUS_CC_CAPABLE 1 + +typedef struct nvmlConfComputeSystemCaps_st { + unsigned int cpuCaps; + unsigned int gpusCaps; +} nvmlConfComputeSystemCaps_t; + +/** + * Confidential Compute DevTools Mode values + */ +#define NVML_CC_SYSTEM_DEVTOOLS_MODE_OFF 0 +#define NVML_CC_SYSTEM_DEVTOOLS_MODE_ON 1 + +/** + * Confidential Compute Environment values + */ +#define NVML_CC_SYSTEM_ENVIRONMENT_UNAVAILABLE 0 +#define NVML_CC_SYSTEM_ENVIRONMENT_SIM 1 +#define NVML_CC_SYSTEM_ENVIRONMENT_PROD 2 + +/** + * Confidential Compute Feature Status values + */ +#define NVML_CC_SYSTEM_FEATURE_DISABLED 0 +#define NVML_CC_SYSTEM_FEATURE_ENABLED 1 + +typedef struct nvmlConfComputeSystemState_st { + unsigned int environment; + unsigned int ccFeature; + unsigned int devToolsMode; +} nvmlConfComputeSystemState_t; + +/** + * Confidential Compute Multigpu mode values + */ +#define NVML_CC_SYSTEM_MULTIGPU_NONE 0 +#define NVML_CC_SYSTEM_MULTIGPU_PROTECTED_PCIE 1 +#define NVML_CC_SYSTEM_MULTIGPU_NVLE 2 + +/** + * Confidential Compute System settings + */ +typedef struct { + unsigned int version; + unsigned int environment; + unsigned int ccFeature; + unsigned int devToolsMode; + unsigned int multiGpuMode; +} nvmlSystemConfComputeSettings_v1_t; + +typedef nvmlSystemConfComputeSettings_v1_t nvmlSystemConfComputeSettings_t; +#define nvmlSystemConfComputeSettings_v1 NVML_STRUCT_VERSION(SystemConfComputeSettings, 1) + +/** + * Protected memory size + */ +typedef struct +nvmlConfComputeMemSizeInfo_st +{ + unsigned long long protectedMemSizeKib; + unsigned long long unprotectedMemSizeKib; +} nvmlConfComputeMemSizeInfo_t; + +/** + * Confidential Compute GPUs/System Ready State values + */ +#define NVML_CC_ACCEPTING_CLIENT_REQUESTS_FALSE 0 +#define NVML_CC_ACCEPTING_CLIENT_REQUESTS_TRUE 1 + +/** + * GPU Certificate Details + */ +#define NVML_GPU_CERT_CHAIN_SIZE 0x1000 +#define NVML_GPU_ATTESTATION_CERT_CHAIN_SIZE 0x1400 + +typedef struct nvmlConfComputeGpuCertificate_st { + unsigned int certChainSize; + unsigned int attestationCertChainSize; + unsigned char certChain[NVML_GPU_CERT_CHAIN_SIZE]; + unsigned char attestationCertChain[NVML_GPU_ATTESTATION_CERT_CHAIN_SIZE]; +} nvmlConfComputeGpuCertificate_t; + +/** + * GPU Attestation Report + */ +#define NVML_CC_GPU_CEC_NONCE_SIZE 0x20 +#define NVML_CC_GPU_ATTESTATION_REPORT_SIZE 0x2000 +#define NVML_CC_GPU_CEC_ATTESTATION_REPORT_SIZE 0x1000 +#define NVML_CC_CEC_ATTESTATION_REPORT_NOT_PRESENT 0 +#define NVML_CC_CEC_ATTESTATION_REPORT_PRESENT 1 +#define NVML_CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MIN 50 +#define NVML_CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MAX 65 + +typedef struct nvmlConfComputeGpuAttestationReport_st { + unsigned int isCecAttestationReportPresent; //!< output + unsigned int attestationReportSize; //!< output + unsigned int cecAttestationReportSize; //!< output + unsigned char nonce[NVML_CC_GPU_CEC_NONCE_SIZE]; //!< input: spdm supports 32 bytes on nonce + unsigned char attestationReport[NVML_CC_GPU_ATTESTATION_REPORT_SIZE]; //!< output + unsigned char cecAttestationReport[NVML_CC_GPU_CEC_ATTESTATION_REPORT_SIZE]; //!< output +} nvmlConfComputeGpuAttestationReport_t; + +typedef struct nvmlConfComputeSetKeyRotationThresholdInfo_st { + unsigned int version; + unsigned long long maxAttackerAdvantage; +} nvmlConfComputeSetKeyRotationThresholdInfo_v1_t; + +typedef nvmlConfComputeSetKeyRotationThresholdInfo_v1_t nvmlConfComputeSetKeyRotationThresholdInfo_t; +#define nvmlConfComputeSetKeyRotationThresholdInfo_v1 \ + NVML_STRUCT_VERSION(ConfComputeSetKeyRotationThresholdInfo, 1) + +typedef struct nvmlConfComputeGetKeyRotationThresholdInfo_st { + unsigned int version; + unsigned long long attackerAdvantage; +} nvmlConfComputeGetKeyRotationThresholdInfo_v1_t; + +typedef nvmlConfComputeGetKeyRotationThresholdInfo_v1_t nvmlConfComputeGetKeyRotationThresholdInfo_t; +#define nvmlConfComputeGetKeyRotationThresholdInfo_v1 \ + NVML_STRUCT_VERSION(ConfComputeGetKeyRotationThresholdInfo, 1) + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlFabricDefs Fabric definitions + * @{ + */ +/***************************************************************************************************/ + +#define NVML_GPU_FABRIC_UUID_LEN 16 //!< Length of Fabric UUID + +/** + * Fabric Probe States + */ +#define NVML_GPU_FABRIC_STATE_NOT_SUPPORTED 0 //!< Fabric Probe State not supported +#define NVML_GPU_FABRIC_STATE_NOT_STARTED 1 //!< Fabric Probe has not started +#define NVML_GPU_FABRIC_STATE_IN_PROGRESS 2 //!< Fabric Probe in progress +#define NVML_GPU_FABRIC_STATE_COMPLETED 3 //!< Fabric Probe State completed + +/** + * Probe State of GPU registration process + */ +typedef unsigned char nvmlGpuFabricState_t; + +/** + * Contains the device fabric information + */ +typedef struct +{ + unsigned char clusterUuid[NVML_GPU_FABRIC_UUID_LEN]; //!< Uuid of the cluster to which this GPU belongs + nvmlReturn_t status; //!< Error status, if any. Must be checked only if state returns "complete". + unsigned int cliqueId; //!< ID of the fabric clique to which this GPU belongs + nvmlGpuFabricState_t state; //!< Current state of GPU registration process. See NVML_GPU_FABRIC_STATE_* +} nvmlGpuFabricInfo_t; + +/** + * Fabric Degraded BW + */ +#define NVML_GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_NOT_SUPPORTED 0 //!< Fabric Health Mask: Degraded Bandwidth not supported +#define NVML_GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_TRUE 1 //!< Fabric Health Mask: Bandwidth degraded +#define NVML_GPU_FABRIC_HEALTH_MASK_DEGRADED_BW_FALSE 2 //!< Fabric Health Mask: Bandwidth not degraded + +#define NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_DEGRADED_BW 0 //!< Fabric Health Mask Bit Shift for Degraded Bandwidth +#define NVML_GPU_FABRIC_HEALTH_MASK_WIDTH_DEGRADED_BW 0x3 //!< Fabric Health Mask Width for Degraded Bandwidth + +/** + * Fabric Route Recovery + */ +#define NVML_GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_NOT_SUPPORTED 0 //!< Fabric Health Mask: Route Recovery not supported +#define NVML_GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_TRUE 1 //!< Fabric Health Mask: Route Recovery in progress +#define NVML_GPU_FABRIC_HEALTH_MASK_ROUTE_RECOVERY_FALSE 2 //!< Fabric Health Mask: Route Recovery not in progress + +#define NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_RECOVERY 2 //!< Fabric Health Mask Bit Shift for Route Recovery +#define NVML_GPU_FABRIC_HEALTH_MASK_WIDTH_ROUTE_RECOVERY 0x3 //!< Fabric Health Mask Width for Route Recovery + +/** + * Nvlink Fabric Route Unhealthy + */ +#define NVML_GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_NOT_SUPPORTED 0 //!< Fabric Health Mask: Route Unhealthy not supported +#define NVML_GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_TRUE 1 //!< Fabric Health Mask: Route is unhealthy +#define NVML_GPU_FABRIC_HEALTH_MASK_ROUTE_UNHEALTHY_FALSE 2 //!< Fabric Health Mask: Route is healthy + +#define NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_UNHEALTHY 4 //!< Fabric Health Mask Bit Shift for Route Unhealthy +#define NVML_GPU_FABRIC_HEALTH_MASK_WIDTH_ROUTE_UNHEALTHY 0x3 //!< Fabric Health Mask Width for Route Unhealthy + +/** + * Fabric Access Timeout Recovery + */ +#define NVML_GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_NOT_SUPPORTED 0 //!< Fabric Health Mask: Access Timeout Recovery not supported +#define NVML_GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_TRUE 1 //!< Fabric Health Mask: Access Timeout Recovery in progress +#define NVML_GPU_FABRIC_HEALTH_MASK_ACCESS_TIMEOUT_RECOVERY_FALSE 2 //!< Fabric Health Mask: Access Timeout Recovery not in progress + +#define NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_ACCESS_TIMEOUT_RECOVERY 6 //!< Fabric Health Mask Bit Shift for Access Timeout Recovery +#define NVML_GPU_FABRIC_HEALTH_MASK_WIDTH_ACCESS_TIMEOUT_RECOVERY 0x3 //!< Fabric Health Mask Width for Access Timeout Recovery + +/** + * Fabric Incorrect Configuration + */ +#define NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NOT_SUPPORTED 0 //!< Fabric Health Mask: Incorrect Configuration not supported +#define NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NONE 1 //!< Fabric Health Mask: Correct Configuration +#define NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INCORRECT_SYSGUID 2 //!< Fabric Health Mask: Incorrect Configuration - SysGUID +#define NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INCORRECT_CHASSIS_SN 3 //!< Fabric Health Mask: Incorrect Configuration - Chassis Serial Number +#define NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_NO_PARTITION 4 //!< Fabric Health Mask: Incorrect Configuration - No Partition +#define NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INSUFFICIENT_NVLINKS 5 //!< Fabric Health Mask: Incorrect Configuration - Insufficient Nvlinks + +#define NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_INCORRECT_CONFIGURATION 8 //!< Fabric Health Mask Bit Shift for Incorrect Configuration +#define NVML_GPU_FABRIC_HEALTH_MASK_WIDTH_INCORRECT_CONFIGURATION 0xf //!< Fabric Health Mask Width for Incorrect Configuration + +/** + * Fabric Health + */ +#define NVML_GPU_FABRIC_HEALTH_SUMMARY_NOT_SUPPORTED 0 //!< Fabric Health Summary: Not supported +#define NVML_GPU_FABRIC_HEALTH_SUMMARY_HEALTHY 1 //!< Fabric Health Summary: Healthy +#define NVML_GPU_FABRIC_HEALTH_SUMMARY_UNHEALTHY 2 //!< Fabric Health Summary: Unhealthy +#define NVML_GPU_FABRIC_HEALTH_SUMMARY_LIMITED_CAPACITY 3 //!< Fabric Health Summary: Limited Capacity + +/** + * GPU Fabric Health Status Mask for various fields can be obtained + * using the below macro. + * Ex - NVML_GPU_FABRIC_HEALTH_GET(var, _DEGRADED_BW) + */ +#define NVML_GPU_FABRIC_HEALTH_GET(var, type) \ + (((var) >> NVML_GPU_FABRIC_HEALTH_MASK_SHIFT##type) & \ + (NVML_GPU_FABRIC_HEALTH_MASK_WIDTH##type)) + +/** + * GPU Fabric Health Status Mask for various fields can be tested + * using the below macro. + * Ex - NVML_GPU_FABRIC_HEALTH_TEST(var, _DEGRADED_BW, _TRUE) + */ +#define NVML_GPU_FABRIC_HEALTH_TEST(var, type, val) \ + (NVML_GPU_FABRIC_HEALTH_GET(var, type) == \ + NVML_GPU_FABRIC_HEALTH_MASK##type##val) + +/** +* GPU Fabric information (v2). +* +* @deprecated nvmlGpuFabricInfo_v2_t is deprecated and will be removed in a future release. +* Use nvmlGpuFabricInfo_v3_t instead +* +* Version 2 adds the \ref nvmlGpuFabricInfo_v2_t.version field +* to the start of the structure, and the \ref nvmlGpuFabricInfo_v2_t.healthMask +* field to the end. This structure is not backwards-compatible with +* \ref nvmlGpuFabricInfo_t. +*/ +typedef struct +{ + unsigned int version; //!< Structure version identifier (set to nvmlGpuFabricInfo_v2) + unsigned char clusterUuid[NVML_GPU_FABRIC_UUID_LEN]; //!< Uuid of the cluster to which this GPU belongs + nvmlReturn_t status; //!< Probe Error status, if any. Must be checked only if Probe state returns "complete". + unsigned int cliqueId; //!< ID of the fabric clique to which this GPU belongs + nvmlGpuFabricState_t state; //!< Current Probe State of GPU registration process. See NVML_GPU_FABRIC_STATE_* + unsigned int healthMask; //!< GPU Fabric health Status Mask. See NVML_GPU_FABRIC_HEALTH_MASK_* +} nvmlGpuFabricInfo_v2_t; + +/** +* Version identifier value for \ref nvmlGpuFabricInfo_v2_t.version. +*/ +#define nvmlGpuFabricInfo_v2 NVML_STRUCT_VERSION(GpuFabricInfo, 2) + +/** +* GPU Fabric information (v3). +*/ +typedef struct +{ + unsigned int version; //!< Structure version identifier (set to nvmlGpuFabricInfo_v2) + unsigned char clusterUuid[NVML_GPU_FABRIC_UUID_LEN]; //!< Uuid of the cluster to which this GPU belongs + nvmlReturn_t status; //!< Probe Error status, if any. Must be checked only if Probe state returns "complete". + unsigned int cliqueId; //!< ID of the fabric clique to which this GPU belongs + nvmlGpuFabricState_t state; //!< Current Probe State of GPU registration process. See NVML_GPU_FABRIC_STATE_* + unsigned int healthMask; //!< GPU Fabric health Status Mask. See NVML_GPU_FABRIC_HEALTH_MASK_* + unsigned char healthSummary; //!< GPU Fabric health summary. See NVML_GPU_FABRIC_HEALTH_SUMMARY_* +} nvmlGpuFabricInfo_v3_t; + +typedef nvmlGpuFabricInfo_v3_t nvmlGpuFabricInfoV_t; + +/** +* Version identifier value for \ref nvmlGpuFabricInfo_v3_t.version. +*/ +#define nvmlGpuFabricInfo_v3 NVML_STRUCT_VERSION(GpuFabricInfo, 3) + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlInitializationAndCleanup Initialization and Cleanup + * This chapter describes the methods that handle NVML initialization and cleanup. + * It is the user's responsibility to call \ref nvmlInit_v2() before calling any other methods, and + * nvmlShutdown() once NVML is no longer being used. + * @{ + */ +/***************************************************************************************************/ + +#define NVML_INIT_FLAG_NO_GPUS 1 //!< Don't fail nvmlInit() when no GPUs are found +#define NVML_INIT_FLAG_NO_ATTACH 2 //!< Don't attach GPUs + +/** + * Initialize NVML, but don't initialize any GPUs yet. + * + * \note nvmlInit_v3 introduces a "flags" argument, that allows passing boolean values + * modifying the behaviour of nvmlInit(). + * \note In NVML 5.319 new nvmlInit_v2 has replaced nvmlInit"_v1" (default in NVML 4.304 and older) that + * did initialize all GPU devices in the system. + * + * This allows NVML to communicate with a GPU + * when other GPUs in the system are unstable or in a bad state. When using this API, GPUs are + * discovered and initialized in nvmlDeviceGetHandleBy* functions instead. + * + * \note To contrast nvmlInit_v2 with nvmlInit"_v1", NVML 4.304 nvmlInit"_v1" will fail when any detected GPU is in + * a bad or unstable state. + * + * For all products. + * + * This method, should be called once before invoking any other methods in the library. + * A reference count of the number of initializations is maintained. Shutdown only occurs + * when the reference count reaches zero. + * + * @return + * - \ref NVML_SUCCESS if NVML has been properly initialized + * - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running + * - \ref NVML_ERROR_NO_PERMISSION if NVML does not have permission to talk to the driver + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlInit_v2(void); + +/** + * nvmlInitWithFlags is a variant of nvmlInit(), that allows passing a set of boolean values + * modifying the behaviour of nvmlInit(). + * Other than the "flags" parameter it is completely similar to \ref nvmlInit_v2. + * + * For all products. + * + * @param flags behaviour modifier flags + * + * @return + * - \ref NVML_SUCCESS if NVML has been properly initialized + * - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running + * - \ref NVML_ERROR_NO_PERMISSION if NVML does not have permission to talk to the driver + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlInitWithFlags(unsigned int flags); + +/** + * Shut down NVML by releasing all GPU resources previously allocated with \ref nvmlInit_v2(). + * + * For all products. + * + * This method should be called after NVML work is done, once for each call to \ref nvmlInit_v2() + * A reference count of the number of initializations is maintained. Shutdown only occurs + * when the reference count reaches zero. For backwards compatibility, no error is reported if + * nvmlShutdown() is called more times than nvmlInit(). + * + * @return + * - \ref NVML_SUCCESS if NVML has been properly shut down + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlShutdown(void); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlErrorReporting Error reporting + * This chapter describes helper functions for error reporting routines. + * @{ + */ +/***************************************************************************************************/ + +/** + * Helper method for converting NVML error codes into readable strings. + * + * For all products. + * + * @param result NVML error code to convert + * + * @return String representation of the error. + * + */ +const DECLDIR char* nvmlErrorString(nvmlReturn_t result); +/** @} */ + + +/***************************************************************************************************/ +/** @defgroup nvmlConstants Constants + * @{ + */ +/***************************************************************************************************/ + +/** + * Buffer size guaranteed to be large enough for \ref nvmlDeviceGetInforomVersion and \ref nvmlDeviceGetInforomImageVersion + */ +#define NVML_DEVICE_INFOROM_VERSION_BUFFER_SIZE 16 + +/** + * Buffer size guaranteed to be large enough for storing GPU identifiers. + */ +#define NVML_DEVICE_UUID_BUFFER_SIZE 80 + +/** + * Buffer size guaranteed to be large enough for \ref nvmlDeviceGetUUID + */ +#define NVML_DEVICE_UUID_V2_BUFFER_SIZE 96 + +/** + * Buffer size guaranteed to be large enough for \ref nvmlDeviceGetBoardPartNumber + */ +#define NVML_DEVICE_PART_NUMBER_BUFFER_SIZE 80 + +/** + * Buffer size guaranteed to be large enough for \ref nvmlSystemGetDriverVersion + */ +#define NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE 80 + +/** + * Buffer size guaranteed to be large enough for \ref nvmlSystemGetNVMLVersion + */ +#define NVML_SYSTEM_NVML_VERSION_BUFFER_SIZE 80 + +/** + * Buffer size guaranteed to be large enough for storing GPU device names. + */ +#define NVML_DEVICE_NAME_BUFFER_SIZE 64 + +/** + * Buffer size guaranteed to be large enough for \ref nvmlDeviceGetName + */ +#define NVML_DEVICE_NAME_V2_BUFFER_SIZE 96 + +/** + * Buffer size guaranteed to be large enough for \ref nvmlDeviceGetSerial + */ +#define NVML_DEVICE_SERIAL_BUFFER_SIZE 30 + +/** + * Buffer size guaranteed to be large enough for \ref nvmlDeviceGetVbiosVersion + */ +#define NVML_DEVICE_VBIOS_VERSION_BUFFER_SIZE 32 + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlSystemQueries System Queries + * This chapter describes the queries that NVML can perform against the local system. These queries + * are not device-specific. + * @{ + */ +/***************************************************************************************************/ + +/** + * Retrieves the version of the system's graphics driver. + * + * For all products. + * + * The version identifier is an alphanumeric string. It will not exceed 80 characters in length + * (including the NULL terminator). See \ref nvmlConstants::NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE. + * + * @param version Reference in which to return the version identifier + * @param length The maximum allowed length of the string returned in \a version + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a version is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + */ +nvmlReturn_t DECLDIR nvmlSystemGetDriverVersion(char *version, unsigned int length); + +/** + * Retrieves the version of the NVML library. + * + * For all products. + * + * The version identifier is an alphanumeric string. It will not exceed 80 characters in length + * (including the NULL terminator). See \ref nvmlConstants::NVML_SYSTEM_NVML_VERSION_BUFFER_SIZE. + * + * @param version Reference in which to return the version identifier + * @param length The maximum allowed length of the string returned in \a version + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a version is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + */ +nvmlReturn_t DECLDIR nvmlSystemGetNVMLVersion(char *version, unsigned int length); + +/** + * Retrieves the version of the CUDA driver. + * + * For all products. + * + * The CUDA driver version returned will be retreived from the currently installed version of CUDA. + * If the cuda library is not found, this function will return a known supported version number. + * + * @param cudaDriverVersion Reference in which to return the version identifier + * + * @return + * - \ref NVML_SUCCESS if \a cudaDriverVersion has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a cudaDriverVersion is NULL + */ +nvmlReturn_t DECLDIR nvmlSystemGetCudaDriverVersion(int *cudaDriverVersion); + +/** + * Retrieves the version of the CUDA driver from the shared library. + * + * For all products. + * + * The returned CUDA driver version by calling cuDriverGetVersion() + * + * @param cudaDriverVersion Reference in which to return the version identifier + * + * @return + * - \ref NVML_SUCCESS if \a cudaDriverVersion has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a cudaDriverVersion is NULL + * - \ref NVML_ERROR_LIBRARY_NOT_FOUND if \a libcuda.so.1 or libcuda.dll is not found + * - \ref NVML_ERROR_FUNCTION_NOT_FOUND if \a cuDriverGetVersion() is not found in the shared library + */ +nvmlReturn_t DECLDIR nvmlSystemGetCudaDriverVersion_v2(int *cudaDriverVersion); + +/** + * Macros for converting the CUDA driver version number to Major and Minor version numbers. + */ +#define NVML_CUDA_DRIVER_VERSION_MAJOR(v) ((v)/1000) +#define NVML_CUDA_DRIVER_VERSION_MINOR(v) (((v)%1000)/10) + +/** + * Gets name of the process with provided process id + * + * For all products. + * + * Returned process name is cropped to provided length. + * name string is encoded in ANSI. + * + * @param pid The identifier of the process + * @param name Reference in which to return the process name + * @param length The maximum allowed length of the string returned in \a name + * + * @return + * - \ref NVML_SUCCESS if \a name has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a name is NULL or \a length is 0. + * - \ref NVML_ERROR_NOT_FOUND if process doesn't exists + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlSystemGetProcessName(unsigned int pid, char *name, unsigned int length); + +/** + * Retrieves the IDs and firmware versions for any Host Interface Cards (HICs) in the system. + * + * For S-class products. + * + * The \a hwbcCount argument is expected to be set to the size of the input \a hwbcEntries array. + * The HIC must be connected to an S-class system for it to be reported by this function. + * + * @param hwbcCount Size of hwbcEntries array + * @param hwbcEntries Array holding information about hwbc + * + * @return + * - \ref NVML_SUCCESS if \a hwbcCount and \a hwbcEntries have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if either \a hwbcCount or \a hwbcEntries is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a hwbcCount indicates that the \a hwbcEntries array is too small + */ +nvmlReturn_t DECLDIR nvmlSystemGetHicVersion(unsigned int *hwbcCount, nvmlHwbcEntry_t *hwbcEntries); + +/** + * Retrieve the set of GPUs that have a CPU affinity with the given CPU number + * For all products. + * Supported on Linux only. + * + * @param cpuNumber The CPU number + * @param count When zero, is set to the number of matching GPUs such that \a deviceArray + * can be malloc'd. When non-zero, \a deviceArray will be filled with \a count + * number of device handles. + * @param deviceArray An array of device handles for GPUs found with affinity to \a cpuNumber + * + * @return + * - \ref NVML_SUCCESS if \a deviceArray or \a count (if initially zero) has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a cpuNumber, or \a count is invalid, or \a deviceArray is NULL with a non-zero \a count + * - \ref NVML_ERROR_NOT_SUPPORTED if the device or OS does not support this feature + * - \ref NVML_ERROR_UNKNOWN an error has occurred in underlying topology discovery + */ +nvmlReturn_t DECLDIR nvmlSystemGetTopologyGpuSet(unsigned int cpuNumber, unsigned int *count, nvmlDevice_t *deviceArray); + +/** + * Structure to store Driver branch information + */ +typedef struct +{ + unsigned int version; //!< The version number of this struct + char branch[NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE]; //!< driver branch +} nvmlSystemDriverBranchInfo_v1_t; +typedef nvmlSystemDriverBranchInfo_v1_t nvmlSystemDriverBranchInfo_t; +#define nvmlSystemDriverBranchInfo_v1 NVML_STRUCT_VERSION(SystemDriverBranchInfo, 1) + +/** + * Retrieves the driver branch of the NVIDIA driver installed on the system. + * + * For all products. + * + * The branch identifier is an alphanumeric string. It will not exceed 80 characters in length + * (including the NULL terminator). See \ref nvmlConstants::NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE. + * + * @param branchInfo Pointer to the driver branch information structure \a nvmlSystemDriverBranchInfo_t + * @param length The maximum allowed length of the driver branch string + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a branchInfo is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlSystemGetDriverBranch(nvmlSystemDriverBranchInfo_t *branchInfo, unsigned int length); + + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlUnitQueries Unit Queries + * This chapter describes that queries that NVML can perform against each unit. For S-class systems only. + * In each case the device is identified with an nvmlUnit_t handle. This handle is obtained by + * calling \ref nvmlUnitGetHandleByIndex(). + * @{ + */ +/***************************************************************************************************/ + + /** + * Retrieves the number of units in the system. + * + * For S-class products. + * + * @param unitCount Reference in which to return the number of units + * + * @return + * - \ref NVML_SUCCESS if \a unitCount has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unitCount is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlUnitGetCount(unsigned int *unitCount); + +/** + * Acquire the handle for a particular unit, based on its index. + * + * For S-class products. + * + * Valid indices are derived from the \a unitCount returned by \ref nvmlUnitGetCount(). + * For example, if \a unitCount is 2 the valid indices are 0 and 1, corresponding to UNIT 0 and UNIT 1. + * + * The order in which NVML enumerates units has no guarantees of consistency between reboots. + * + * @param index The index of the target unit, >= 0 and < \a unitCount + * @param unit Reference in which to return the unit handle + * + * @return + * - \ref NVML_SUCCESS if \a unit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a index is invalid or \a unit is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlUnitGetHandleByIndex(unsigned int index, nvmlUnit_t *unit); + +/** + * Retrieves the static information associated with a unit. + * + * For S-class products. + * + * See \ref nvmlUnitInfo_t for details on available unit info. + * + * @param unit The identifier of the target unit + * @param info Reference in which to return the unit information + * + * @return + * - \ref NVML_SUCCESS if \a info has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a info is NULL + */ +nvmlReturn_t DECLDIR nvmlUnitGetUnitInfo(nvmlUnit_t unit, nvmlUnitInfo_t *info); + +/** + * Retrieves the LED state associated with this unit. + * + * For S-class products. + * + * See \ref nvmlLedState_t for details on allowed states. + * + * @param unit The identifier of the target unit + * @param state Reference in which to return the current LED state + * + * @return + * - \ref NVML_SUCCESS if \a state has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a state is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlUnitSetLedState() + */ +nvmlReturn_t DECLDIR nvmlUnitGetLedState(nvmlUnit_t unit, nvmlLedState_t *state); + +/** + * Retrieves the PSU stats for the unit. + * + * For S-class products. + * + * See \ref nvmlPSUInfo_t for details on available PSU info. + * + * @param unit The identifier of the target unit + * @param psu Reference in which to return the PSU information + * + * @return + * - \ref NVML_SUCCESS if \a psu has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a psu is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlUnitGetPsuInfo(nvmlUnit_t unit, nvmlPSUInfo_t *psu); + +/** + * Retrieves the temperature readings for the unit, in degrees C. + * + * For S-class products. + * + * Depending on the product, readings may be available for intake (type=0), + * exhaust (type=1) and board (type=2). + * + * @param unit The identifier of the target unit + * @param type The type of reading to take + * @param temp Reference in which to return the intake temperature + * + * @return + * - \ref NVML_SUCCESS if \a temp has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit or \a type is invalid or \a temp is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlUnitGetTemperature(nvmlUnit_t unit, unsigned int type, unsigned int *temp); + +/** + * Retrieves the fan speed readings for the unit. + * + * For S-class products. + * + * See \ref nvmlUnitFanSpeeds_t for details on available fan speed info. + * + * @param unit The identifier of the target unit + * @param fanSpeeds Reference in which to return the fan speed information + * + * @return + * - \ref NVML_SUCCESS if \a fanSpeeds has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a fanSpeeds is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlUnitGetFanSpeedInfo(nvmlUnit_t unit, nvmlUnitFanSpeeds_t *fanSpeeds); + +/** + * Retrieves the set of GPU devices that are attached to the specified unit. + * + * For S-class products. + * + * The \a deviceCount argument is expected to be set to the size of the input \a devices array. + * + * @param unit The identifier of the target unit + * @param deviceCount Reference in which to provide the \a devices array size, and + * to return the number of attached GPU devices + * @param devices Reference in which to return the references to the attached GPU devices + * + * @return + * - \ref NVML_SUCCESS if \a deviceCount and \a devices have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a deviceCount indicates that the \a devices array is too small + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid, either of \a deviceCount or \a devices is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlUnitGetDevices(nvmlUnit_t unit, unsigned int *deviceCount, nvmlDevice_t *devices); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlDeviceQueries Device Queries + * This chapter describes that queries that NVML can perform against each device. + * In each case the device is identified with an nvmlDevice_t handle. This handle is obtained by + * calling one of \ref nvmlDeviceGetHandleByIndex_v2(), \ref nvmlDeviceGetHandleBySerial(), + * \ref nvmlDeviceGetHandleByPciBusId_v2(). or \ref nvmlDeviceGetHandleByUUID(). + * @{ + */ +/***************************************************************************************************/ + + /** + * Retrieves the number of compute devices in the system. A compute device is a single GPU. + * + * For all products. + * + * Note: New nvmlDeviceGetCount_v2 (default in NVML 5.319) returns count of all devices in the system + * even if nvmlDeviceGetHandleByIndex_v2 returns NVML_ERROR_NO_PERMISSION for such device. + * Update your code to handle this error, or use NVML 4.304 or older nvml header file. + * For backward binary compatibility reasons _v1 version of the API is still present in the shared + * library. + * Old _v1 version of nvmlDeviceGetCount doesn't count devices that NVML has no permission to talk to. + * + * @param deviceCount Reference in which to return the number of accessible devices + * + * @return + * - \ref NVML_SUCCESS if \a deviceCount has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a deviceCount is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCount_v2(unsigned int *deviceCount); + +/** + * Get attributes (engine counts etc.) for the given NVML device handle. + * + * @note This API currently only supports MIG device handles. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device NVML device handle + * @param attributes Device attributes + * + * @return + * - \ref NVML_SUCCESS if \a device attributes were successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device handle is invalid + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAttributes_v2(nvmlDevice_t device, nvmlDeviceAttributes_t *attributes); + +/** + * Acquire the handle for a particular device, based on its index. + * + * For all products. + * + * Valid indices are derived from the \a accessibleDevices count returned by + * \ref nvmlDeviceGetCount_v2(). For example, if \a accessibleDevices is 2 the valid indices + * are 0 and 1, corresponding to GPU 0 and GPU 1. + * + * The order in which NVML enumerates devices has no guarantees of consistency between reboots. For that reason it + * is recommended that devices be looked up by their PCI ids or UUID. See + * \ref nvmlDeviceGetHandleByUUID() and \ref nvmlDeviceGetHandleByPciBusId_v2(). + * + * Note: The NVML index may not correlate with other APIs, such as the CUDA device index. + * + * Starting from NVML 5, this API causes NVML to initialize the target GPU + * NVML may initialize additional GPUs if: + * - The target GPU is an SLI slave + * + * Note: New nvmlDeviceGetCount_v2 (default in NVML 5.319) returns count of all devices in the system + * even if nvmlDeviceGetHandleByIndex_v2 returns NVML_ERROR_NO_PERMISSION for such device. + * Update your code to handle this error, or use NVML 4.304 or older nvml header file. + * For backward binary compatibility reasons _v1 version of the API is still present in the shared + * library. + * Old _v1 version of nvmlDeviceGetCount doesn't count devices that NVML has no permission to talk to. + * + * This means that nvmlDeviceGetHandleByIndex_v2 and _v1 can return different devices for the same index. + * If you don't touch macros that map old (_v1) versions to _v2 versions at the top of the file you don't + * need to worry about that. + * + * @param index The index of the target GPU, >= 0 and < \a accessibleDevices + * @param device Reference in which to return the device handle + * + * @return + * - \ref NVML_SUCCESS if \a device has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a index is invalid or \a device is NULL + * - \ref NVML_ERROR_INSUFFICIENT_POWER if any attached devices have improperly attached external power cables + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to talk to this device + * - \ref NVML_ERROR_IRQ_ISSUE if NVIDIA kernel detected an interrupt issue with the attached GPUs + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetIndex + * @see nvmlDeviceGetCount + */ +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByIndex_v2(unsigned int index, nvmlDevice_t *device); + +/** + * Acquire the handle for a particular device, based on its board serial number. + * + * For Fermi &tm; or newer fully supported devices. + * + * This number corresponds to the value printed directly on the board, and to the value returned by + * \ref nvmlDeviceGetSerial(). + * + * @deprecated Since more than one GPU can exist on a single board this function is deprecated in favor + * of \ref nvmlDeviceGetHandleByUUID. + * For dual GPU boards this function will return NVML_ERROR_INVALID_ARGUMENT. + * + * Starting from NVML 5, this API causes NVML to initialize the target GPU + * NVML may initialize additional GPUs as it searches for the target GPU + * + * @param serial The board serial number of the target GPU + * @param device Reference in which to return the device handle + * + * @return + * - \ref NVML_SUCCESS if \a device has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a serial is invalid, \a device is NULL or more than one + * device has the same serial (dual GPU boards) + * - \ref NVML_ERROR_NOT_FOUND if \a serial does not match a valid device on the system + * - \ref NVML_ERROR_INSUFFICIENT_POWER if any attached devices have improperly attached external power cables + * - \ref NVML_ERROR_IRQ_ISSUE if NVIDIA kernel detected an interrupt issue with the attached GPUs + * - \ref NVML_ERROR_GPU_IS_LOST if any GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetSerial + * @see nvmlDeviceGetHandleByUUID + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetHandleBySerial(const char *serial, nvmlDevice_t *device); + +/** + * Acquire the handle for a particular device, based on its globally unique immutable UUID (in ASCII format) associated with each device. + * + * For all products. + * + * @param uuid The UUID of the target GPU or MIG instance + * @param device Reference in which to return the device handle or MIG device handle + * + * Starting from NVML 5, this API causes NVML to initialize the target GPU + * NVML may initialize additional GPUs as it searches for the target GPU + * + * @return + * - \ref NVML_SUCCESS if \a device has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a uuid is invalid or \a device is null + * - \ref NVML_ERROR_NOT_FOUND if \a uuid does not match a valid device on the system + * - \ref NVML_ERROR_INSUFFICIENT_POWER if any attached devices have improperly attached external power cables + * - \ref NVML_ERROR_IRQ_ISSUE if NVIDIA kernel detected an interrupt issue with the attached GPUs + * - \ref NVML_ERROR_GPU_IS_LOST if any GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetUUID + */ +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByUUID(const char *uuid, nvmlDevice_t *device); + +/** + * Acquire the handle for a particular device, based on its globally unique immutable UUID (in either ASCII or binary format) associated with each device. + * See \ref nvmlUUID_v1_t for more information on the UUID struct. The caller must set the appropriate version prior to calling this API. + * + * For all products. + * + * @param[in] uuid The UUID of the target GPU or MIG instance + * @param[out] device Reference in which to return the device handle or MIG device handle + * + * This API causes NVML to initialize the target GPU + * NVML may initialize additional GPUs as it searches for the target GPU + * + * @return + * - \ref NVML_SUCCESS if \a device has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a uuid is invalid, \a device is null or \a uuid->type is invalid + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the provided version is invalid/unsupported + * - \ref NVML_ERROR_NOT_FOUND if \a uuid does not match a valid device on the system + * - \ref NVML_ERROR_GPU_IS_LOST if any GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByUUIDV(const nvmlUUID_t *uuid, nvmlDevice_t *device); + +/** + * Acquire the handle for a particular device, based on its PCI bus id. + * + * For all products. + * + * This value corresponds to the nvmlPciInfo_t::busId returned by \ref nvmlDeviceGetPciInfo_v3(). + * + * Starting from NVML 5, this API causes NVML to initialize the target GPU + * NVML may initialize additional GPUs if: + * - The target GPU is an SLI slave + * + * \note NVML 4.304 and older version of nvmlDeviceGetHandleByPciBusId"_v1" returns NVML_ERROR_NOT_FOUND + * instead of NVML_ERROR_NO_PERMISSION. + * + * @param pciBusId The PCI bus id of the target GPU + * Accept the following formats (all numbers in hexadecimal): + * domain:bus:device.function in format %x:%x:%x.%x + * domain:bus:device in format %x:%x:%x + * bus:device.function in format %x:%x.%x + * + * @param device Reference in which to return the device handle + * + * @return + * - \ref NVML_SUCCESS if \a device has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pciBusId is invalid or \a device is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a pciBusId does not match a valid device on the system + * - \ref NVML_ERROR_INSUFFICIENT_POWER if the attached device has improperly attached external power cables + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to talk to this device + * - \ref NVML_ERROR_IRQ_ISSUE if NVIDIA kernel detected an interrupt issue with the attached GPUs + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByPciBusId_v2(const char *pciBusId, nvmlDevice_t *device); + +/** + * Retrieves the name of this device. + * + * For all products. + * + * The name is an alphanumeric string that denotes a particular product, e.g. Tesla &tm; C2070. It will not + * exceed 96 characters in length (including the NULL terminator). See \ref + * nvmlConstants::NVML_DEVICE_NAME_V2_BUFFER_SIZE. + * + * When used with MIG device handles the API returns MIG device names which can be used to identify devices + * based on their attributes. + * + * @param device The identifier of the target device + * @param name Reference in which to return the product name + * @param length The maximum allowed length of the string returned in \a name + * + * @return + * - \ref NVML_SUCCESS if \a name has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a name is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetName(nvmlDevice_t device, char *name, unsigned int length); + +/** + * Retrieves the brand of this device. + * + * For all products. + * + * The type is a member of \ref nvmlBrandType_t defined above. + * + * @param device The identifier of the target device + * @param type Reference in which to return the product brand type + * + * @return + * - \ref NVML_SUCCESS if \a name has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a type is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBrand(nvmlDevice_t device, nvmlBrandType_t *type); + +/** + * Retrieves the NVML index of this device. + * + * For all products. + * + * Valid indices are derived from the \a accessibleDevices count returned by + * \ref nvmlDeviceGetCount_v2(). For example, if \a accessibleDevices is 2 the valid indices + * are 0 and 1, corresponding to GPU 0 and GPU 1. + * + * The order in which NVML enumerates devices has no guarantees of consistency between reboots. For that reason it + * is recommended that devices be looked up by their PCI ids or GPU UUID. See + * \ref nvmlDeviceGetHandleByPciBusId_v2() and \ref nvmlDeviceGetHandleByUUID(). + * + * When used with MIG device handles this API returns indices that can be + * passed to \ref nvmlDeviceGetMigDeviceHandleByIndex to retrieve an identical handle. + * MIG device indices are unique within a device. + * + * Note: The NVML index may not correlate with other APIs, such as the CUDA device index. + * + * @param device The identifier of the target device + * @param index Reference in which to return the NVML index of the device + * + * @return + * - \ref NVML_SUCCESS if \a index has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a index is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetHandleByIndex() + * @see nvmlDeviceGetCount() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetIndex(nvmlDevice_t device, unsigned int *index); + +/** + * Retrieves the globally unique board serial number associated with this device's board. + * + * For all products with an inforom. + * + * The serial number is an alphanumeric string that will not exceed 30 characters (including the NULL terminator). + * This number matches the serial number tag that is physically attached to the board. See \ref + * nvmlConstants::NVML_DEVICE_SERIAL_BUFFER_SIZE. + * + * @param device The identifier of the target device + * @param serial Reference in which to return the board/module serial number + * @param length The maximum allowed length of the string returned in \a serial + * + * @return + * - \ref NVML_SUCCESS if \a serial has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a serial is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSerial(nvmlDevice_t device, char *serial, unsigned int length); + +/** + * Get a unique identifier for the device module on the baseboard + * + * This API retrieves a unique identifier for each GPU module that exists on a given baseboard. + * For non-baseboard products, this ID would always be 0. + * + * @param device The identifier of the target device + * @param moduleId Unique identifier for the GPU module + * + * @return + * - \ref NVML_SUCCESS if \a moduleId has been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a moduleId is invalid + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetModuleId(nvmlDevice_t device, unsigned int *moduleId); + +/** + * Retrieves the Device's C2C Mode information + * + * @param device The identifier of the target device + * @param c2cModeInfo Output struct containing the device's C2C Mode info + * + * @return + * - \ref NVML_SUCCESS if \a C2C Mode Infor query is successful + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a serial is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetC2cModeInfoV(nvmlDevice_t device, nvmlC2cModeInfo_v1_t *c2cModeInfo); + +/***************************************************************************************************/ + +/** @defgroup nvmlAffinity CPU and Memory Affinity + * This chapter describes NVML operations that are associated with CPU and memory + * affinity. + * @{ + */ +/***************************************************************************************************/ + +//! Scope of NUMA node for affinity queries +#define NVML_AFFINITY_SCOPE_NODE 0 +//! Scope of processor socket for affinity queries +#define NVML_AFFINITY_SCOPE_SOCKET 1 + +typedef unsigned int nvmlAffinityScope_t; + +/** + * Retrieves an array of unsigned ints (sized to nodeSetSize) of bitmasks with + * the ideal memory affinity within node or socket for the device. + * For example, if NUMA node 0, 1 are ideal within the socket for the device and nodeSetSize == 1, + * result[0] = 0x3 + * + * \note If requested scope is not applicable to the target topology, the API + * will fall back to reporting the memory affinity for the immediate non-I/O + * ancestor of the device. + * + * For Kepler &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param nodeSetSize The size of the nodeSet array that is safe to access + * @param nodeSet Array reference in which to return a bitmask of NODEs, 64 NODEs per + * unsigned long on 64-bit machines, 32 on 32-bit machines + * @param scope Scope that change the default behavior + * + * @return + * - \ref NVML_SUCCESS if \a NUMA node Affinity has been filled + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, nodeSetSize == 0, nodeSet is NULL or scope is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ + +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryAffinity(nvmlDevice_t device, unsigned int nodeSetSize, unsigned long *nodeSet, nvmlAffinityScope_t scope); + +/** + * Retrieves an array of unsigned ints (sized to cpuSetSize) of bitmasks with the + * ideal CPU affinity within node or socket for the device. + * For example, if processors 0, 1, 32, and 33 are ideal for the device and cpuSetSize == 2, + * result[0] = 0x3, result[1] = 0x3 + * + * \note If requested scope is not applicable to the target topology, the API + * will fall back to reporting the CPU affinity for the immediate non-I/O + * ancestor of the device. + * + * For Kepler &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param cpuSetSize The size of the cpuSet array that is safe to access + * @param cpuSet Array reference in which to return a bitmask of CPUs, 64 CPUs per + * unsigned long on 64-bit machines, 32 on 32-bit machines + * @param scope Scope that change the default behavior + * + * @return + * - \ref NVML_SUCCESS if \a cpuAffinity has been filled + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, cpuSetSize == 0, cpuSet is NULL or sope is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ + +nvmlReturn_t DECLDIR nvmlDeviceGetCpuAffinityWithinScope(nvmlDevice_t device, unsigned int cpuSetSize, unsigned long *cpuSet, nvmlAffinityScope_t scope); + +/** + * Retrieves an array of unsigned ints (sized to cpuSetSize) of bitmasks with the ideal CPU affinity for the device + * For example, if processors 0, 1, 32, and 33 are ideal for the device and cpuSetSize == 2, + * result[0] = 0x3, result[1] = 0x3 + * This is equivalent to calling \ref nvmlDeviceGetCpuAffinityWithinScope with \ref NVML_AFFINITY_SCOPE_NODE. + * + * For Kepler &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param cpuSetSize The size of the cpuSet array that is safe to access + * @param cpuSet Array reference in which to return a bitmask of CPUs, 64 CPUs per + * unsigned long on 64-bit machines, 32 on 32-bit machines + * + * @return + * - \ref NVML_SUCCESS if \a cpuAffinity has been filled + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, cpuSetSize == 0, or cpuSet is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCpuAffinity(nvmlDevice_t device, unsigned int cpuSetSize, unsigned long *cpuSet); + +/** + * Sets the ideal affinity for the calling thread and device using the guidelines + * given in nvmlDeviceGetCpuAffinity(). Note, this is a change as of version 8.0. + * Older versions set the affinity for a calling process and all children. + * Currently supports up to 1024 processors. + * + * For Kepler &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if the calling process has been successfully bound + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetCpuAffinity(nvmlDevice_t device); + +/** + * Clear all affinity bindings for the calling thread. Note, this is a change as of version + * 8.0 as older versions cleared the affinity for a calling process and all children. + * + * For Kepler &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if the calling process has been successfully unbound + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceClearCpuAffinity(nvmlDevice_t device); + +/** + * Get the NUMA node of the given GPU device. + * This only applies to platforms where the GPUs are NUMA nodes. + * + * @param[in] device The device handle + * @param[out] node NUMA node ID of the device + * + * @returns + * - \ref NVML_SUCCESS if the NUMA node is retrieved successfully + * - \ref NVML_ERROR_NOT_SUPPORTED if request is not supported on the current platform + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device \a node is invalid + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNumaNodeId(nvmlDevice_t device, unsigned int *node); + +/** + * Get the addressing mode for a given GPU. Addressing modes can be one of: + * 1. HMM: System allocated memory (malloc, mmap) is addressable from the device (GPU), + * via software-based mirroring of the CPU's page tables, on the GPU. + * 2. ATS: System allocated memory (malloc, mmap) is addressable from the device (GPU), + * via Address Translation Services. This means that there is (effectively) + * a single set of page tables, and the CPU and GPU both use them. + * 3. None: Neither HMM nor ATS is active. + * + * %TURING_OR_NEWER% + * Supported on Linux only. + * + * @param[in] device The device handle + * @param[out] mode Pointer to addressing mode of the device + * + * @returns + * - \ref NVML_SUCCESS if \a mode is retrieved successfully + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the provided version is invalid/unsupported + * - \ref NVML_ERROR_NOT_SUPPORTED if request is not supported on the current platform + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device \a node is invalid + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAddressingMode(nvmlDevice_t device, nvmlDeviceAddressingMode_t *mode); + +/** + * Get the repair status for TPC/Channel repair + * + * For Ampere &tm; or newer fully supported devices. + * + * @param[in] device The identifier of the target device + * @param[out] repairStatus Reference to \a nvmlRepairStatus_t + * + * @return + * - \ref NVML_SUCCESS if the query was successful + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the provided version is invalid/unsupported + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRepairStatus(nvmlDevice_t device, nvmlRepairStatus_t *repairStatus); + +/** + * Retrieve the common ancestor for two devices + * For all products. + * Supported on Linux only. + * + * @param device1 The identifier of the first device + * @param device2 The identifier of the second device + * @param pathInfo A \ref nvmlGpuTopologyLevel_t that gives the path type + * + * @return + * - \ref NVML_SUCCESS if \a pathInfo has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device1, or \a device2 is invalid, or \a pathInfo is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device or OS does not support this feature + * - \ref NVML_ERROR_UNKNOWN an error has occurred in underlying topology discovery + */ + +/** @} */ +nvmlReturn_t DECLDIR nvmlDeviceGetTopologyCommonAncestor(nvmlDevice_t device1, nvmlDevice_t device2, nvmlGpuTopologyLevel_t *pathInfo); + +/** + * Retrieve the set of GPUs that are nearest to a given device at a specific interconnectivity level + * For all products. + * Supported on Linux only. + * + * @param device The identifier of the first device + * @param level The \ref nvmlGpuTopologyLevel_t level to search for other GPUs + * @param count When zero, is set to the number of matching GPUs such that \a deviceArray + * can be malloc'd. When non-zero, \a deviceArray will be filled with \a count + * number of device handles. + * @param deviceArray An array of device handles for GPUs found at \a level + * + * @return + * - \ref NVML_SUCCESS if \a deviceArray or \a count (if initially zero) has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a level, or \a count is invalid, or \a deviceArray is NULL with a non-zero \a count + * - \ref NVML_ERROR_NOT_SUPPORTED if the device or OS does not support this feature + * - \ref NVML_ERROR_UNKNOWN an error has occurred in underlying topology discovery + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTopologyNearestGpus(nvmlDevice_t device, nvmlGpuTopologyLevel_t level, unsigned int *count, nvmlDevice_t *deviceArray); + +/** + * Retrieve the status for a given p2p capability index between a given pair of GPU + * + * @param device1 The first device + * @param device2 The second device + * @param p2pIndex p2p Capability Index being looked for between \a device1 and \a device2 + * @param p2pStatus Reference in which to return the status of the \a p2pIndex + * between \a device1 and \a device2 + * @return + * - \ref NVML_SUCCESS if \a p2pStatus has been populated + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device1 or \a device2 or \a p2pIndex is invalid or \a p2pStatus is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetP2PStatus(nvmlDevice_t device1, nvmlDevice_t device2, nvmlGpuP2PCapsIndex_t p2pIndex,nvmlGpuP2PStatus_t *p2pStatus); + +/** + * Retrieves the globally unique immutable UUID associated with this device, as a 5 part hexadecimal string, + * that augments the immutable, board serial identifier. + * + * For all products. + * + * The UUID is a globally unique identifier. It is the only available identifier for pre-Fermi-architecture products. + * It does NOT correspond to any identifier printed on the board. It will not exceed 96 characters in length + * (including the NULL terminator). See \ref nvmlConstants::NVML_DEVICE_UUID_V2_BUFFER_SIZE. + * + * When used with MIG device handles the API returns globally unique UUIDs which can be used to identify MIG + * devices across both GPU and MIG devices. UUIDs are immutable for the lifetime of a MIG device. + * + * @param device The identifier of the target device + * @param uuid Reference in which to return the GPU UUID + * @param length The maximum allowed length of the string returned in \a uuid + * + * @return + * - \ref NVML_SUCCESS if \a uuid has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a uuid is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetUUID(nvmlDevice_t device, char *uuid, unsigned int length); + +/** + * Retrieves minor number for the device. The minor number for the device is such that the Nvidia device node file for + * each GPU will have the form /dev/nvidia[minor number]. + * + * For all products. + * Supported only for Linux + * + * @param device The identifier of the target device + * @param minorNumber Reference in which to return the minor number for the device + * @return + * - \ref NVML_SUCCESS if the minor number is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minorNumber is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMinorNumber(nvmlDevice_t device, unsigned int *minorNumber); + +/** + * Retrieves the the device board part number which is programmed into the board's InfoROM + * + * For all products. + * + * @param device Identifier of the target device + * @param partNumber Reference to the buffer to return + * @param length Length of the buffer reference + * + * @return + * - \ref NVML_SUCCESS if \a partNumber has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NOT_SUPPORTED if the needed VBIOS fields have not been filled + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a serial is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBoardPartNumber(nvmlDevice_t device, char* partNumber, unsigned int length); + +/** + * Retrieves the version information for the device's infoROM object. + * + * For all products with an inforom. + * + * Fermi and higher parts have non-volatile on-board memory for persisting device info, such as aggregate + * ECC counts. The version of the data structures in this memory may change from time to time. It will not + * exceed 16 characters in length (including the NULL terminator). + * See \ref nvmlConstants::NVML_DEVICE_INFOROM_VERSION_BUFFER_SIZE. + * + * See \ref nvmlInforomObject_t for details on the available infoROM objects. + * + * @param device The identifier of the target device + * @param object The target infoROM object + * @param version Reference in which to return the infoROM version + * @param length The maximum allowed length of the string returned in \a version + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a version is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have an infoROM + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetInforomImageVersion + */ +nvmlReturn_t DECLDIR nvmlDeviceGetInforomVersion(nvmlDevice_t device, nvmlInforomObject_t object, char *version, unsigned int length); + +/** + * Retrieves the global infoROM image version + * + * For all products with an inforom. + * + * Image version just like VBIOS version uniquely describes the exact version of the infoROM flashed on the board + * in contrast to infoROM object version which is only an indicator of supported features. + * Version string will not exceed 16 characters in length (including the NULL terminator). + * See \ref nvmlConstants::NVML_DEVICE_INFOROM_VERSION_BUFFER_SIZE. + * + * @param device The identifier of the target device + * @param version Reference in which to return the infoROM image version + * @param length The maximum allowed length of the string returned in \a version + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a version is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have an infoROM + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetInforomVersion + */ +nvmlReturn_t DECLDIR nvmlDeviceGetInforomImageVersion(nvmlDevice_t device, char *version, unsigned int length); + +/** + * Retrieves the checksum of the configuration stored in the device's infoROM. + * + * For all products with an inforom. + * + * Can be used to make sure that two GPUs have the exact same configuration. + * Current checksum takes into account configuration stored in PWR and ECC infoROM objects. + * Checksum can change between driver releases or when user changes configuration (e.g. disable/enable ECC) + * + * @param device The identifier of the target device + * @param checksum Reference in which to return the infoROM configuration checksum + * + * @return + * - \ref NVML_SUCCESS if \a checksum has been set + * - \ref NVML_ERROR_CORRUPTED_INFOROM if the device's checksum couldn't be retrieved due to infoROM corruption + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a checksum is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetInforomConfigurationChecksum(nvmlDevice_t device, unsigned int *checksum); + +/** + * Reads the infoROM from the flash and verifies the checksums. + * + * For all products with an inforom. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if infoROM is not corrupted + * - \ref NVML_ERROR_CORRUPTED_INFOROM if the device's infoROM is corrupted + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceValidateInforom(nvmlDevice_t device); + +/** + * Retrieves the timestamp and the duration of the last flush of the BBX (blackbox) infoROM object during the current run. + * + * For all products with an inforom. + * + * @param device The identifier of the target device + * @param timestamp The start timestamp of the last BBX Flush + * @param durationUs The duration (us) of the last BBX Flush + * + * @return + * - \ref NVML_SUCCESS if \a timestamp and \a durationUs are successfully retrieved + * - \ref NVML_ERROR_NOT_READY if the BBX object has not been flushed yet + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have an infoROM + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetInforomVersion + */ +nvmlReturn_t DECLDIR nvmlDeviceGetLastBBXFlushTime(nvmlDevice_t device, unsigned long long *timestamp, + unsigned long *durationUs); + +/** + * Retrieves the display mode for the device. + * + * For all products. + * + * This method indicates whether a physical display (e.g. monitor) is currently connected to + * any of the device's connectors. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param display Reference in which to return the display mode + * + * @return + * - \ref NVML_SUCCESS if \a display has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a display is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDisplayMode(nvmlDevice_t device, nvmlEnableState_t *display); + +/** + * Retrieves the display active state for the device. + * + * For all products. + * + * This method indicates whether a display is initialized on the device. + * For example whether X Server is attached to this device and has allocated memory for the screen. + * + * Display can be active even when no monitor is physically attached. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param isActive Reference in which to return the display active state + * + * @return + * - \ref NVML_SUCCESS if \a isActive has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a isActive is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDisplayActive(nvmlDevice_t device, nvmlEnableState_t *isActive); + +/** + * Retrieves the persistence mode associated with this device. + * + * For all products. + * For Linux only. + * + * When driver persistence mode is enabled the driver software state is not torn down when the last + * client disconnects. By default this feature is disabled. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param mode Reference in which to return the current driver persistence mode + * + * @return + * - \ref NVML_SUCCESS if \a mode has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetPersistenceMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPersistenceMode(nvmlDevice_t device, nvmlEnableState_t *mode); + +/** + * Retrieves PCI attributes of this device. + * + * For all products. + * + * See \ref nvmlPciInfoExt_v1_t for details on the available PCI info. + * + * @param device The identifier of the target device + * @param pci Reference in which to return the PCI info + * + * @return + * - \ref NVML_SUCCESS if \a pci has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pci is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPciInfoExt(nvmlDevice_t device, nvmlPciInfoExt_t *pci); + +/** + * Retrieves the PCI attributes of this device. + * + * For all products. + * + * See \ref nvmlPciInfo_t for details on the available PCI info. + * + * @param device The identifier of the target device + * @param pci Reference in which to return the PCI info + * + * @return + * - \ref NVML_SUCCESS if \a pci has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pci is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo_v3(nvmlDevice_t device, nvmlPciInfo_t *pci); + +/** + * Retrieves the maximum PCIe link generation possible with this device and system + * + * I.E. for a generation 2 PCIe device attached to a generation 1 PCIe bus the max link generation this function will + * report is generation 1. + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param maxLinkGen Reference in which to return the max PCIe link generation + * + * @return + * - \ref NVML_SUCCESS if \a maxLinkGen has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a maxLinkGen is null + * - \ref NVML_ERROR_NOT_SUPPORTED if PCIe link information is not available + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMaxPcieLinkGeneration(nvmlDevice_t device, unsigned int *maxLinkGen); + +/** + * Retrieves the maximum PCIe link generation supported by this device + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param maxLinkGenDevice Reference in which to return the max PCIe link generation + * + * @return + * - \ref NVML_SUCCESS if \a maxLinkGenDevice has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a maxLinkGenDevice is null + * - \ref NVML_ERROR_NOT_SUPPORTED if PCIe link information is not available + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuMaxPcieLinkGeneration(nvmlDevice_t device, unsigned int *maxLinkGenDevice); + +/** + * Retrieves the maximum PCIe link width possible with this device and system + * + * I.E. for a device with a 16x PCIe bus width attached to a 8x PCIe system bus this function will report + * a max link width of 8. + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param maxLinkWidth Reference in which to return the max PCIe link generation + * + * @return + * - \ref NVML_SUCCESS if \a maxLinkWidth has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a maxLinkWidth is null + * - \ref NVML_ERROR_NOT_SUPPORTED if PCIe link information is not available + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMaxPcieLinkWidth(nvmlDevice_t device, unsigned int *maxLinkWidth); + +/** + * Retrieves the current PCIe link generation + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param currLinkGen Reference in which to return the current PCIe link generation + * + * @return + * - \ref NVML_SUCCESS if \a currLinkGen has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a currLinkGen is null + * - \ref NVML_ERROR_NOT_SUPPORTED if PCIe link information is not available + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCurrPcieLinkGeneration(nvmlDevice_t device, unsigned int *currLinkGen); + +/** + * Retrieves the current PCIe link width + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param currLinkWidth Reference in which to return the current PCIe link generation + * + * @return + * - \ref NVML_SUCCESS if \a currLinkWidth has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a currLinkWidth is null + * - \ref NVML_ERROR_NOT_SUPPORTED if PCIe link information is not available + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCurrPcieLinkWidth(nvmlDevice_t device, unsigned int *currLinkWidth); + +/** + * Retrieve PCIe utilization information. + * This function is querying a byte counter over a 20ms interval and thus is the + * PCIe throughput over that interval. + * + * For Maxwell &tm; or newer fully supported devices. + * + * This method is not supported in virtual machines running virtual GPU (vGPU). + * + * @param device The identifier of the target device + * @param counter The specific counter that should be queried \ref nvmlPcieUtilCounter_t + * @param value Reference in which to return throughput in KB/s + * + * @return + * - \ref NVML_SUCCESS if \a value has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a counter is invalid, or \a value is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPcieThroughput(nvmlDevice_t device, nvmlPcieUtilCounter_t counter, unsigned int *value); + +/** + * Retrieve the PCIe replay counter. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param value Reference in which to return the counter's value + * + * @return + * - \ref NVML_SUCCESS if \a value has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a value is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPcieReplayCounter(nvmlDevice_t device, unsigned int *value); + +/** + * Retrieves the current clock speeds for the device. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlClockType_t for details on available clock information. + * + * @param device The identifier of the target device + * @param type Identify which clock domain to query + * @param clock Reference in which to return the clock speed in MHz + * + * @return + * - \ref NVML_SUCCESS if \a clock has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clock is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device cannot report the specified clock + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetClockInfo(nvmlDevice_t device, nvmlClockType_t type, unsigned int *clock); + +/** + * Retrieves the maximum clock speeds for the device. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlClockType_t for details on available clock information. + * + * \note On GPUs from Fermi family current P0 clocks (reported by \ref nvmlDeviceGetClockInfo) can differ from max clocks + * by few MHz. + * + * @param device The identifier of the target device + * @param type Identify which clock domain to query + * @param clock Reference in which to return the clock speed in MHz + * + * @return + * - \ref NVML_SUCCESS if \a clock has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clock is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device cannot report the specified clock + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMaxClockInfo(nvmlDevice_t device, nvmlClockType_t type, unsigned int *clock); + +/** + * Retrieve the GPCCLK VF offset value + * @param[in] device The identifier of the target device + * @param[out] offset The retrieved GPCCLK VF offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpcClkVfOffset(nvmlDevice_t device, int *offset); + +/** + * @deprecated Applications clocks are deprecated and will be removed in CUDA 14.0. + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetApplicationsClock(nvmlDevice_t device, nvmlClockType_t clockType, unsigned int *clockMHz); + +/** + * @deprecated Applications clocks are deprecated and will be removed in CUDA 14.0. + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetDefaultApplicationsClock(nvmlDevice_t device, nvmlClockType_t clockType, unsigned int *clockMHz); + +/** + * Retrieves the clock speed for the clock specified by the clock type and clock ID. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param clockType Identify which clock domain to query + * @param clockId Identify which clock in the domain to query + * @param clockMHz Reference in which to return the clock in MHz + * + * @return + * - \ref NVML_SUCCESS if \a clockMHz has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clockMHz is NULL or \a clockType is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetClock(nvmlDevice_t device, nvmlClockType_t clockType, nvmlClockId_t clockId, unsigned int *clockMHz); + +/** + * Retrieves the customer defined maximum boost clock speed specified by the given clock type. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param clockType Identify which clock domain to query + * @param clockMHz Reference in which to return the clock in MHz + * + * @return + * - \ref NVML_SUCCESS if \a clockMHz has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clockMHz is NULL or \a clockType is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device or the \a clockType on this device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMaxCustomerBoostClock(nvmlDevice_t device, nvmlClockType_t clockType, unsigned int *clockMHz); + +/** + * Retrieves the list of possible memory clocks that can be used as an argument for \ref nvmlDeviceSetMemoryLockedClocks. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param count Reference in which to provide the \a clocksMHz array size, and + * to return the number of elements + * @param clocksMHz Reference in which to return the clock in MHz + * + * @return + * - \ref NVML_SUCCESS if \a count and \a clocksMHz have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a count is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small (\a count is set to the number of + * required elements) + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetMemoryLockedClocks + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedMemoryClocks(nvmlDevice_t device, unsigned int *count, unsigned int *clocksMHz); + +/** + * Retrieves the list of possible graphics clocks that can be used as an argument for \ref nvmlDeviceSetGpuLockedClocks. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param memoryClockMHz Memory clock for which to return possible graphics clocks + * @param count Reference in which to provide the \a clocksMHz array size, and + * to return the number of elements + * @param clocksMHz Reference in which to return the clocks in MHz + * + * @return + * - \ref NVML_SUCCESS if \a count and \a clocksMHz have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NOT_FOUND if the specified \a memoryClockMHz is not a supported frequency + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clock is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetGpuLockedClocks + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedGraphicsClocks(nvmlDevice_t device, unsigned int memoryClockMHz, unsigned int *count, unsigned int *clocksMHz); + +/** + * Retrieve the current state of Auto Boosted clocks on a device and store it in \a isEnabled + * + * For Kepler &tm; or newer fully supported devices. + * + * Auto Boosted clocks are enabled by default on some hardware, allowing the GPU to run at higher clock rates + * to maximize performance as thermal limits allow. + * + * On Pascal and newer hardware, Auto Aoosted clocks are controlled through application clocks. + * Use \ref nvmlDeviceSetApplicationsClocks and \ref nvmlDeviceResetApplicationsClocks to control Auto Boost + * behavior. + * + * @param device The identifier of the target device + * @param isEnabled Where to store the current state of Auto Boosted clocks of the target device + * @param defaultIsEnabled Where to store the default Auto Boosted clocks behavior of the target device that the device will + * revert to when no applications are using the GPU + * + * @return + * - \ref NVML_SUCCESS If \a isEnabled has been been set with the Auto Boosted clocks state of \a device + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a isEnabled is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support Auto Boosted clocks + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAutoBoostedClocksEnabled(nvmlDevice_t device, nvmlEnableState_t *isEnabled, nvmlEnableState_t *defaultIsEnabled); + +/** + * Retrieves the intended operating speed of the device's fan. + * + * Note: The reported speed is the intended fan speed. If the fan is physically blocked and unable to spin, the + * output will not match the actual fan speed. + * + * For all discrete products with dedicated fans. + * + * The fan speed is expressed as a percentage of the product's maximum noise tolerance fan speed. + * This value may exceed 100% in certain cases. + * + * @param device The identifier of the target device + * @param speed Reference in which to return the fan speed percentage + * + * @return + * - \ref NVML_SUCCESS if \a speed has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a speed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFanSpeed(nvmlDevice_t device, unsigned int *speed); + +/** + * Retrieves the intended operating speed of the device's specified fan. + * + * Note: The reported speed is the intended fan speed. If the fan is physically blocked and unable to spin, the + * output will not match the actual fan speed. + * + * For all discrete products with dedicated fans. + * + * The fan speed is expressed as a percentage of the product's maximum noise tolerance fan speed. + * This value may exceed 100% in certain cases. + * + * @param device The identifier of the target device + * @param fan The index of the target fan, zero indexed. + * @param speed Reference in which to return the fan speed percentage + * + * @return + * - \ref NVML_SUCCESS if \a speed has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a fan is not an acceptable index, or \a speed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan or is newer than Maxwell + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFanSpeed_v2(nvmlDevice_t device, unsigned int fan, unsigned int * speed); + +/** + * Retrieves the intended operating speed in rotations per minute (RPM) of the device's specified fan. + * + * For Maxwell &tm; or newer fully supported devices. + * + * For all discrete products with dedicated fans. + * + * Note: The reported speed is the intended fan speed. If the fan is physically blocked and unable to spin, the + * output will not match the actual fan speed. + * + * @param device The identifier of the target device + * @param fanSpeed Structure specifying the index of the target fan (input) and + * retrieved fan speed value (output) + * + * @return + * - \ref NVML_SUCCESS If everything worked + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid, \a fan is not an acceptable + * index, or \a speed is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the provided version is invalid/unsupported + * - \ref NVML_ERROR_NOT_SUPPORTED If the \a device does not support this feature + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFanSpeedRPM(nvmlDevice_t device, nvmlFanSpeedInfo_t *fanSpeed); + +/** + * Retrieves the intended target speed of the device's specified fan. + * + * Normally, the driver dynamically adjusts the fan based on + * the needs of the GPU. But when user set fan speed using nvmlDeviceSetFanSpeed_v2, + * the driver will attempt to make the fan achieve the setting in + * nvmlDeviceSetFanSpeed_v2. The actual current speed of the fan + * is reported in nvmlDeviceGetFanSpeed_v2. + * + * For all discrete products with dedicated fans. + * + * The fan speed is expressed as a percentage of the product's maximum noise tolerance fan speed. + * This value may exceed 100% in certain cases. + * + * @param device The identifier of the target device + * @param fan The index of the target fan, zero indexed. + * @param targetSpeed Reference in which to return the fan speed percentage + * + * @return + * - \ref NVML_SUCCESS if \a speed has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a fan is not an acceptable index, or \a speed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan or is newer than Maxwell + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTargetFanSpeed(nvmlDevice_t device, unsigned int fan, unsigned int *targetSpeed); + +/** + * Retrieves the min and max fan speed that user can set for the GPU fan. + * + * For all cuda-capable discrete products with fans + * + * @param device The identifier of the target device + * @param minSpeed The minimum speed allowed to set + * @param maxSpeed The maximum speed allowed to set + * + * return + * NVML_SUCCESS if speed has been adjusted + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if device is invalid + * NVML_ERROR_NOT_SUPPORTED if the device does not support this + * (doesn't have fans) + * NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMinMaxFanSpeed(nvmlDevice_t device, unsigned int * minSpeed, + unsigned int * maxSpeed); + +/** + * Gets current fan control policy. + * + * For Maxwell &tm; or newer fully supported devices. + * + * For all cuda-capable discrete products with fans + * + * device The identifier of the target \a device + * policy Reference in which to return the fan control \a policy + * + * return + * NVML_SUCCESS if \a policy has been populated + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a policy is null or the \a fan given doesn't reference + * a fan that exists. + * NVML_ERROR_NOT_SUPPORTED if the \a device is older than Maxwell + * NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFanControlPolicy_v2(nvmlDevice_t device, unsigned int fan, + nvmlFanControlPolicy_t *policy); + +/** + * Retrieves the number of fans on the device. + * + * For all discrete products with dedicated fans. + * + * @param device The identifier of the target device + * @param numFans The number of fans + * + * @return + * - \ref NVML_SUCCESS if \a fan number query was successful + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a numFans is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNumFans(nvmlDevice_t device, unsigned int *numFans); + +/** + * @deprecated Use \ref nvmlDeviceGetTemperatureV instead + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetTemperature(nvmlDevice_t device, nvmlTemperatureSensors_t sensorType, unsigned int *temp); + +/** + * Retrieves the cooler's information. + * Returns a cooler's control signal characteristics. The possible types are restricted, Variable and Toggle. + * See \ref nvmlCoolerControl_t for details on available signal types. + * Returns objects that cooler cools. Targets may be GPU, Memory, Power Supply or All of these. + * See \ref nvmlCoolerTarget_t for details on available targets. + * + * For Maxwell &tm; or newer fully supported devices. + * + * For all discrete products with dedicated fans. + * + * @param[in] device The identifier of the target device + * @param[out] coolerInfo Structure specifying the cooler's control signal characteristics (out) + * and the target that cooler cools (out) + * + * @return + * - \ref NVML_SUCCESS If everything worked + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid, \a signalType or \a target is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the provided version is invalid/unsupported + * - \ref NVML_ERROR_NOT_SUPPORTED If the \a device does not support this feature + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCoolerInfo(nvmlDevice_t device, nvmlCoolerInfo_t *coolerInfo); + +/** + * Structure used to encapsulate temperature info + */ +typedef struct +{ + unsigned int version; + nvmlTemperatureSensors_t sensorType; + int temperature; +} nvmlTemperature_v1_t; + +typedef nvmlTemperature_v1_t nvmlTemperature_t; + +#define nvmlTemperature_v1 NVML_STRUCT_VERSION(Temperature, 1) + +/** + * Retrieves the current temperature readings (in degrees C) for the given device. + * + * For all products. + * + * @param[in] device Target device identifier. + * @param[in,out] temperature Structure specifying the sensor type (input) and retrieved + * temperature value (output). + * + * @return + * - \ref NVML_SUCCESS if \a temp has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a sensorType is invalid or \a temp is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have the specified sensor + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTemperatureV(nvmlDevice_t device, nvmlTemperature_t *temperature); + + +/** + * Retrieves the temperature threshold for the GPU with the specified threshold type in degrees C. + * + * For Kepler &tm; or newer fully supported devices. + * + * See \ref nvmlTemperatureThresholds_t for details on available temperature thresholds. + * + * Note: This API is no longer the preferred interface for retrieving the following temperature thresholds + * on Ada and later architectures: NVML_TEMPERATURE_THRESHOLD_SHUTDOWN, NVML_TEMPERATURE_THRESHOLD_SLOWDOWN, + * NVML_TEMPERATURE_THRESHOLD_MEM_MAX and NVML_TEMPERATURE_THRESHOLD_GPU_MAX. + * + * Support for reading these temperature thresholds for Ada and later architectures would be removed from this + * API in future releases. Please use \ref nvmlDeviceGetFieldValues with NVML_FI_DEV_TEMPERATURE_* fields to retrieve + * temperature thresholds on these architectures. + * + * @param device The identifier of the target device + * @param thresholdType The type of threshold value queried + * @param temp Reference in which to return the temperature reading + * @return + * - \ref NVML_SUCCESS if \a temp has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a thresholdType is invalid or \a temp is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a temperature sensor or is unsupported + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTemperatureThreshold(nvmlDevice_t device, nvmlTemperatureThresholds_t thresholdType, unsigned int *temp); + +/** + * Retrieves the thermal margin temperature (distance to nearest slowdown threshold). + * + * @param[in] device The identifier of the target device + * @param[in,out] marginTempInfo Versioned structure in which to return the temperature reading + * + * @returns + * - \ref NVML_SUCCESS if the margin temperature was retrieved successfully + * - \ref NVML_ERROR_NOT_SUPPORTED if request is not supported on the current platform + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a temperature is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the right versioned structure is not used + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMarginTemperature(nvmlDevice_t device, nvmlMarginTemperature_t *marginTempInfo); + +/** + * Used to execute a list of thermal system instructions. + * + * @param device The identifier of the target device + * @param sensorIndex The index of the thermal sensor + * @param pThermalSettings Reference in which to return the thermal sensor information + * + * @return + * - \ref NVML_SUCCESS if \a pThermalSettings has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pThermalSettings is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetThermalSettings(nvmlDevice_t device, unsigned int sensorIndex, nvmlGpuThermalSettings_t *pThermalSettings); + +/** + * Retrieves the current performance state for the device. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlPstates_t for details on allowed performance states. + * + * @param device The identifier of the target device + * @param pState Reference in which to return the performance state reading + * + * @return + * - \ref NVML_SUCCESS if \a pState has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pState is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPerformanceState(nvmlDevice_t device, nvmlPstates_t *pState); + +/** + * Retrieves current clocks event reasons. + * + * For all fully supported products. + * + * \note More than one bit can be enabled at the same time. Multiple reasons can be affecting clocks at once. + * + * @param device The identifier of the target device + * @param clocksEventReasons Reference in which to return bitmask of active clocks event + * reasons + * + * @return + * - \ref NVML_SUCCESS if \a clocksEventReasons has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clocksEventReasons is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlClocksEventReasons + * @see nvmlDeviceGetSupportedClocksEventReasons + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCurrentClocksEventReasons(nvmlDevice_t device, unsigned long long *clocksEventReasons); + +/** + * @deprecated Use \ref nvmlDeviceGetCurrentClocksEventReasons instead + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetCurrentClocksThrottleReasons(nvmlDevice_t device, unsigned long long *clocksThrottleReasons); + +/** + * Retrieves bitmask of supported clocks event reasons that can be returned by + * \ref nvmlDeviceGetCurrentClocksEventReasons + * + * For all fully supported products. + * + * This method is not supported in virtual machines running virtual GPU (vGPU). + * + * @param device The identifier of the target device + * @param supportedClocksEventReasons Reference in which to return bitmask of supported + * clocks event reasons + * + * @return + * - \ref NVML_SUCCESS if \a supportedClocksEventReasons has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a supportedClocksEventReasons is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlClocksEventReasons + * @see nvmlDeviceGetCurrentClocksEventReasons + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedClocksEventReasons(nvmlDevice_t device, unsigned long long *supportedClocksEventReasons); + +/** + * @deprecated Use \ref nvmlDeviceGetSupportedClocksEventReasons instead + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetSupportedClocksThrottleReasons(nvmlDevice_t device, unsigned long long *supportedClocksThrottleReasons); + +/** + * @deprecated Use \ref nvmlDeviceGetPerformanceState. This function exposes an incorrect generalization. + * + * Retrieve the current performance state for the device. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlPstates_t for details on allowed performance states. + * + * @param device The identifier of the target device + * @param pState Reference in which to return the performance state reading + * + * @return + * - \ref NVML_SUCCESS if \a pState has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pState is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetPowerState(nvmlDevice_t device, nvmlPstates_t *pState); + +/** + * Retrieve performance monitor samples from the associated subdevice. + * + * @param device + * @param pDynamicPstatesInfo + * + * @return + * - \ref NVML_SUCCESS if \a pDynamicPstatesInfo has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pDynamicPstatesInfo is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDynamicPstatesInfo(nvmlDevice_t device, nvmlGpuDynamicPstatesInfo_t *pDynamicPstatesInfo); + +/** + * Retrieve the MemClk (Memory Clock) VF offset value. + * @param[in] device The identifier of the target device + * @param[out] offset The retrieved MemClk VF offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemClkVfOffset(nvmlDevice_t device, int *offset); + +/** + * Retrieve min and max clocks of some clock domain for a given PState + * + * @param device The identifier of the target device + * @param type Clock domain + * @param pstate PState to query + * @param minClockMHz Reference in which to return min clock frequency + * @param maxClockMHz Reference in which to return max clock frequency + * + * @return + * - \ref NVML_SUCCESS if everything worked + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a type or \a pstate are invalid or both + * \a minClockMHz and \a maxClockMHz are NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMinMaxClockOfPState(nvmlDevice_t device, nvmlClockType_t type, nvmlPstates_t pstate, + unsigned int * minClockMHz, unsigned int * maxClockMHz); + +/** + * Get all supported Performance States (P-States) for the device. + * + * The returned array would contain a contiguous list of valid P-States supported by + * the device. If the number of supported P-States is fewer than the size of the array + * supplied missing elements would contain \a NVML_PSTATE_UNKNOWN. + * + * The number of elements in the returned list will never exceed \a NVML_MAX_GPU_PERF_PSTATES. + * + * @param device The identifier of the target device + * @param pstates Container to return the list of performance states + * supported by device + * @param size Size of the supplied \a pstates array in bytes + * + * @return + * - \ref NVML_SUCCESS if \a pstates array has been retrieved + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if the the container supplied was not large enough to + * hold the resulting list + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a pstates is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support performance state readings + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedPerformanceStates(nvmlDevice_t device, + nvmlPstates_t *pstates, unsigned int size); + +/** + * Retrieve the GPCCLK min max VF offset value. + * @param[in] device The identifier of the target device + * @param[out] minOffset The retrieved GPCCLK VF min offset value + * @param[out] maxOffset The retrieved GPCCLK VF max offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpcClkMinMaxVfOffset(nvmlDevice_t device, + int *minOffset, int *maxOffset); + +/** + * Retrieve the MemClk (Memory Clock) min max VF offset value. + * @param[in] device The identifier of the target device + * @param[out] minOffset The retrieved MemClk VF min offset value + * @param[out] maxOffset The retrieved MemClk VF max offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemClkMinMaxVfOffset(nvmlDevice_t device, + int *minOffset, int *maxOffset); + +/** + * Retrieve min, max and current clock offset of some clock domain for a given PState + * + * For Maxwell &tm; or newer fully supported devices. + * + * Note: \ref nvmlDeviceGetGpcClkVfOffset, \ref nvmlDeviceGetMemClkVfOffset, \ref nvmlDeviceGetGpcClkMinMaxVfOffset and + * \ref nvmlDeviceGetMemClkMinMaxVfOffset will be deprecated in a future release. + Use \ref nvmlDeviceGetClockOffsets instead. + * + * @param device The identifier of the target device + * @param info Structure specifying the clock type (input) and the pstate (input) + * retrieved clock offset value (output), min clock offset (output) + * and max clock offset (output) + * + * @return + * - \ref NVML_SUCCESS If everything worked + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a type or \a pstate are invalid or both + * \a minClockOffsetMHz and \a maxClockOffsetMHz are NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the provided version is invalid/unsupported + * - \ref NVML_ERROR_NOT_SUPPORTED If the device does not support this feature + */ +nvmlReturn_t DECLDIR nvmlDeviceGetClockOffsets(nvmlDevice_t device, nvmlClockOffset_t *info); + +/** + * Control current clock offset of some clock domain for a given PState + * + * For Maxwell &tm; or newer fully supported devices. + * + * Requires privileged user. + * + * @param device The identifier of the target device + * @param info Structure specifying the clock type (input), the pstate (input) + * and clock offset value (input) + * + * @return + * - \ref NVML_SUCCESS If everything worked + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_NO_PERMISSION If the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a type or \a pstate are invalid or both + * \a clockOffsetMHz is out of allowed range. + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the provided version is invalid/unsupported + * - \ref NVML_ERROR_NOT_SUPPORTED If the device does not support this feature + */ +nvmlReturn_t DECLDIR nvmlDeviceSetClockOffsets(nvmlDevice_t device, nvmlClockOffset_t *info); + +/** + * Retrieves a performance mode string with all the + * performance modes defined for this device along with their associated + * GPU Clock and Memory Clock values. + * Not all tokens will be reported on all GPUs, and additional tokens + * may be added in the future. + * For backwards compatibility we still provide nvclock and memclock; + * those are the same as nvclockmin and memclockmin. + * + * Note: These clock values take into account the offset + * set by clients through /ref nvmlDeviceSetClockOffsets. + * + * Maximum available Pstate (P15) shows the minimum performance level (0) and vice versa. + * + * Each performance modes are returned as a comma-separated list of + * "token=value" pairs. Each set of performance mode tokens are separated + * by a ";". Valid tokens: + * + * Token Value + * "perf" unsigned int - the Performance level + * "nvclock" unsigned int - the GPU clocks (in MHz) for the perf level + * "nvclockmin" unsigned int - the GPU clocks min (in MHz) for the perf level + * "nvclockmax" unsigned int - the GPU clocks max (in MHz) for the perf level + * "nvclockeditable" unsigned int - if the GPU clock domain is editable for the perf level + * "memclock" unsigned int - the memory clocks (in MHz) for the perf level + * "memclockmin" unsigned int - the memory clocks min (in MHz) for the perf level + * "memclockmax" unsigned int - the memory clocks max (in MHz) for the perf level + * "memclockeditable" unsigned int - if the memory clock domain is editable for the perf level + * "memtransferrate" unsigned int - the memory transfer rate (in MHz) for the perf level + * "memtransferratemin" unsigned int - the memory transfer rate min (in MHz) for the perf level + * "memtransferratemax" unsigned int - the memory transfer rate max (in MHz) for the perf level + * "memtransferrateeditable" unsigned int - if the memory transfer rate is editable for the perf level + * + * Example: + * + * perf=0, nvclock=324, nvclockmin=324, nvclockmax=324, nvclockeditable=0, + * memclock=324, memclockmin=324, memclockmax=324, memclockeditable=0, + * memtransferrate=648, memtransferratemin=648, memtransferratemax=648, + * memtransferrateeditable=0 ; + * perf=1, nvclock=324, nvclockmin=324, nvclockmax=640, nvclockeditable=0, + * memclock=810, memclockmin=810, memclockmax=810, memclockeditable=0, + * memtransferrate=1620, memtransferrate=1620, memtransferrate=1620, + * memtransferrateeditable=0 ; + * + * + * @param device The identifier of the target device + * @param perfModes Reference in which to return the performance level string + * + * @return + * - \ref NVML_SUCCESS if \a perfModes has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a name is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPerformanceModes(nvmlDevice_t device, nvmlDevicePerfModes_t *perfModes); + +/** + * Retrieves a string with the associated current GPU Clock and Memory Clock values. + * + * Not all tokens will be reported on all GPUs, and additional tokens + * may be added in the future. + * + * Note: These clock values take into account the offset + * set by clients through /ref nvmlDeviceSetClockOffsets. + * + * Clock values are returned as a comma-separated list of + * "token=value" pairs. + * Valid tokens: + * + * Token Value + * "perf" unsigned int - the Performance level + * "nvclock" unsigned int - the GPU clocks (in MHz) for the perf level + * "nvclockmin" unsigned int - the GPU clocks min (in MHz) for the perf level + * "nvclockmax" unsigned int - the GPU clocks max (in MHz) for the perf level + * "nvclockeditable" unsigned int - if the GPU clock domain is editable for the perf level + * "memclock" unsigned int - the memory clocks (in MHz) for the perf level + * "memclockmin" unsigned int - the memory clocks min (in MHz) for the perf level + * "memclockmax" unsigned int - the memory clocks max (in MHz) for the perf level + * "memclockeditable" unsigned int - if the memory clock domain is editable for the perf level + * "memtransferrate" unsigned int - the memory transfer rate (in MHz) for the perf level + * "memtransferratemin" unsigned int - the memory transfer rate min (in MHz) for the perf level + * "memtransferratemax" unsigned int - the memory transfer rate max (in MHz) for the perf level + * "memtransferrateeditable" unsigned int - if the memory transfer rate is editable for the perf level + * + * Example: + * + * nvclock=324, nvclockmin=324, nvclockmax=324, nvclockeditable=0, + * memclock=324, memclockmin=324, memclockmax=324, memclockeditable=0, + * memtransferrate=648, memtransferratemin=648, memtransferratemax=648, + * memtransferrateeditable=0 ; + * + * + * @param device The identifier of the target device + * @param currentClockFreqs Reference in which to return the performance level string + * + * @return + * - \ref NVML_SUCCESS if \a currentClockFreqs has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a name is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCurrentClockFreqs(nvmlDevice_t device, nvmlDeviceCurrentClockFreqs_t *currentClockFreqs); + +/** + * @deprecated This API has been deprecated. + * + * Retrieves the power management mode associated with this device. + * + * For products from the Fermi family. + * - Requires \a NVML_INFOROM_POWER version 3.0 or higher. + * + * For from the Kepler or newer families. + * - Does not require \a NVML_INFOROM_POWER object. + * + * This flag indicates whether any power management algorithm is currently active on the device. An + * enabled state does not necessarily mean the device is being actively throttled -- only that + * that the driver will do so if the appropriate conditions are met. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param mode Reference in which to return the current power management mode + * + * @return + * - \ref NVML_SUCCESS if \a mode has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementMode(nvmlDevice_t device, nvmlEnableState_t *mode); + +/** + * Retrieves the power management limit associated with this device. + * + * For Fermi &tm; or newer fully supported devices. + * + * The power limit defines the upper boundary for the card's power draw. If + * the card's total power draw reaches this limit the power management algorithm kicks in. + * + * This reading is only available if power management mode is supported. + * See \ref nvmlDeviceGetPowerManagementMode. + * + * @param device The identifier of the target device + * @param limit Reference in which to return the power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a limit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementLimit(nvmlDevice_t device, unsigned int *limit); + +/** + * Retrieves information about possible values of power management limits on this device. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param minLimit Reference in which to return the minimum power management limit in milliwatts + * @param maxLimit Reference in which to return the maximum power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a minLimit and \a maxLimit have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minLimit or \a maxLimit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetPowerManagementLimit + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementLimitConstraints(nvmlDevice_t device, unsigned int *minLimit, unsigned int *maxLimit); + +/** + * Retrieves default power management limit on this device, in milliwatts. + * Default power management limit is a power management limit that the device boots with. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param defaultLimit Reference in which to return the default power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a defaultLimit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a defaultLimit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementDefaultLimit(nvmlDevice_t device, unsigned int *defaultLimit); + +/** + * Retrieves power usage for this GPU in milliwatts and its associated circuitry (e.g. memory) + * + * For Fermi &tm; or newer fully supported devices. + * + * On Fermi and Kepler GPUs the reading is accurate to within +/- 5% of current power draw. On Ampere + * (except GA100) or newer GPUs, the API returns power averaged over 1 sec interval. On GA100 and + * older architectures, instantaneous power is returned. + * + * See \ref NVML_FI_DEV_POWER_AVERAGE and \ref NVML_FI_DEV_POWER_INSTANT to query specific power + * values. + * + * It is only available if power management mode is supported. See \ref nvmlDeviceGetPowerManagementMode. + * + * @param device The identifier of the target device + * @param power Reference in which to return the power usage information + * + * @return + * - \ref NVML_SUCCESS if \a power has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a power is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support power readings + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerUsage(nvmlDevice_t device, unsigned int *power); + +/** + * Retrieves current power mizer mode on this device. + * + * PowerMizerMode provides a hint to the driver as to how to manage the performance of the GPU. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param powerMizerMode Reference in which to return the power mizer mode + * @param supportedPowerMizerModes Reference in which to return the bitmask of supported power mizer modes on this device. + * The supported modes can be combined using the bitwise OR operator '|'. + * For example, if a device supports all PowerMizer modes, the bitmask would be: + * supportedPowerMizerModes = ((1 << NVML_POWER_MIZER_MODE_ADAPTIVE) | + * (1 << NVML_POWER_MIZER_MODE_PREFER_MAXIMUM_PERFORMANCE) | + * (1 << NVML_POWER_MIZER_MODE_AUTO) | + * (1 << NVML_POWER_MIZER_MODE_PREFER_CONSISTENT_PERFORMANCE)); + * This bitmask can be used to check which power mizer modes are available on the device by performing + * a bitwise AND operation with the specific mode you want to check. + * + * @return + * - \ref NVML_SUCCESS if \a powerMizerMode has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a powerMizerMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support powerMizerMode readings + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ + +nvmlReturn_t DECLDIR nvmlDeviceGetPowerMizerMode_v1(nvmlDevice_t device, nvmlDevicePowerMizerModes_v1_t *powerMizerMode); + +/** + * Sets the new power mizer mode. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param powerMizerMode Reference in which to set the power mizer mode. + * + * @return + * - \ref NVML_SUCCESS if \a powerMizerMode has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a powerMizerMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support powerMizerMode readings + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ + +nvmlReturn_t DECLDIR nvmlDeviceSetPowerMizerMode_v1(nvmlDevice_t device, nvmlDevicePowerMizerModes_v1_t *powerMizerMode); + + +/** + * Retrieves total energy consumption for this GPU in millijoules (mJ) since the driver was last reloaded + * + * For Volta &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param energy Reference in which to return the energy consumption information + * + * @return + * - \ref NVML_SUCCESS if \a energy has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a energy is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support energy readings + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTotalEnergyConsumption(nvmlDevice_t device, unsigned long long *energy); + +/** + * Get the effective power limit that the driver enforces after taking into account all limiters + * + * Note: This can be different from the \ref nvmlDeviceGetPowerManagementLimit if other limits are set elsewhere + * This includes the out of band power limit interface + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The device to communicate with + * @param limit Reference in which to return the power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a limit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEnforcedPowerLimit(nvmlDevice_t device, unsigned int *limit); + +/** + * Retrieves the current GOM and pending GOM (the one that GPU will switch to after reboot). + * + * For GK110 M-class and X-class Tesla &tm; products from the Kepler family. + * Modes \ref NVML_GOM_LOW_DP and \ref NVML_GOM_ALL_ON are supported on fully supported GeForce products. + * Not supported on Quadro ® and Tesla &tm; C-class products. + * + * @param device The identifier of the target device + * @param current Reference in which to return the current GOM + * @param pending Reference in which to return the pending GOM + * + * @return + * - \ref NVML_SUCCESS if \a mode has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a current or \a pending is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlGpuOperationMode_t + * @see nvmlDeviceSetGpuOperationMode + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuOperationMode(nvmlDevice_t device, nvmlGpuOperationMode_t *current, nvmlGpuOperationMode_t *pending); + +/** + * Retrieves the amount of used, free, reserved and total memory available on the device, in bytes. + * The reserved amount is supported on version 2 only. + * + * For all products. + * + * Enabling ECC reduces the amount of total available memory, due to the extra required parity bits. + * Under WDDM most device memory is allocated and managed on startup by Windows. + * + * Under Linux and Windows TCC, the reported amount of used memory is equal to the sum of memory allocated + * by all active channels on the device. + * + * See \ref nvmlMemory_v2_t for details on available memory info. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate + * information, only if the caller has appropriate privileges. Per-instance + * information can be queried by using specific MIG device handles. + * + * @note nvmlDeviceGetMemoryInfo_v2 adds additional memory information. + * + * @note On systems where GPUs are NUMA nodes, the accuracy of FB memory utilization + * provided by this API depends on the memory accounting of the operating system. + * This is because FB memory is managed by the operating system instead of the NVIDIA GPU driver. + * Typically, pages allocated from FB memory are not released even after + * the process terminates to enhance performance. In scenarios where + * the operating system is under memory pressure, it may resort to utilizing FB memory. + * Such actions can result in discrepancies in the accuracy of memory reporting. + * + * @param device The identifier of the target device + * @param memory Reference in which to return the memory information + * + * @return + * - \ref NVML_SUCCESS if \a memory has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryInfo(nvmlDevice_t device, nvmlMemory_t *memory); + +/** + * nvmlDeviceGetMemoryInfo_v2 accounts separately for reserved memory and includes it in the used memory amount. + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryInfo_v2(nvmlDevice_t device, nvmlMemory_v2_t *memory); + +/** + * Retrieves the current compute mode for the device. + * + * For all products. + * + * See \ref nvmlComputeMode_t for details on allowed compute modes. + * + * @param device The identifier of the target device + * @param mode Reference in which to return the current compute mode + * + * @return + * - \ref NVML_SUCCESS if \a mode has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetComputeMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetComputeMode(nvmlDevice_t device, nvmlComputeMode_t *mode); + +/** + * Retrieves the CUDA compute capability of the device. + * + * For all products. + * + * Returns the major and minor compute capability version numbers of the + * device. The major and minor versions are equivalent to the + * CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MINOR and + * CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MAJOR attributes that would be + * returned by CUDA's cuDeviceGetAttribute(). + * + * @param device The identifier of the target device + * @param major Reference in which to return the major CUDA compute capability + * @param minor Reference in which to return the minor CUDA compute capability + * + * @return + * - \ref NVML_SUCCESS if \a major and \a minor have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a major or \a minor are NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCudaComputeCapability(nvmlDevice_t device, int *major, int *minor); + +/** + * Retrieves the current and pending DRAM Encryption modes for the device. + * + * %BLACKWELL_OR_NEWER% + * Only applicable to devices that support DRAM Encryption + * Requires \a NVML_INFOROM_DEN version 1.0 or higher. + * + * Changing DRAM Encryption modes requires a reboot. The "pending" DRAM Encryption mode refers to the target mode following + * the next reboot. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param current Reference in which to return the current DRAM Encryption mode + * @param pending Reference in which to return the pending DRAM Encryption mode + * + * @return + * - \ref NVML_SUCCESS if \a current and \a pending have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or either \a current or \a pending is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the argument version is not supported + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetDramEncryptionMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDramEncryptionMode(nvmlDevice_t device, nvmlDramEncryptionInfo_t *current, nvmlDramEncryptionInfo_t *pending); + +/** + * Set the DRAM Encryption mode for the device. + * + * For Kepler &tm; or newer fully supported devices. + * Only applicable to devices that support DRAM Encryption. + * Requires \a NVML_INFOROM_DEN version 1.0 or higher. + * Requires root/admin permissions. + * + * The DRAM Encryption mode determines whether the GPU enables its DRAM Encryption support. + * + * This operation takes effect after the next reboot. + * + * See \ref nvmlEnableState_t for details on available modes. + * + * @param device The identifier of the target device + * @param dramEncryption The target DRAM Encryption mode + * + * @return + * - \ref NVML_SUCCESS if the DRAM Encryption mode was set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a DRAM Encryption is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the argument version is not supported + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetDramEncryptionMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetDramEncryptionMode(nvmlDevice_t device, const nvmlDramEncryptionInfo_t *dramEncryption); + +/** + * Retrieves the current and pending ECC modes for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * + * Changing ECC modes requires a reboot. The "pending" ECC mode refers to the target mode following + * the next reboot. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param current Reference in which to return the current ECC mode + * @param pending Reference in which to return the pending ECC mode + * + * @return + * - \ref NVML_SUCCESS if \a current and \a pending have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or either \a current or \a pending is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetEccMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEccMode(nvmlDevice_t device, nvmlEnableState_t *current, nvmlEnableState_t *pending); + +/** + * Retrieves the default ECC modes for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param defaultMode Reference in which to return the default ECC mode + * + * @return + * - \ref NVML_SUCCESS if \a current and \a pending have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a default is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetEccMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDefaultEccMode(nvmlDevice_t device, nvmlEnableState_t *defaultMode); + +/** + * Retrieves the device boardId from 0-N. + * Devices with the same boardId indicate GPUs connected to the same PLX. Use in conjunction with + * \ref nvmlDeviceGetMultiGpuBoard() to decide if they are on the same board as well. + * The boardId returned is a unique ID for the current configuration. Uniqueness and ordering across + * reboots and system configurations is not guaranteed (i.e. if a Tesla K40c returns 0x100 and + * the two GPUs on a Tesla K10 in the same system returns 0x200 it is not guaranteed they will + * always return those values but they will always be different from each other). + * + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param boardId Reference in which to return the device's board ID + * + * @return + * - \ref NVML_SUCCESS if \a boardId has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a boardId is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBoardId(nvmlDevice_t device, unsigned int *boardId); + +/** + * Retrieves whether the device is on a Multi-GPU Board + * Devices that are on multi-GPU boards will set \a multiGpuBool to a non-zero value. + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param multiGpuBool Reference in which to return a zero or non-zero value + * to indicate whether the device is on a multi GPU board + * + * @return + * - \ref NVML_SUCCESS if \a multiGpuBool has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a multiGpuBool is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMultiGpuBoard(nvmlDevice_t device, unsigned int *multiGpuBool); + +/** + * Retrieves the total ECC error counts for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * Requires ECC Mode to be enabled. + * + * The total error count is the sum of errors across each of the separate memory systems, i.e. the total set of + * errors across the entire device. + * + * See \ref nvmlMemoryErrorType_t for a description of available error types.\n + * See \ref nvmlEccCounterType_t for a description of available counter types. + * + * @param device The identifier of the target device + * @param errorType Flag that specifies the type of the errors. + * @param counterType Flag that specifies the counter-type of the errors. + * @param eccCounts Reference in which to return the specified ECC errors + * + * @return + * - \ref NVML_SUCCESS if \a eccCounts has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a errorType or \a counterType is invalid, or \a eccCounts is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceClearEccErrorCounts() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTotalEccErrors(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, nvmlEccCounterType_t counterType, unsigned long long *eccCounts); + +/** + * Retrieves the detailed ECC error counts for the device. + * + * @deprecated This API supports only a fixed set of ECC error locations + * On different GPU architectures different locations are supported + * See \ref nvmlDeviceGetMemoryErrorCounter + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 2.0 or higher to report aggregate location-based ECC counts. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher to report all other ECC counts. + * Requires ECC Mode to be enabled. + * + * Detailed errors provide separate ECC counts for specific parts of the memory system. + * + * Reports zero for unsupported ECC error counters when a subset of ECC error counters are supported. + * + * See \ref nvmlMemoryErrorType_t for a description of available bit types.\n + * See \ref nvmlEccCounterType_t for a description of available counter types.\n + * See \ref nvmlEccErrorCounts_t for a description of provided detailed ECC counts. + * + * @param device The identifier of the target device + * @param errorType Flag that specifies the type of the errors. + * @param counterType Flag that specifies the counter-type of the errors. + * @param eccCounts Reference in which to return the specified ECC errors + * + * @return + * - \ref NVML_SUCCESS if \a eccCounts has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a errorType or \a counterType is invalid, or \a eccCounts is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceClearEccErrorCounts() + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetDetailedEccErrors(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, nvmlEccCounterType_t counterType, nvmlEccErrorCounts_t *eccCounts); + +/** + * Retrieves the requested memory error counter for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Requires \a NVML_INFOROM_ECC version 2.0 or higher to report aggregate location-based memory error counts. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher to report all other memory error counts. + * + * Only applicable to devices with ECC. + * + * Requires ECC Mode to be enabled. + * + * @note On MIG-enabled GPUs, per instance information can be queried using specific + * MIG device handles. Per instance information is currently only supported for + * non-DRAM uncorrectable volatile errors. Querying volatile errors using device + * handles is currently not supported. + * + * See \ref nvmlMemoryErrorType_t for a description of available memory error types.\n + * See \ref nvmlEccCounterType_t for a description of available counter types.\n + * See \ref nvmlMemoryLocation_t for a description of available counter locations.\n + * + * @param device The identifier of the target device + * @param errorType Flag that specifies the type of error. + * @param counterType Flag that specifies the counter-type of the errors. + * @param locationType Specifies the location of the counter. + * @param count Reference in which to return the ECC counter + * + * @return + * - \ref NVML_SUCCESS if \a count has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a bitTyp,e \a counterType or \a locationType is + * invalid, or \a count is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support ECC error reporting in the specified memory + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryErrorCounter(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, + nvmlEccCounterType_t counterType, + nvmlMemoryLocation_t locationType, unsigned long long *count); + +/** + * Retrieves the current utilization rates for the device's major subsystems. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlUtilization_t for details on available utilization rates. + * + * \note During driver initialization when ECC is enabled one can see high GPU and Memory Utilization readings. + * This is caused by ECC Memory Scrubbing mechanism that is performed during driver initialization. + * + * @note On MIG-enabled GPUs, querying device utilization rates is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference in which to return the utilization information + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a utilization is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetUtilizationRates(nvmlDevice_t device, nvmlUtilization_t *utilization); + +/** + * Retrieves the current utilization and sampling size in microseconds for the Encoder + * + * For Kepler &tm; or newer fully supported devices. + * + * @note On MIG-enabled GPUs, querying encoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for encoder utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** + * Retrieves the current capacity of the device's encoder, as a percentage of maximum encoder capacity with valid values in the range 0-100. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param encoderQueryType Type of encoder to query + * @param encoderCapacity Reference to an unsigned int for the encoder capacity + * + * @return + * - \ref NVML_SUCCESS if \a encoderCapacity is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a encoderCapacity is NULL, or \a device or \a encoderQueryType + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if device does not support the encoder specified in \a encodeQueryType + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderCapacity (nvmlDevice_t device, nvmlEncoderType_t encoderQueryType, unsigned int *encoderCapacity); + +/** + * Retrieves the current encoder statistics for a given device. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param sessionCount Reference to an unsigned int for count of active encoder sessions + * @param averageFps Reference to an unsigned int for trailing average FPS of all active sessions + * @param averageLatency Reference to an unsigned int for encode latency in microseconds + * + * @return + * - \ref NVML_SUCCESS if \a sessionCount, \a averageFps and \a averageLatency is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount, or \a device or \a averageFps, + * or \a averageLatency is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderStats (nvmlDevice_t device, unsigned int *sessionCount, + unsigned int *averageFps, unsigned int *averageLatency); + +/** + * Retrieves information about active encoder sessions on a target device. + * + * An array of active encoder sessions is returned in the caller-supplied buffer pointed at by \a sessionInfos. The + * array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions + * written to the buffer. + * + * If the supplied buffer is not large enough to accommodate the active session array, the function returns + * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlEncoderSessionInfo_t array required in \a sessionCount. + * To query the number of active encoder sessions, call this function with *sessionCount = 0. The code will return + * NVML_SUCCESS with number of active encoder sessions updated in *sessionCount. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param sessionCount Reference to caller supplied array size, and returns the number of sessions. + * @param sessionInfos Reference in which to return the session information + * + * @return + * - \ref NVML_SUCCESS if \a sessionInfos is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is returned in \a sessionCount + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL. + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderSessions(nvmlDevice_t device, unsigned int *sessionCount, nvmlEncoderSessionInfo_t *sessionInfos); + +/** + * Retrieves the current utilization and sampling size in microseconds for the Decoder + * + * For Kepler &tm; or newer fully supported devices. + * + * @note On MIG-enabled GPUs, querying decoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for decoder utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDecoderUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** + * Retrieves the current utilization and sampling size in microseconds for the JPG + * + * %TURING_OR_NEWER% + * + * @note On MIG-enabled GPUs, querying decoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for jpg utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetJpgUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** + * Retrieves the current utilization and sampling size in microseconds for the OFA (Optical Flow Accelerator) + * + * %TURING_OR_NEWER% + * + * @note On MIG-enabled GPUs, querying decoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for ofa utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetOfaUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** +* Retrieves the active frame buffer capture sessions statistics for a given device. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @param device The identifier of the target device +* @param fbcStats Reference to nvmlFBCStats_t structure containing NvFBC stats +* +* @return +* - \ref NVML_SUCCESS if \a fbcStats is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a fbcStats is NULL +* - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetFBCStats(nvmlDevice_t device, nvmlFBCStats_t *fbcStats); + +/** +* Retrieves information about active frame buffer capture sessions on a target device. +* +* An array of active FBC sessions is returned in the caller-supplied buffer pointed at by \a sessionInfo. The +* array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions +* written to the buffer. +* +* If the supplied buffer is not large enough to accommodate the active session array, the function returns +* NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlFBCSessionInfo_t array required in \a sessionCount. +* To query the number of active FBC sessions, call this function with *sessionCount = 0. The code will return +* NVML_SUCCESS with number of active FBC sessions updated in *sessionCount. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @note hResolution, vResolution, averageFPS and averageLatency data for a FBC session returned in \a sessionInfo may +* be zero if there are no new frames captured since the session started. +* +* @param device The identifier of the target device +* @param sessionCount Reference to caller supplied array size, and returns the number of sessions. +* @param sessionInfo Reference in which to return the session information +* +* @return +* - \ref NVML_SUCCESS if \a sessionInfo is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is returned in \a sessionCount +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL. +* - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetFBCSessions(nvmlDevice_t device, unsigned int *sessionCount, nvmlFBCSessionInfo_t *sessionInfo); + +/** + * Retrieves the current and pending driver model for the device. + * + * For Kepler &tm; or newer fully supported devices. + * For windows only. + * + * On Windows platforms the device driver can run in either WDDM, MCDM or WDM (TCC) modes. If a display is attached + * to the device it must run in WDDM mode. MCDM mode is preferred if a display is not attached. TCC mode is deprecated. + * + * See \ref nvmlDriverModel_t for details on available driver models. + * + * @param device The identifier of the target device + * @param current Reference in which to return the current driver model + * @param pending Reference in which to return the pending driver model + * + * @return + * - \ref NVML_SUCCESS if either \a current and/or \a pending have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or both \a current and \a pending are NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the platform is not windows + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetDriverModel_v2() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDriverModel_v2(nvmlDevice_t device, nvmlDriverModel_t *current, nvmlDriverModel_t *pending); + +/** + * Get VBIOS version of the device. + * + * For all products. + * + * The VBIOS version may change from time to time. It will not exceed 32 characters in length + * (including the NULL terminator). See \ref nvmlConstants::NVML_DEVICE_VBIOS_VERSION_BUFFER_SIZE. + * + * @param device The identifier of the target device + * @param version Reference to which to return the VBIOS version + * @param length The maximum allowed length of the string returned in \a version + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a version is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVbiosVersion(nvmlDevice_t device, char *version, unsigned int length); + +/** + * Get Bridge Chip Information for all the bridge chips on the board. + * + * For all fully supported products. + * Only applicable to multi-GPU products. + * + * @param device The identifier of the target device + * @param bridgeHierarchy Reference to the returned bridge chip Hierarchy + * + * @return + * - \ref NVML_SUCCESS if bridge chip exists + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a bridgeInfo is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if bridge chip not supported on the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBridgeChipInfo(nvmlDevice_t device, nvmlBridgeChipHierarchy_t *bridgeHierarchy); + +/** + * Get information about processes with a compute context on a device + * + * For Fermi &tm; or newer fully supported devices. + * + * This function returns information only about compute running processes (e.g. CUDA application which have + * active context). Any graphics applications (e.g. using OpenGL, DirectX) won't be listed by this function. + * + * To query the current number of running compute processes, call this function with *infoCount = 0. The + * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call + * \a infos is allowed to be NULL. + * + * The usedGpuMemory field returned is all of the memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a infos table in case new compute processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in vGPU Host virtualization mode. + * + * @param device The device handle or MIG device handle + * @param infoCount Reference in which to provide the \a infos array size, and + * to return the number of returned elements + * @param infos Reference in which to return the process information + * + * @return + * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small + * \a infoCount will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see \ref nvmlSystemGetProcessName + */ +nvmlReturn_t DECLDIR nvmlDeviceGetComputeRunningProcesses_v3(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); + +/** + * Get information about processes with a graphics context on a device + * + * For Kepler &tm; or newer fully supported devices. + * + * This function returns information only about graphics based processes + * (eg. applications using OpenGL, DirectX) + * + * To query the current number of running graphics processes, call this function with *infoCount = 0. The + * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call + * \a infos is allowed to be NULL. + * + * The usedGpuMemory field returned is all of the memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a infos table in case new graphics processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in vGPU Host virtualization mode. + * + * @param device The device handle or MIG device handle + * @param infoCount Reference in which to provide the \a infos array size, and + * to return the number of returned elements + * @param infos Reference in which to return the process information + * + * @return + * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small + * \a infoCount will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see \ref nvmlSystemGetProcessName + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGraphicsRunningProcesses_v3(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); + +/** + * Get information about processes with a Multi-Process Service (MPS) compute context on a device + * + * For Volta &tm; or newer fully supported devices. + * + * This function returns information only about compute running processes (e.g. CUDA application which have + * active context) utilizing MPS. Any graphics applications (e.g. using OpenGL, DirectX) won't be listed by + * this function. + * + * To query the current number of running compute processes, call this function with *infoCount = 0. The + * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call + * \a infos is allowed to be NULL. + * + * The usedGpuMemory field returned is all of the memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a infos table in case new compute processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in vGPU Host virtualization mode. + * + * @param device The device handle or MIG device handle + * @param infoCount Reference in which to provide the \a infos array size, and + * to return the number of returned elements + * @param infos Reference in which to return the process information + * + * @return + * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small + * \a infoCount will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see \ref nvmlSystemGetProcessName + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMPSComputeRunningProcesses_v3(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); + +/** + * Get information about running processes on a device for input context + * + * For Hopper &tm; or newer fully supported devices. + * + * This function returns information only about running processes (e.g. CUDA application which have + * active context). + * + * To determine the size of the \a plist->procArray array to allocate, call the function with + * \a plist->numProcArrayEntries set to zero and \a plist->procArray set to NULL. The return + * code will be either NVML_ERROR_INSUFFICIENT_SIZE (if there are valid processes of type + * \a plist->mode to report on, in which case the \a plist->numProcArrayEntries field will + * indicate the required number of entries in the array) or NVML_SUCCESS (if no processes of type + * \a plist->mode exist). + * + * The usedGpuMemory field returned is all of the memory used by the application. + * The usedGpuCcProtectedMemory field returned is all of the protected memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a plist->procArray table in case new processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in + * vGPU Host virtualization mode. + * Protected memory usage is currently not available in MIG mode and in windows. + * + * @param device The device handle or MIG device handle + * @param plist Reference in which to process detail list + * \a plist->version The api version + * \a plist->mode The process mode + * \a plist->procArray Reference in which to return the process information + * \a plist->numProcArrayEntries Proc array size of returned entries + * + * @return + * - \ref NVML_SUCCESS if \a plist->numprocArrayEntries and \a plist->procArray have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a plist->numprocArrayEntries indicates that the \a plist->procArray is too small + * \a plist->numprocArrayEntries will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a plist is NULL, \a plist->version is invalid, + * \a plist->mode is invalid, + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRunningProcessDetailList(nvmlDevice_t device, nvmlProcessDetailList_t *plist); + +/** + * Check if the GPU devices are on the same physical board. + * + * For all fully supported products. + * + * @param device1 The first GPU device + * @param device2 The second GPU device + * @param onSameBoard Reference in which to return the status. + * Non-zero indicates that the GPUs are on the same board. + * + * @return + * - \ref NVML_SUCCESS if \a onSameBoard has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a dev1 or \a dev2 are invalid or \a onSameBoard is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this check is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the either GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceOnSameBoard(nvmlDevice_t device1, nvmlDevice_t device2, int *onSameBoard); + +/** + * Retrieves the root/admin permissions on the target API. See \a nvmlRestrictedAPI_t for the list of supported APIs. + * If an API is restricted only root users can call that API. See \a nvmlDeviceSetAPIRestriction to change current permissions. + * + * For all fully supported products. + * + * @param device The identifier of the target device + * @param apiType Target API type for this operation + * @param isRestricted Reference in which to return the current restriction + * NVML_FEATURE_ENABLED indicates that the API is root-only + * NVML_FEATURE_DISABLED indicates that the API is accessible to all users + * + * @return + * - \ref NVML_SUCCESS if \a isRestricted has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a apiType incorrect or \a isRestricted is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device or the device does not support + * the feature that is being queried (E.G. Enabling/disabling Auto Boosted clocks is + * not supported by the device) + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlRestrictedAPI_t + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAPIRestriction(nvmlDevice_t device, nvmlRestrictedAPI_t apiType, nvmlEnableState_t *isRestricted); + +/** + * Gets recent samples for the GPU. + * + * For Kepler &tm; or newer fully supported devices. + * + * Based on type, this method can be used to fetch the power, utilization or clock samples maintained in the buffer by + * the driver. + * + * Power, Utilization and Clock samples are returned as type "unsigned int" for the union nvmlValue_t. + * + * To get the size of samples that user needs to allocate, the method is invoked with samples set to NULL. + * The returned samplesCount will provide the number of samples that can be queried. The user needs to + * allocate the buffer with size as samplesCount * sizeof(nvmlSample_t). + * + * lastSeenTimeStamp represents CPU timestamp in microseconds. Set it to 0 to fetch all the samples maintained by the + * underlying buffer. Set lastSeenTimeStamp to one of the timeStamps retrieved from the date of the previous query + * to get more recent samples. + * + * This method fetches the number of entries which can be accommodated in the provided samples array, and the + * reference samplesCount is updated to indicate how many samples were actually retrieved. The advantage of using this + * method for samples in contrast to polling via existing methods is to get get higher frequency data at lower polling cost. + * + * @note On MIG-enabled GPUs, querying the following sample types, NVML_GPU_UTILIZATION_SAMPLES, NVML_MEMORY_UTILIZATION_SAMPLES + * NVML_ENC_UTILIZATION_SAMPLES and NVML_DEC_UTILIZATION_SAMPLES, is not currently supported. + * + * @param device The identifier for the target device + * @param type Type of sampling event + * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. + * @param sampleValType Output parameter to represent the type of sample value as described in nvmlSampleVal_t + * @param sampleCount Reference to provide the number of elements which can be queried in samples array + * @param samples Reference in which samples are returned + + * @return + * - \ref NVML_SUCCESS if samples are successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a samplesCount is NULL or + * reference to \a sampleCount is 0 for non null \a samples + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_FOUND if sample entries are not found + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSamples(nvmlDevice_t device, nvmlSamplingType_t type, unsigned long long lastSeenTimeStamp, + nvmlValueType_t *sampleValType, unsigned int *sampleCount, nvmlSample_t *samples); + +/** + * Gets Total, Available and Used size of BAR1 memory. + * + * BAR1 is used to map the FB (device memory) so that it can be directly accessed by the CPU or by 3rd party + * devices (peer-to-peer on the PCIE bus). + * + * @note In MIG mode, if device handle is provided, the API returns aggregate + * information, only if the caller has appropriate privileges. Per-instance + * information can be queried by using specific MIG device handles. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param bar1Memory Reference in which BAR1 memory + * information is returned. + * + * @return + * - \ref NVML_SUCCESS if BAR1 memory is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a bar1Memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBAR1MemoryInfo(nvmlDevice_t device, nvmlBAR1Memory_t *bar1Memory); + +/** + * @deprecated Use \ref nvmlDeviceGetFieldValues to query this data. + * This API will be removed in CUDA 14.0. + * + * Translations are as follows: + + * + * NVML_PERF_POLICY_POWER -> NVML_FI_DEV_CLOCKS_EVENT_REASON_SW_POWER_CAP + * NVML_PERF_POLICY_THERMAL -> NVML_FI_DEV_CLOCKS_EVENT_REASON_SW_THERM_SLOWDOWN + * NVML_PERF_POLICY_SYNC_BOOST -> NVML_FI_DEV_CLOCKS_EVENT_REASON_SYNC_BOOST + * NVML_PERF_POLICY_BOARD_LIMIT -> NVML_FI_DEV_PERF_POLICY_BOARD_LIMIT + * NVML_PERF_POLICY_LOW_UTILIZATION -> NVML_FI_DEV_PERF_POLICY_LOW_UTILIZATION + * NVML_PERF_POLICY_RELIABILITY -> NVML_FI_DEV_PERF_POLICY_RELIABILITY + * NVML_PERF_POLICY_TOTAL_APP_CLOCKS -> DEPRECATED, Do not use + * NVML_PERF_POLICY_TOTAL_BASE_CLOCKS -> NVML_FI_DEV_PERF_POLICY_TOTAL_BASE_CLOCKS + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetViolationStatus(nvmlDevice_t device, nvmlPerfPolicyType_t perfPolicyType, nvmlViolationTime_t *violTime); + +/** + * Gets the device's interrupt number + * + * @param device The identifier of the target device + * @param irqNum The interrupt number associated with the specified device + * + * @return + * - \ref NVML_SUCCESS if irq number is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a irqNum is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetIrqNum(nvmlDevice_t device, unsigned int *irqNum); + +/** + * Gets the device's core count + * + * @note On MIG-enabled GPUs, querying the device's core count is currently not supported using this API. + * Please use \ref nvmlDeviceGetGpuInstanceProfileInfo to fetch the MIG device's core count. + * + * @param device The identifier of the target device + * @param numCores The number of cores for the specified device + * + * @return + * - \ref NVML_SUCCESS if GPU core count is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a numCores is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device or a mig device. + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNumGpuCores(nvmlDevice_t device, unsigned int *numCores); + +/** + * Gets the devices power source + * + * @param device The identifier of the target device + * @param powerSource The power source of the device + * + * @return + * - \ref NVML_SUCCESS if the current power source was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a powerSource is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerSource(nvmlDevice_t device, nvmlPowerSource_t *powerSource); + +/** + * Gets the device's memory bus width + * + * @param device The identifier of the target device + * @param busWidth The devices's memory bus width + * + * @return + * - \ref NVML_SUCCESS if the memory bus width is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a busWidth is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryBusWidth(nvmlDevice_t device, unsigned int *busWidth); + +/** + * Gets the device's PCIE Max Link speed in MBPS + * + * @param device The identifier of the target device + * @param maxSpeed The devices's PCIE Max Link speed in MBPS + * + * @return + * - \ref NVML_SUCCESS if PCIe Max Link Speed is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a maxSpeed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPcieLinkMaxSpeed(nvmlDevice_t device, unsigned int *maxSpeed); + +/** + * Gets the device's PCIe Link speed in Mbps + * + * @param device The identifier of the target device + * @param pcieSpeed The devices's PCIe Max Link speed in Mbps + * + * @return + * - \ref NVML_SUCCESS if \a pcieSpeed has been retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pcieSpeed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support PCIe speed getting + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPcieSpeed(nvmlDevice_t device, unsigned int *pcieSpeed); + +/** + * Gets the device's Adaptive Clock status + * + * @param device The identifier of the target device + * @param adaptiveClockStatus The current adaptive clocking status, either + * NVML_ADAPTIVE_CLOCKING_INFO_STATUS_DISABLED + * or NVML_ADAPTIVE_CLOCKING_INFO_STATUS_ENABLED + * + * @return + * - \ref NVML_SUCCESS if the current adaptive clocking status is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a adaptiveClockStatus is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAdaptiveClockInfoStatus(nvmlDevice_t device, unsigned int *adaptiveClockStatus); + +/** + * Get the type of the GPU Bus (PCIe, PCI, ...) + * + * @param device The identifier of the target device + * @param type The PCI Bus type + * + * return + * - \ref NVML_SUCCESS if the bus \a type is successfully retreived + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a type is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBusType(nvmlDevice_t device, nvmlBusType_t *type); + + + /** + * @deprecated Will be deprecated in a future release. Use \ref nvmlDeviceGetGpuFabricInfoV instead + * + * Get fabric information associated with the device. + * + * For Hopper &tm; or newer fully supported devices. + * + * On Hopper + NVSwitch systems, GPU is registered with the NVIDIA Fabric Manager + * Upon successful registration, the GPU is added to the NVLink fabric to enable + * peer-to-peer communication. + * This API reports the current state of the GPU in the NVLink fabric + * along with other useful information. + * + * + * @param device The identifier of the target device + * @param gpuFabricInfo Information about GPU fabric state + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support gpu fabric + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetGpuFabricInfo(nvmlDevice_t device, nvmlGpuFabricInfo_t *gpuFabricInfo); + +/** +* Versioned wrapper around \ref nvmlDeviceGetGpuFabricInfo that accepts a versioned +* \ref nvmlGpuFabricInfo_v2_t or later output structure. +* +* @note The caller must set the \ref nvmlGpuFabricInfoV_t.version field to the +* appropriate version prior to calling this function. For example: +* \code +* nvmlGpuFabricInfoV_t fabricInfo = +* { .version = nvmlGpuFabricInfo_v2 }; +* nvmlReturn_t result = nvmlDeviceGetGpuFabricInfoV(device,&fabricInfo); +* \endcode +* +* For Hopper &tm; or newer fully supported devices. +* +* @param device The identifier of the target device +* @param gpuFabricInfo Information about GPU fabric state +* +* @return +* - \ref NVML_SUCCESS Upon success +* - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support gpu fabric +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuFabricInfoV(nvmlDevice_t device, + nvmlGpuFabricInfoV_t *gpuFabricInfo); + +/** + * Get Conf Computing System capabilities. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param capabilities System CC capabilities + * + * @return + * - \ref NVML_SUCCESS if \a capabilities were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a capabilities is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeCapabilities(nvmlConfComputeSystemCaps_t *capabilities); + +/** + * Get Conf Computing System State. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param state System CC State + * + * @return + * - \ref NVML_SUCCESS if \a state were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a state is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeState(nvmlConfComputeSystemState_t *state); + +/** + * Get Conf Computing Protected and Unprotected Memory Sizes. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device Device handle + * @param memInfo Protected/Unprotected Memory sizes + * + * @return + * - \ref NVML_SUCCESS if \a memInfo were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a memInfo or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeMemSizeInfo(nvmlDevice_t device, nvmlConfComputeMemSizeInfo_t *memInfo); + +/** + * Get Conf Computing GPUs ready state. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param isAcceptingWork Returns GPU current work accepting state, + * NVML_CC_ACCEPTING_CLIENT_REQUESTS_TRUE or + * NVML_CC_ACCEPTING_CLIENT_REQUESTS_FALSE + * + * return + * - \ref NVML_SUCCESS if \a current GPUs ready state were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a isAcceptingWork is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeGpusReadyState(unsigned int *isAcceptingWork); + +/** + * Get Conf Computing protected memory usage. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param memory Reference in which to return the memory information + * + * @return + * - \ref NVML_SUCCESS if \a memory has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeProtectedMemoryUsage(nvmlDevice_t device, nvmlMemory_t *memory); + +/** + * Get Conf Computing GPU certificate details. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param gpuCert Reference in which to return the gpu certificate information + * + * @return + * - \ref NVML_SUCCESS if \a gpu certificate info has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeGpuCertificate(nvmlDevice_t device, + nvmlConfComputeGpuCertificate_t *gpuCert); + +/** + * Get Conf Computing GPU attestation report. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param gpuAtstReport Reference in which to return the gpu attestation report + * + * @return + * - \ref NVML_SUCCESS if \a gpu attestation report has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeGpuAttestationReport(nvmlDevice_t device, + nvmlConfComputeGpuAttestationReport_t *gpuAtstReport); +/** + * Get Conf Computing key rotation threshold detail. + * + * For Hopper &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param pKeyRotationThrInfo Reference in which to return the key rotation threshold data + * + * @return + * - \ref NVML_SUCCESS if \a gpu key rotation threshold info has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeKeyRotationThresholdInfo( + nvmlConfComputeGetKeyRotationThresholdInfo_t *pKeyRotationThrInfo); + +/** + * Set Conf Computing Unprotected Memory Size. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device Device Handle + * @param sizeKiB Unprotected Memory size to be set in KiB + * + * @return + * - \ref NVML_SUCCESS if \a sizeKiB successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlDeviceSetConfComputeUnprotectedMemSize(nvmlDevice_t device, unsigned long long sizeKiB); + +/** + * Set Conf Computing GPUs ready state. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param isAcceptingWork GPU accepting new work, NVML_CC_ACCEPTING_CLIENT_REQUESTS_TRUE or + * NVML_CC_ACCEPTING_CLIENT_REQUESTS_FALSE + * + * return + * - \ref NVML_SUCCESS if \a current GPUs ready state is successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a isAcceptingWork is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemSetConfComputeGpusReadyState(unsigned int isAcceptingWork); + +/** + * Set Conf Computing key rotation threshold. + * + * For Hopper &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * This function is to set the confidential compute key rotation threshold parameters. + * \a pKeyRotationThrInfo->maxAttackerAdvantage should be in the range from + * NVML_CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MIN to NVML_CC_KEY_ROTATION_THRESHOLD_ATTACKER_ADVANTAGE_MAX. + * Default value is 60. + * + * @param pKeyRotationThrInfo Reference to the key rotation threshold data + * + * @return + * - \ref NVML_SUCCESS if \a key rotation threashold max attacker advantage has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_INVALID_STATE if confidential compute GPU ready state is enabled + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlSystemSetConfComputeKeyRotationThresholdInfo( + nvmlConfComputeSetKeyRotationThresholdInfo_t *pKeyRotationThrInfo); + +/** + * Get Conf Computing System Settings. + * + * For Hopper &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param settings System CC settings + * + * @return + * - \ref NVML_SUCCESS If the query is success + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid or \a counters is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST If the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the provided version is invalid/unsupported + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeSettings(nvmlSystemConfComputeSettings_t *settings); + +/** + * Retrieve GSP firmware version. + * + * The caller passes in buffer via \a version and corresponding GSP firmware numbered version + * is returned with the same parameter in string format. + * + * @param device Device handle + * @param version The retrieved GSP firmware version + * + * @return + * - \ref NVML_SUCCESS if GSP firmware version is sucessfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or GSP \a version pointer is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if GSP firmware is not enabled for GPU + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGspFirmwareVersion(nvmlDevice_t device, char *version); + +/** + * Retrieve GSP firmware mode. + * + * The caller passes in integer pointers. GSP firmware enablement and default mode information is returned with + * corresponding parameters. The return value in \a isEnabled and \a defaultMode should be treated as boolean. + * + * @param device Device handle + * @param isEnabled Pointer to specify if GSP firmware is enabled + * @param defaultMode Pointer to specify if GSP firmware is supported by default on \a device + * + * @return + * - \ref NVML_SUCCESS if GSP firmware mode is sucessfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or any of \a isEnabled or \a defaultMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if GSP firmware is not enabled for GPU + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGspFirmwareMode(nvmlDevice_t device, unsigned int *isEnabled, unsigned int *defaultMode); + +/** + * Get SRAM ECC error status of this device. + * + * For Ampere &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * See \ref nvmlEccSramErrorStatus_v1_t for more information on the struct. + * + * @param device The identifier of the target device + * @param status Returns SRAM ECC error status + * + * @return + * - \ref NVML_SUCCESS If \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid or \a counters is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST If the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a nvmlEccSramErrorStatus_t is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSramEccErrorStatus(nvmlDevice_t device, + nvmlEccSramErrorStatus_t *status); + +/** + * Set new power limit of this device. + * + * For Kepler &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * See \ref nvmlDeviceGetPowerManagementLimitConstraints to check the allowed ranges of values. + * + * See \ref nvmlPowerValue_v2_t for more information on the struct. + * + * \note Limit is not persistent across reboots or driver unloads. + * Enable persistent mode to prevent driver from unloading when no application is using the device. + * + * This API replaces nvmlDeviceSetPowerManagementLimit. It can be used as a drop-in replacement for the older version. + * + * @param device The identifier of the target device + * @param powerValue Power management limit in milliwatts to set + * + * @return + * - \ref NVML_SUCCESS if \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a powerValue is NULL or contains invalid values + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see NVML_FI_DEV_POWER_AVERAGE + * @see NVML_FI_DEV_POWER_INSTANT + * @see NVML_FI_DEV_POWER_MIN_LIMIT + * @see NVML_FI_DEV_POWER_MAX_LIMIT + * @see NVML_FI_DEV_POWER_CURRENT_LIMIT + */ +nvmlReturn_t DECLDIR nvmlDeviceSetPowerManagementLimit_v2(nvmlDevice_t device, nvmlPowerValue_v2_t *powerValue); + +/** + * @} // @defgroup nvmlDeviceQueries Device Queries + */ + +/** @addtogroup nvmlAccountingStats + * @{ + */ + +/** + * Queries the state of per process accounting mode. + * + * For Kepler &tm; or newer fully supported devices. + * + * See \ref nvmlDeviceGetAccountingStats for more details. + * See \ref nvmlDeviceSetAccountingMode + * + * @param device The identifier of the target device + * @param mode Reference in which to return the current accounting mode + * + * @return + * - \ref NVML_SUCCESS if the mode has been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode are NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingMode(nvmlDevice_t device, nvmlEnableState_t *mode); + +/** + * Queries process's accounting stats. + * + * For Kepler &tm; or newer fully supported devices. + * + * Accounting stats capture GPU utilization and other statistics across the lifetime of a process. + * Accounting stats can be queried during life time of the process and after its termination. + * The time field in \ref nvmlAccountingStats_t is reported as 0 during the lifetime of the process and + * updated to actual running time after its termination. + * Accounting stats are kept in a circular buffer, newly created processes overwrite information about old + * processes. + * + * See \ref nvmlAccountingStats_t for description of each returned metric. + * List of processes that can be queried can be retrieved from \ref nvmlDeviceGetAccountingPids. + * + * @note Accounting Mode needs to be on. See \ref nvmlDeviceGetAccountingMode. + * @note Only compute and graphics applications stats can be queried. Monitoring applications stats can't be + * queried since they don't contribute to GPU utilization. + * @note In case of pid collision stats of only the latest process (that terminated last) will be reported + * + * @warning On Kepler devices per process statistics are accurate only if there's one process running on a GPU. + * + * @param device The identifier of the target device + * @param pid Process Id of the target process to query stats for + * @param stats Reference in which to return the process's accounting stats + * + * @return + * - \ref NVML_SUCCESS if stats have been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a stats are NULL + * - \ref NVML_ERROR_NOT_FOUND if process stats were not found + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device doesn't support this feature or accounting mode is disabled + * or on vGPU host. + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetAccountingBufferSize + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingStats(nvmlDevice_t device, unsigned int pid, nvmlAccountingStats_t *stats); + +/** + * Queries list of processes that can be queried for accounting stats. The list of processes returned + * can be in running or terminated state. + * + * For Kepler &tm; or newer fully supported devices. + * + * To query the number of processes under Accounting Mode, call this function with *count = 0 and pids=NULL. + * The return code will be NVML_ERROR_INSUFFICIENT_SIZE with an updated count value indicating the number of processes. + * + * For more details see \ref nvmlDeviceGetAccountingStats. + * + * @note In case of PID collision some processes might not be accessible before the circular buffer is full. + * + * @param device The identifier of the target device + * @param count Reference in which to provide the \a pids array size, and + * to return the number of elements ready to be queried + * @param pids Reference in which to return list of process ids + * + * @return + * - \ref NVML_SUCCESS if pids were successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a count is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device doesn't support this feature or accounting mode is disabled + * or on vGPU host. + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small (\a count is set to + * expected value) + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetAccountingBufferSize + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingPids(nvmlDevice_t device, unsigned int *count, unsigned int *pids); + +/** + * Returns the number of processes that the circular buffer with accounting pids can hold. + * + * For Kepler &tm; or newer fully supported devices. + * + * This is the maximum number of processes that accounting information will be stored for before information + * about oldest processes will get overwritten by information about new processes. + * + * @param device The identifier of the target device + * @param bufferSize Reference in which to provide the size (in number of elements) + * of the circular buffer for accounting stats. + * + * @return + * - \ref NVML_SUCCESS if buffer size was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a bufferSize is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetAccountingStats + * @see nvmlDeviceGetAccountingPids + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingBufferSize(nvmlDevice_t device, unsigned int *bufferSize); + +/** @} */ + +/** @addtogroup nvmlDeviceQueries + * @{ + */ + +/** + * Returns the list of retired pages by source, including pages that are pending retirement + * The address information provided from this API is the hardware address of the page that was retired. Note + * that this does not match the virtual address used in CUDA, but will match the address information in Xid 63 + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param cause Filter page addresses by cause of retirement + * @param pageCount Reference in which to provide the \a addresses buffer size, and + * to return the number of retired pages that match \a cause + * Set to 0 to query the size without allocating an \a addresses buffer + * @param addresses Buffer to write the page addresses into + * + * @return + * - \ref NVML_SUCCESS if \a pageCount was populated and \a addresses was filled + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a pageCount indicates the buffer is not large enough to store all the + * matching page addresses. \a pageCount is set to the needed size. + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a pageCount is NULL, \a cause is invalid, or + * \a addresses is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPages(nvmlDevice_t device, nvmlPageRetirementCause_t cause, + unsigned int *pageCount, unsigned long long *addresses); + +/** + * Returns the list of retired pages by source, including pages that are pending retirement + * The address information provided from this API is the hardware address of the page that was retired. Note + * that this does not match the virtual address used in CUDA, but will match the address information in Xid 63 + * + * \note nvmlDeviceGetRetiredPages_v2 adds an additional timestamps parameter to return the time of each page's + * retirement. This is supported for Pascal and newer architecture. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param cause Filter page addresses by cause of retirement + * @param pageCount Reference in which to provide the \a addresses buffer size, and + * to return the number of retired pages that match \a cause + * Set to 0 to query the size without allocating an \a addresses buffer + * @param addresses Buffer to write the page addresses into + * @param timestamps Buffer to write the timestamps of page retirement, additional for _v2 + * + * @return + * - \ref NVML_SUCCESS if \a pageCount was populated and \a addresses was filled + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a pageCount indicates the buffer is not large enough to store all the + * matching page addresses. \a pageCount is set to the needed size. + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a pageCount is NULL, \a cause is invalid, or + * \a addresses is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPages_v2(nvmlDevice_t device, nvmlPageRetirementCause_t cause, + unsigned int *pageCount, unsigned long long *addresses, unsigned long long *timestamps); + +/** + * Check if any pages are pending retirement and need a reboot to fully retire. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param isPending Reference in which to return the pending status + * + * @return + * - \ref NVML_SUCCESS if \a isPending was populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a isPending is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPagesPendingStatus(nvmlDevice_t device, nvmlEnableState_t *isPending); + +/** + * Get number of remapped rows. The number of rows reported will be based on + * the cause of the remapping. isPending indicates whether or not there are + * pending remappings. A reset will be required to actually remap the row. + * failureOccurred will be set if a row remapping ever failed in the past. A + * pending remapping won't affect future work on the GPU since + * error-containment and dynamic page blacklisting will take care of that. + * + * @note On MIG-enabled GPUs with active instances, querying the number of + * remapped rows is not supported + * + * For Ampere &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param corrRows Reference for number of rows remapped due to correctable errors + * @param uncRows Reference for number of rows remapped due to uncorrectable errors + * @param isPending Reference for whether or not remappings are pending + * @param failureOccurred Reference that is set when a remapping has failed in the past + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a corrRows, \a uncRows, \a isPending or \a failureOccurred is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN Unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRemappedRows(nvmlDevice_t device, unsigned int *corrRows, unsigned int *uncRows, + unsigned int *isPending, unsigned int *failureOccurred); + +/** + * Get the row remapper histogram. Returns the remap availability for each bank + * on the GPU. + * + * @param device Device handle + * @param values Histogram values + * + * @return + * - \ref NVML_SUCCESS On success + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRowRemapperHistogram(nvmlDevice_t device, nvmlRowRemapperHistogramValues_t *values); + +/** + * Get architecture for device + * + * @param device The identifier of the target device + * @param arch Reference where architecture is returned, if call successful. + * Set to NVML_DEVICE_ARCH_* upon success + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device or \a arch (output refererence) are invalid + */ +nvmlReturn_t DECLDIR nvmlDeviceGetArchitecture(nvmlDevice_t device, nvmlDeviceArchitecture_t *arch); + +/** + * Retrieves the frequency monitor fault status for the device. + * + * For Ampere &tm; or newer fully supported devices. + * Requires root user. + * + * See \ref nvmlClkMonStatus_t for details on decoding the status output. + * + * @param device The identifier of the target device + * @param status Reference in which to return the clkmon fault status + * + * @return + * - \ref NVML_SUCCESS if \a status has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a status is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetClkMonStatus() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetClkMonStatus(nvmlDevice_t device, nvmlClkMonStatus_t *status); + +/** + * Retrieves the current utilization and process ID + * + * For Maxwell &tm; or newer fully supported devices. + * + * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, and video decoder for processes running. + * Utilization values are returned as an array of utilization sample structures in the caller-supplied buffer pointed at + * by \a utilization. One utilization sample structure is returned per process running, that had some non-zero utilization + * during the last sample period. It includes the CPU timestamp at which the samples were recorded. Individual utilization values + * are returned as "unsigned int" values. If no valid sample entries are found since the lastSeenTimeStamp, NVML_ERROR_NOT_FOUND + * is returned. + * + * To read utilization values, first determine the size of buffer required to hold the samples by invoking the function with + * \a utilization set to NULL. The caller should allocate a buffer of size + * processSamplesCount * sizeof(nvmlProcessUtilizationSample_t). Invoke the function again with the allocated buffer passed + * in \a utilization, and \a processSamplesCount set to the number of entries the buffer is sized for. + * + * On successful return, the function updates \a processSamplesCount with the number of process utilization sample + * structures that were actually written. This may differ from a previously read value as instances are created or + * destroyed. + * + * lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 + * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set lastSeenTimeStamp + * to a timeStamp retrieved from a previous query to read utilization since the previous query. + * + * @note On MIG-enabled GPUs, querying process utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Pointer to caller-supplied buffer in which guest process utilization samples are returned + * @param processSamplesCount Pointer to caller-supplied array size, and returns number of processes running + * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. + + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NOT_FOUND if sample entries are not found + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetProcessUtilization(nvmlDevice_t device, nvmlProcessUtilizationSample_t *utilization, + unsigned int *processSamplesCount, unsigned long long lastSeenTimeStamp); + +/** + * Retrieves the recent utilization and process ID for all running processes + * + * For Maxwell &tm; or newer fully supported devices. + * + * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, and video decoder, jpeg decoder, OFA (Optical Flow Accelerator) + * for all running processes. Utilization values are returned as an array of utilization sample structures in the caller-supplied buffer pointed at + * by \a procesesUtilInfo->procUtilArray. One utilization sample structure is returned per process running, that had some non-zero utilization + * during the last sample period. It includes the CPU timestamp at which the samples were recorded. Individual utilization values + * are returned as "unsigned int" values. + * + * The caller should allocate a buffer of size processSamplesCount * sizeof(nvmlProcessUtilizationInfo_t). If the buffer is too small, the API will + * return \a NVML_ERROR_INSUFFICIENT_SIZE, with the recommended minimal buffer size at \a procesesUtilInfo->processSamplesCount. The caller should + * invoke the function again with the allocated buffer passed in \a procesesUtilInfo->procUtilArray, and \a procesesUtilInfo->processSamplesCount + * set to the number no less than the recommended value by the previous API return. + * + * On successful return, the function updates \a procesesUtilInfo->processSamplesCount with the number of process utilization info structures + * that were actually written. This may differ from a previously read value as instances are created or destroyed. + * + * \a procesesUtilInfo->lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 + * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set \a procesesUtilInfo->lastSeenTimeStamp + * to a timeStamp retrieved from a previous query to read utilization since the previous query. + * + * \a procesesUtilInfo->version is the version number of the structure nvmlProcessesUtilizationInfo_t, the caller should set the correct version + * number to retrieve the specific version of processes utilization information. + * + * @note On MIG-enabled GPUs, querying process utilization is not currently supported. + * + * @param device The identifier of the target device + * @param procesesUtilInfo Pointer to the caller-provided structure of nvmlProcessesUtilizationInfo_t. + + * @return + * - \ref NVML_SUCCESS If \a procesesUtilInfo->procUtilArray has been populated + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid, or \a procesesUtilInfo is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If the device does not support this feature + * - \ref NVML_ERROR_NOT_FOUND If sample entries are not found + * - \ref NVML_ERROR_GPU_IS_LOST If the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a procesesUtilInfo is invalid + * - \ref NVML_ERROR_INSUFFICIENT_SIZE If \a procesesUtilInfo->procUtilArray is NULL, or the buffer size of procesesUtilInfo->procUtilArray is too small. + * The caller should check the minimul array size from the returned procesesUtilInfo->processSamplesCount, and call + * the function again with a buffer no smaller than procesesUtilInfo->processSamplesCount * sizeof(nvmlProcessUtilizationInfo_t) + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetProcessesUtilizationInfo(nvmlDevice_t device, nvmlProcessesUtilizationInfo_t *procesesUtilInfo); + +/** + * Get platform information of this device. + * + * %BLACKWELL_OR_NEWER% + * + * See \ref nvmlPlatformInfo_v2_t for more information on the struct. + * + * @param device The identifier of the target device + * @param platformInfo Pointer to the caller-provided structure of nvmlPlatformInfo_t. + * + * @return + * - \ref NVML_SUCCESS If \a platformInfo has been retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid or \a platformInfo is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If the device does not support this feature + * - \ref NVML_ERROR_MEMORY if system memory is insufficient + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a nvmlPlatformInfo_t is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPlatformInfo(nvmlDevice_t device, nvmlPlatformInfo_t *platformInfo); + +/** + * Retrieves the Per Device Identifier (PDI) associated with this device. + * + * For Pascal &tm; or newer fully supported devices. + * + * See \ref nvmlPdi_v1_t for more information on the struct. + * + * @param[in] device The identifier of the target device + * @param[out] pdi Reference to the caller-provided structure to return the GPU PDI + * + * @return + * - \ref NVML_SUCCESS if \a pdi has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a pdi is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the version is invalid/unsupported + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPdi(nvmlDevice_t device, nvmlPdi_t *pdi); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlUnitCommands Unit Commands + * This chapter describes NVML operations that change the state of the unit. For S-class products. + * Each of these requires root/admin access. Non-admin users will see an NVML_ERROR_NO_PERMISSION + * error code when invoking any of these methods. + * @{ + */ +/***************************************************************************************************/ + +/** + * Set the LED state for the unit. The LED can be either green (0) or amber (1). + * + * For S-class products. + * Requires root/admin permissions. + * + * This operation takes effect immediately. + * + * + * Current S-Class products don't provide unique LEDs for each unit. As such, both front + * and back LEDs will be toggled in unison regardless of which unit is specified with this command. + * + * See \ref nvmlLedColor_t for available colors. + * + * @param unit The identifier of the target unit + * @param color The target LED color + * + * @return + * - \ref NVML_SUCCESS if the LED color has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit or \a color is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlUnitGetLedState() + */ +nvmlReturn_t DECLDIR nvmlUnitSetLedState(nvmlUnit_t unit, nvmlLedColor_t color); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlDeviceCommands Device Commands + * This chapter describes NVML operations that change the state of the device. + * Each of these requires root/admin access. Non-admin users will see an NVML_ERROR_NO_PERMISSION + * error code when invoking any of these methods. + * @{ + */ +/***************************************************************************************************/ + +/** + * Set the persistence mode for the device. + * + * For all products. + * For Linux only. + * Requires root/admin permissions. + * + * The persistence mode determines whether the GPU driver software is torn down after the last client + * exits. + * + * This operation takes effect immediately. It is not persistent across reboots. After each reboot the + * persistence mode is reset to "Disabled". + * + * See \ref nvmlEnableState_t for available modes. + * + * After calling this API with mode set to NVML_FEATURE_DISABLED on a device that has its own NUMA + * memory, the given device handle will no longer be valid, and to continue to interact with this + * device, a new handle should be obtained from one of the nvmlDeviceGetHandleBy*() APIs. This + * limitation is currently only applicable to devices that have a coherent NVLink connection to + * system memory. + * + * @param device The identifier of the target device + * @param mode The target persistence mode + * + * @return + * - \ref NVML_SUCCESS if the persistence mode was set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetPersistenceMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetPersistenceMode(nvmlDevice_t device, nvmlEnableState_t mode); + +/** + * Set the compute mode for the device. + * + * For all products. + * Requires root/admin permissions. + * + * The compute mode determines whether a GPU can be used for compute operations and whether it can + * be shared across contexts. + * + * This operation takes effect immediately. Under Linux it is not persistent across reboots and + * always resets to "Default". Under windows it is persistent. + * + * Under windows compute mode may only be set to DEFAULT when running in WDDM + * + * @note On MIG-enabled GPUs, compute mode would be set to DEFAULT and changing it is not supported. + * + * See \ref nvmlComputeMode_t for details on available compute modes. + * + * @param device The identifier of the target device + * @param mode The target compute mode + * + * @return + * - \ref NVML_SUCCESS if the compute mode was set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetComputeMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetComputeMode(nvmlDevice_t device, nvmlComputeMode_t mode); + +/** + * Set the ECC mode for the device. + * + * For Kepler &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * Requires root/admin permissions. + * + * The ECC mode determines whether the GPU enables its ECC support. + * + * This operation takes effect after the next reboot. + * + * See \ref nvmlEnableState_t for details on available modes. + * + * @param device The identifier of the target device + * @param ecc The target ECC mode + * + * @return + * - \ref NVML_SUCCESS if the ECC mode was set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a ecc is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetEccMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetEccMode(nvmlDevice_t device, nvmlEnableState_t ecc); + +/** + * Clear the ECC error and other memory error counts for the device. + * + * For Kepler &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 2.0 or higher to clear aggregate location-based ECC counts. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher to clear all other ECC counts. + * Requires root/admin permissions. + * Requires ECC Mode to be enabled. + * + * Sets all of the specified ECC counters to 0, including both detailed and total counts. + * + * This operation takes effect immediately. + * + * See \ref nvmlMemoryErrorType_t for details on available counter types. + * + * @param device The identifier of the target device + * @param counterType Flag that indicates which type of errors should be cleared. + * + * @return + * - \ref NVML_SUCCESS if the error counts were cleared + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a counterType is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see + * - nvmlDeviceGetDetailedEccErrors() + * - nvmlDeviceGetTotalEccErrors() + */ +nvmlReturn_t DECLDIR nvmlDeviceClearEccErrorCounts(nvmlDevice_t device, nvmlEccCounterType_t counterType); + +/** + * Set the driver model for the device. + * + * For Fermi &tm; or newer fully supported devices. + * For windows only. + * Requires root/admin permissions. + * + * On Windows platforms the device driver can run in either WDDM or WDM (TCC) mode. If a display is attached + * to the device it must run in WDDM mode. + * + * It is possible to force the change to WDM (TCC) while the display is still attached with a force flag (nvmlFlagForce). + * This should only be done if the host is subsequently powered down and the display is detached from the device + * before the next reboot. + * + * This operation takes effect after the next reboot. + * + * Windows driver model may only be set to WDDM when running in DEFAULT compute mode. + * + * Change driver model to WDDM is not supported when GPU doesn't support graphics acceleration or + * will not support it after reboot. See \ref nvmlDeviceSetGpuOperationMode. + * + * See \ref nvmlDriverModel_t for details on available driver models. + * See \ref nvmlFlagDefault and \ref nvmlFlagForce + * + * @param device The identifier of the target device + * @param driverModel The target driver model + * @param flags Flags that change the default behavior + * + * @return + * - \ref NVML_SUCCESS if the driver model has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a driverModel is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the platform is not windows or the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetDriverModel() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetDriverModel(nvmlDevice_t device, nvmlDriverModel_t driverModel, unsigned int flags); + +typedef enum nvmlClockLimitId_enum { + NVML_CLOCK_LIMIT_ID_RANGE_START = 0xffffff00, + NVML_CLOCK_LIMIT_ID_TDP, + NVML_CLOCK_LIMIT_ID_UNLIMITED +} nvmlClockLimitId_t; + +/** + * Set clocks that device will lock to. + * + * Sets the clocks that the device will be running at to the value in the range of minGpuClockMHz to maxGpuClockMHz. + * + * Can be used as a setting to request constant performance. + * + * This can be called with a pair of integer clock frequencies in MHz, or a pair of /ref nvmlClockLimitId_t values. + * See the table below for valid combinations of these values. + * + * minGpuClock | maxGpuClock | Effect + * ------------+-------------+-------------------------------------------------- + * tdp | tdp | Lock clock to TDP + * unlimited | tdp | Upper bound is TDP but clock may drift below this + * tdp | unlimited | Lower bound is TDP but clock may boost above this + * unlimited | unlimited | Unlocked (== nvmlDeviceResetGpuLockedClocks) + * + * If one arg takes one of these values, the other must be one of these values as + * well. Mixed numeric and symbolic calls return NVML_ERROR_INVALID_ARGUMENT. + * + * Requires root/admin permissions. + * + * After system reboot or driver reload GPU clocks go back to their default value. + * See \ref nvmlDeviceResetGpuLockedClocks. + * + * For Volta &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param minGpuClockMHz Requested minimum gpu clock in MHz + * @param maxGpuClockMHz Requested maximum gpu clock in MHz + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minGpuClockMHz and \a maxGpuClockMHz + * is not a valid clock combination + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetGpuLockedClocks(nvmlDevice_t device, unsigned int minGpuClockMHz, unsigned int maxGpuClockMHz); + +/** + * Resets the gpu clock to the default value + * + * This is the gpu clock that will be used after system reboot or driver reload. + * Default values are idle clocks. + * + * @see nvmlDeviceSetGpuLockedClocks + * + * For Volta &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceResetGpuLockedClocks(nvmlDevice_t device); + +/** + * Set memory clocks that device will lock to. + * + * Sets the device's memory clocks to the value in the range of minMemClockMHz to maxMemClockMHz. + * + * Can be used as a setting to request constant performance. + * + * Requires root/admin permissions. + * + * After system reboot or driver reload memory clocks go back to their default value. + * See \ref nvmlDeviceResetMemoryLockedClocks. + * + * For Ampere &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param minMemClockMHz Requested minimum memory clock in MHz + * @param maxMemClockMHz Requested maximum memory clock in MHz + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minGpuClockMHz and \a maxGpuClockMHz + * is not a valid clock combination + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetMemoryLockedClocks(nvmlDevice_t device, unsigned int minMemClockMHz, unsigned int maxMemClockMHz); + +/** + * Resets the memory clock to the default value + * + * This is the memory clock that will be used after system reboot or driver reload. + * Default values are idle clocks. + * + * @see nvmlDeviceSetMemoryLockedClocks + * + * For Ampere &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceResetMemoryLockedClocks(nvmlDevice_t device); + +/** + * @deprecated Applications clocks are deprecated and will be removed in CUDA 14.0. + * + * Please use \ref nvmlDeviceSetMemoryLockedClocks for Memory Clocks and + * \ref nvmlDeviceSetGpuLockedClocks for Graphics Clocks. + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceSetApplicationsClocks(nvmlDevice_t device, unsigned int memClockMHz, unsigned int graphicsClockMHz); + +/** + * @deprecated Applications clocks are deprecated and will be removed in CUDA 14.0. + * + * Please use \ref nvmlDeviceResetMemoryLockedClocks for Memory Clocks and + * \ref nvmlDeviceResetGpuLockedClocks for Graphics Clocks. + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceResetApplicationsClocks(nvmlDevice_t device); + +/** + * Try to set the current state of Auto Boosted clocks on a device. + * + * For Kepler &tm; or newer fully supported devices. + * + * Auto Boosted clocks are enabled by default on some hardware, allowing the GPU to run at higher clock rates + * to maximize performance as thermal limits allow. Auto Boosted clocks should be disabled if fixed clock + * rates are desired. + * + * Non-root users may use this API by default but can be restricted by root from using this API by calling + * \ref nvmlDeviceSetAPIRestriction with apiType=NVML_RESTRICTED_API_SET_AUTO_BOOSTED_CLOCKS. + * Note: Persistence Mode is required to modify current Auto Boost settings, therefore, it must be enabled. + * + * On Pascal and newer hardware, Auto Boosted clocks are controlled through application clocks. + * Use \ref nvmlDeviceSetApplicationsClocks and \ref nvmlDeviceResetApplicationsClocks to control Auto Boost + * behavior. + * + * @param device The identifier of the target device + * @param enabled What state to try to set Auto Boosted clocks of the target device to + * + * @return + * - \ref NVML_SUCCESS If the Auto Boosted clocks were successfully set to the state specified by \a enabled + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support Auto Boosted clocks + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceSetAutoBoostedClocksEnabled(nvmlDevice_t device, nvmlEnableState_t enabled); + +/** + * Try to set the default state of Auto Boosted clocks on a device. This is the default state that Auto Boosted clocks will + * return to when no compute running processes (e.g. CUDA application which have an active context) are running + * + * For Kepler &tm; or newer non-GeForce fully supported devices and Maxwell or newer GeForce devices. + * Requires root/admin permissions. + * + * Auto Boosted clocks are enabled by default on some hardware, allowing the GPU to run at higher clock rates + * to maximize performance as thermal limits allow. Auto Boosted clocks should be disabled if fixed clock + * rates are desired. + * + * On Pascal and newer hardware, Auto Boosted clocks are controlled through application clocks. + * Use \ref nvmlDeviceSetApplicationsClocks and \ref nvmlDeviceResetApplicationsClocks to control Auto Boost + * behavior. + * + * @param device The identifier of the target device + * @param enabled What state to try to set default Auto Boosted clocks of the target device to + * @param flags Flags that change the default behavior. Currently Unused. + * + * @return + * - \ref NVML_SUCCESS If the Auto Boosted clock's default state was successfully set to the state specified by \a enabled + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NO_PERMISSION If the calling user does not have permission to change Auto Boosted clock's default state. + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support Auto Boosted clocks + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceSetDefaultAutoBoostedClocksEnabled(nvmlDevice_t device, nvmlEnableState_t enabled, unsigned int flags); + +/** + * Sets the speed of the fan control policy to default. + * + * For all cuda-capable discrete products with fans + * + * @param device The identifier of the target device + * @param fan The index of the fan, starting at zero + * + * return + * NVML_SUCCESS if speed has been adjusted + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if device is invalid + * NVML_ERROR_NOT_SUPPORTED if the device does not support this + * (doesn't have fans) + * NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetDefaultFanSpeed_v2(nvmlDevice_t device, unsigned int fan); + +/** + * Sets current fan control policy. + * + * For Maxwell &tm; or newer fully supported devices. + * + * Requires privileged user. + * + * For all cuda-capable discrete products with fans + * + * device The identifier of the target \a device + * policy The fan control \a policy to set + * + * return + * NVML_SUCCESS if \a policy has been set + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a policy is null or the \a fan given doesn't reference + * a fan that exists. + * NVML_ERROR_NOT_SUPPORTED if the \a device is older than Maxwell + * NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetFanControlPolicy(nvmlDevice_t device, unsigned int fan, + nvmlFanControlPolicy_t policy); + +/** + * Sets the temperature threshold for the GPU with the specified threshold type in degrees C. + * + * For Maxwell &tm; or newer fully supported devices. + * + * See \ref nvmlTemperatureThresholds_t for details on available temperature thresholds. + * + * @param device The identifier of the target device + * @param thresholdType The type of threshold value to be set + * @param temp Reference which hold the value to be set + * @return + * - \ref NVML_SUCCESS if \a temp has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a thresholdType is invalid or \a temp is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a temperature sensor or is unsupported + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetTemperatureThreshold(nvmlDevice_t device, nvmlTemperatureThresholds_t thresholdType, int *temp); + +/** + * Set new power limit of this device. + * + * For Kepler &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * See \ref nvmlDeviceGetPowerManagementLimitConstraints to check the allowed ranges of values. + * + * \note Limit is not persistent across reboots or driver unloads. + * Enable persistent mode to prevent driver from unloading when no application is using the device. + * + * @param device The identifier of the target device + * @param limit Power management limit in milliwatts to set + * + * @return + * - \ref NVML_SUCCESS if \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a defaultLimit is out of range + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetPowerManagementLimitConstraints + * @see nvmlDeviceGetPowerManagementDefaultLimit + */ +nvmlReturn_t DECLDIR nvmlDeviceSetPowerManagementLimit(nvmlDevice_t device, unsigned int limit); + +/** + * Sets new GOM. See \a nvmlGpuOperationMode_t for details. + * + * For GK110 M-class and X-class Tesla &tm; products from the Kepler family. + * Modes \ref NVML_GOM_LOW_DP and \ref NVML_GOM_ALL_ON are supported on fully supported GeForce products. + * Not supported on Quadro ® and Tesla &tm; C-class products. + * Requires root/admin permissions. + * + * Changing GOMs requires a reboot. + * The reboot requirement might be removed in the future. + * + * Compute only GOMs don't support graphics acceleration. Under windows switching to these GOMs when + * pending driver model is WDDM is not supported. See \ref nvmlDeviceSetDriverModel. + * + * @param device The identifier of the target device + * @param mode Target GOM + * + * @return + * - \ref NVML_SUCCESS if \a mode has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode incorrect + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support GOM or specific mode + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlGpuOperationMode_t + * @see nvmlDeviceGetGpuOperationMode + */ +nvmlReturn_t DECLDIR nvmlDeviceSetGpuOperationMode(nvmlDevice_t device, nvmlGpuOperationMode_t mode); + +/** + * Changes the root/admin restructions on certain APIs. See \a nvmlRestrictedAPI_t for the list of supported APIs. + * This method can be used by a root/admin user to give non-root/admin access to certain otherwise-restricted APIs. + * The new setting lasts for the lifetime of the NVIDIA driver; it is not persistent. See \a nvmlDeviceGetAPIRestriction + * to query the current restriction settings. + * + * For Kepler &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * @param device The identifier of the target device + * @param apiType Target API type for this operation + * @param isRestricted The target restriction + * + * @return + * - \ref NVML_SUCCESS if \a isRestricted has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a apiType incorrect + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support changing API restrictions or the device does not support + * the feature that api restrictions are being set for (E.G. Enabling/disabling auto + * boosted clocks is not supported by the device) + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlRestrictedAPI_t + */ +nvmlReturn_t DECLDIR nvmlDeviceSetAPIRestriction(nvmlDevice_t device, nvmlRestrictedAPI_t apiType, nvmlEnableState_t isRestricted); + +/** + * Sets the speed of a specified fan. + * + * WARNING: This function changes the fan control policy to manual. It means that YOU have to monitor + * the temperature and adjust the fan speed accordingly. + * If you set the fan speed too low you can burn your GPU! + * Use nvmlDeviceSetDefaultFanSpeed_v2 to restore default control policy. + * + * For all cuda-capable discrete products with fans that are Maxwell or Newer. + * + * device The identifier of the target device + * fan The index of the fan, starting at zero + * speed The target speed of the fan [0-100] in % of max speed + * + * return + * NVML_SUCCESS if the fan speed has been set + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if the device is not valid, or the speed is outside acceptable ranges, + * or if the fan index doesn't reference an actual fan. + * NVML_ERROR_NOT_SUPPORTED if the device is older than Maxwell. + * NVML_ERROR_UNKNOWN if there was an unexpected error. + */ +nvmlReturn_t DECLDIR nvmlDeviceSetFanSpeed_v2(nvmlDevice_t device, unsigned int fan, unsigned int speed); + +/** + * @deprecated Will be deprecated in a future release. Use \ref nvmlDeviceSetClockOffsets instead. It works + * on Maxwell onwards GPU architectures. + * + * Set the GPCCLK VF offset value + * @param[in] device The identifier of the target device + * @param[in] offset The GPCCLK VF offset value to set + * + * @return + * - \ref NVML_SUCCESS if \a offset has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceSetGpcClkVfOffset(nvmlDevice_t device, int offset); + +/** + * @deprecated Will be deprecated in a future release. Use \ref nvmlDeviceSetClockOffsets instead. It works + * on Maxwell onwards GPU architectures. + * + * Set the MemClk (Memory Clock) VF offset value. It requires elevated privileges. + * @param[in] device The identifier of the target device + * @param[in] offset The MemClk VF offset value to set + * + * @return + * - \ref NVML_SUCCESS if \a offset has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceSetMemClkVfOffset(nvmlDevice_t device, int offset); + +/** + * @} + */ + +/** @addtogroup nvmlAccountingStats + * @{ + */ + +/** + * Enables or disables per process accounting. + * + * For Kepler &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * @note This setting is not persistent and will default to disabled after driver unloads. + * Enable persistence mode to be sure the setting doesn't switch off to disabled. + * + * @note Enabling accounting mode has no negative impact on the GPU performance. + * + * @note Disabling accounting clears all accounting pids information. + * + * @note On MIG-enabled GPUs, accounting mode would be set to DISABLED and changing it is not supported. + * + * See \ref nvmlDeviceGetAccountingMode + * See \ref nvmlDeviceGetAccountingStats + * See \ref nvmlDeviceClearAccountingPids + * + * @param device The identifier of the target device + * @param mode The target accounting mode + * + * @return + * - \ref NVML_SUCCESS if the new mode has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a mode are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetAccountingMode(nvmlDevice_t device, nvmlEnableState_t mode); + +/** + * Clears accounting information about all processes that have already terminated. + * + * For Kepler &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * See \ref nvmlDeviceGetAccountingMode + * See \ref nvmlDeviceGetAccountingStats + * See \ref nvmlDeviceSetAccountingMode + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if accounting information has been cleared + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceClearAccountingPids(nvmlDevice_t device); + +/** @} */ // @addtogroup nvmlAccountingStats + +/***************************************************************************************************/ +/** @defgroup NvLink NvLink Methods + * This chapter describes methods that NVML can perform on NVLINK enabled devices. + * @{ + */ +/***************************************************************************************************/ + +#define NVML_NVLINK_BER_MANTISSA_SHIFT 8 +#define NVML_NVLINK_BER_MANTISSA_WIDTH 0xf + +#define NVML_NVLINK_BER_EXP_SHIFT 0 +#define NVML_NVLINK_BER_EXP_WIDTH 0xff + +/** + * Nvlink Error counter BER can be obtained using the below macros + * Ex - NVML_NVLINK_ERROR_COUNTER_BER_GET(var, BER_MANTISSA) + */ +#define NVML_NVLINK_ERROR_COUNTER_BER_GET(var, type) \ + (((var) >> NVML_NVLINK_##type##_SHIFT) & \ + (NVML_NVLINK_##type##_WIDTH)) \ + +/* + * NVML_FI_DEV_NVLINK_GET_STATE state enums + */ +#define NVML_NVLINK_STATE_INACTIVE 0x0 +#define NVML_NVLINK_STATE_ACTIVE 0x1 +#define NVML_NVLINK_STATE_SLEEP 0x2 + +#define NVML_NVLINK_TOTAL_SUPPORTED_BW_MODES 23 + +typedef struct +{ + unsigned int version; + unsigned char bwModes[NVML_NVLINK_TOTAL_SUPPORTED_BW_MODES]; + unsigned char totalBwModes; +} nvmlNvlinkSupportedBwModes_v1_t; +typedef nvmlNvlinkSupportedBwModes_v1_t nvmlNvlinkSupportedBwModes_t; +#define nvmlNvlinkSupportedBwModes_v1 NVML_STRUCT_VERSION(NvlinkSupportedBwModes, 1) + +typedef struct +{ + unsigned int version; + unsigned int bIsBest; + unsigned char bwMode; +} nvmlNvlinkGetBwMode_v1_t; +typedef nvmlNvlinkGetBwMode_v1_t nvmlNvlinkGetBwMode_t; +#define nvmlNvlinkGetBwMode_v1 NVML_STRUCT_VERSION(NvlinkGetBwMode, 1) + +typedef struct +{ + unsigned int version; + unsigned int bSetBest; + unsigned char bwMode; +} nvmlNvlinkSetBwMode_v1_t; +typedef nvmlNvlinkSetBwMode_v1_t nvmlNvlinkSetBwMode_t; +#define nvmlNvlinkSetBwMode_v1 NVML_STRUCT_VERSION(NvlinkSetBwMode, 1) + +/** + * Struct to represent per device NVLINK information v1 + */ +typedef struct +{ + unsigned int version; //!< IN - the API version number + unsigned int isNvleEnabled; //!< OUT - NVLINK encryption enablement +} nvmlNvLinkInfo_v1_t; +#define nvmlNvLinkInfo_v1 NVML_STRUCT_VERSION(NvLinkInfo, 1) + +#define NVML_NVLINK_FIRMWARE_UCODE_TYPE_MSE 0x1 +#define NVML_NVLINK_FIRMWARE_UCODE_TYPE_NETIR 0x2 +#define NVML_NVLINK_FIRMWARE_UCODE_TYPE_NETIR_UPHY 0x3 +#define NVML_NVLINK_FIRMWARE_UCODE_TYPE_NETIR_CLN 0x4 +#define NVML_NVLINK_FIRMWARE_UCODE_TYPE_NETIR_DLN 0x5 +#define NVML_NVLINK_FIRMWARE_VERSION_LENGTH 100 + +/** + * Struct to represent NVLINK firmware Semantic versioning and ucode type + */ +typedef struct +{ + unsigned char ucodeType; + unsigned int major; + unsigned int minor; + unsigned int subMinor; +} nvmlNvlinkFirmwareVersion_t; + +/** + * Struct to represent NVLINK firmware information + */ +typedef struct +{ + nvmlNvlinkFirmwareVersion_t firmwareVersion[NVML_NVLINK_FIRMWARE_VERSION_LENGTH]; //!< OUT - NVLINK firmware version + unsigned int numValidEntries; //!< OUT - Number of valid firmware entries +} nvmlNvlinkFirmwareInfo_t; + +/** + * Struct to represent per device NVLINK information v2 + */ +typedef struct +{ + unsigned int version; //!< IN - the API version number + unsigned int isNvleEnabled; //!< OUT - NVLINK encryption enablement + nvmlNvlinkFirmwareInfo_t firmwareInfo; //!< OUT - NVLINK Firmware info +} nvmlNvLinkInfo_v2_t; +typedef nvmlNvLinkInfo_v2_t nvmlNvLinkInfo_t; +#define nvmlNvLinkInfo_v2 NVML_STRUCT_VERSION(NvLinkInfo, 2) + +/** + * Retrieves the state of the device's NvLink for the link specified + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * @param isActive \a nvmlEnableState_t where NVML_FEATURE_ENABLED indicates that + * the link is active and NVML_FEATURE_DISABLED indicates it + * is inactive + * + * @return + * - \ref NVML_SUCCESS if \a isActive has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid or \a isActive is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkState(nvmlDevice_t device, unsigned int link, nvmlEnableState_t *isActive); + +/** + * Retrieves the version of the device's NvLink for the link specified + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * @param version Requested NvLink version from nvmlNvlinkVersion_t + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid or \a version is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkVersion(nvmlDevice_t device, unsigned int link, unsigned int *version); + +/** + * Retrieves the requested capability from the device's NvLink for the link specified + * Please refer to the \a nvmlNvLinkCapability_t structure for the specific caps that can be queried + * The return value should be treated as a boolean. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * @param capability Specifies the \a nvmlNvLinkCapability_t to be queried + * @param capResult A boolean for the queried capability indicating that feature is available + * + * @return + * - \ref NVML_SUCCESS if \a capResult has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, or \a capability is invalid or \a capResult is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkCapability(nvmlDevice_t device, unsigned int link, + nvmlNvLinkCapability_t capability, unsigned int *capResult); + +/** + * Retrieves the PCI information for the remote node on a NvLink link + * Note: pciSubSystemId is not filled in this function and is indeterminate + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * @param pci \a nvmlPciInfo_t of the remote node for the specified link + * + * @return + * - \ref NVML_SUCCESS if \a pci has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid or \a pci is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemotePciInfo_v2(nvmlDevice_t device, unsigned int link, nvmlPciInfo_t *pci); + +/** + * Retrieves the specified error counter value + * Please refer to \a nvmlNvLinkErrorCounter_t for error counters that are available + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * @param counter Specifies the NvLink counter to be queried + * @param counterValue Returned counter value + * + * @return + * - \ref NVML_SUCCESS if \a counter has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, or \a counter is invalid or \a counterValue is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkErrorCounter(nvmlDevice_t device, unsigned int link, + nvmlNvLinkErrorCounter_t counter, unsigned long long *counterValue); + +/** + * Resets all error counters to zero + * Please refer to \a nvmlNvLinkErrorCounter_t for the list of error counters that are reset + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * + * @return + * - \ref NVML_SUCCESS if the reset is successful + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceResetNvLinkErrorCounters(nvmlDevice_t device, unsigned int link); + +/** + * @deprecated Setting utilization counter control is no longer supported. + * + * Set the NVLINK utilization counter control information for the specified counter, 0 or 1. + * Please refer to \a nvmlNvLinkUtilizationControl_t for the structure definition. Performs a reset + * of the counters if the reset parameter is non-zero. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param counter Specifies the counter that should be set (0 or 1). + * @param link Specifies the NvLink link to be queried + * @param control A reference to the \a nvmlNvLinkUtilizationControl_t to set + * @param reset Resets the counters on set if non-zero + * + * @return + * - \ref NVML_SUCCESS if the control has been set successfully + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, \a link, or \a control is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceSetNvLinkUtilizationControl(nvmlDevice_t device, unsigned int link, unsigned int counter, + nvmlNvLinkUtilizationControl_t *control, unsigned int reset); + +/** + * @deprecated Getting utilization counter control is no longer supported. + * + * Get the NVLINK utilization counter control information for the specified counter, 0 or 1. + * Please refer to \a nvmlNvLinkUtilizationControl_t for the structure definition + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param counter Specifies the counter that should be set (0 or 1). + * @param link Specifies the NvLink link to be queried + * @param control A reference to the \a nvmlNvLinkUtilizationControl_t to place information + * + * @return + * - \ref NVML_SUCCESS if the control has been set successfully + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, \a link, or \a control is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkUtilizationControl(nvmlDevice_t device, unsigned int link, unsigned int counter, + nvmlNvLinkUtilizationControl_t *control); + + +/** + * @deprecated Use \ref nvmlDeviceGetFieldValues with NVML_FI_DEV_NVLINK_THROUGHPUT_* as field values instead. + * + * Retrieve the NVLINK utilization counter based on the current control for a specified counter. + * In general it is good practice to use \a nvmlDeviceSetNvLinkUtilizationControl + * before reading the utilization counters as they have no default state + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * @param counter Specifies the counter that should be read (0 or 1). + * @param rxcounter Receive counter return value + * @param txcounter Transmit counter return value + * + * @return + * - \ref NVML_SUCCESS if \a rxcounter and \a txcounter have been successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, or \a link is invalid or \a rxcounter or \a txcounter are NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkUtilizationCounter(nvmlDevice_t device, unsigned int link, unsigned int counter, + unsigned long long *rxcounter, unsigned long long *txcounter); + +/** + * @deprecated Freezing NVLINK utilization counters is no longer supported. + * + * Freeze the NVLINK utilization counters + * Both the receive and transmit counters are operated on by this function + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be queried + * @param counter Specifies the counter that should be frozen (0 or 1). + * @param freeze NVML_FEATURE_ENABLED = freeze the receive and transmit counters + * NVML_FEATURE_DISABLED = unfreeze the receive and transmit counters + * + * @return + * - \ref NVML_SUCCESS if counters were successfully frozen or unfrozen + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, \a counter, or \a freeze is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceFreezeNvLinkUtilizationCounter (nvmlDevice_t device, unsigned int link, + unsigned int counter, nvmlEnableState_t freeze); + +/** + * @deprecated Resetting NVLINK utilization counters is no longer supported. + * + * Reset the NVLINK utilization counters + * Both the receive and transmit counters are operated on by this function + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param link Specifies the NvLink link to be reset + * @param counter Specifies the counter that should be reset (0 or 1) + * + * @return + * - \ref NVML_SUCCESS if counters were successfully reset + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, or \a counter is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlDeviceResetNvLinkUtilizationCounter (nvmlDevice_t device, unsigned int link, unsigned int counter); + +/** +* Get the NVLink device type of the remote device connected over the given link. +* +* @param device The device handle of the target GPU +* @param link The NVLink link index on the target GPU +* @param pNvLinkDeviceType Pointer in which the output remote device type is returned +* +* @return +* - \ref NVML_SUCCESS if \a pNvLinkDeviceType has been set +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_NOT_SUPPORTED if NVLink is not supported +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid, or +* \a pNvLinkDeviceType is NULL +* - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is +* otherwise inaccessible +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemoteDeviceType(nvmlDevice_t device, unsigned int link, nvmlIntNvLinkDeviceType_t *pNvLinkDeviceType); + +/** + * Set NvLink Low Power Threshold for device. + * + * For Hopper &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param info Reference to \a nvmlNvLinkPowerThres_t struct + * input parameters + * + * @return + * - \ref NVML_SUCCESS if the \a Threshold is successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a Threshold is not within range + * - \ref NVML_ERROR_NOT_READY if an internal driver setting prevents the threshold from being used + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * + **/ +nvmlReturn_t DECLDIR nvmlDeviceSetNvLinkDeviceLowPowerThreshold(nvmlDevice_t device, nvmlNvLinkPowerThres_t *info); + +/** + * Set the global nvlink bandwith mode + * + * @param nvlinkBwMode nvlink bandwidth mode + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid argument is provided + * - \ref NVML_ERROR_IN_USE if P2P object exists + * - \ref NVML_ERROR_NOT_SUPPORTED if GPU is not Hopper or newer architecture. + * - \ref NVML_ERROR_NO_PERMISSION if not root user + */ +nvmlReturn_t DECLDIR nvmlSystemSetNvlinkBwMode(unsigned int nvlinkBwMode); + +/** + * Get the global nvlink bandwith mode + * + * @param nvlinkBwMode reference of nvlink bandwidth mode + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid pointer is provided + * - \ref NVML_ERROR_NOT_SUPPORTED if GPU is not Hopper or newer architecture. + * - \ref NVML_ERROR_NO_PERMISSION if not root user + */ +nvmlReturn_t DECLDIR nvmlSystemGetNvlinkBwMode(unsigned int *nvlinkBwMode); + +/** + * Get the supported NvLink Reduced Bandwidth Modes of the device + * + * %BLACKWELL_OR_NEWER% + * + * @param device The identifier of the target device + * @param supportedBwMode Reference to \a nvmlNvlinkSupportedBwModes_t + * + * @return + * - \ref NVML_SUCCESS if the query was successful + * - \ref NVML_ERROR_INVALID_ARGUMENT if device is invalid or supportedBwMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this feature is not supported by the device + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the version specified is not supported + **/ +nvmlReturn_t DECLDIR nvmlDeviceGetNvlinkSupportedBwModes(nvmlDevice_t device, + nvmlNvlinkSupportedBwModes_t *supportedBwMode); + +/** + * Get the NvLink Reduced Bandwidth Mode for the device + * + * %BLACKWELL_OR_NEWER% + * + * @param device The identifier of the target device + * @param getBwMode Reference to \a nvmlNvlinkGetBwMode_t + * + * @return + * - \ref NVML_SUCCESS if the query was successful + * - \ref NVML_ERROR_INVALID_ARGUMENT if device is invalid or getBwMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this feature is not supported by the device + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the version specified is not supported + **/ +nvmlReturn_t DECLDIR nvmlDeviceGetNvlinkBwMode(nvmlDevice_t device, + nvmlNvlinkGetBwMode_t *getBwMode); + +/** + * Set the NvLink Reduced Bandwidth Mode for the device + * + * %BLACKWELL_OR_NEWER% + * + * @param device The identifier of the target device + * @param setBwMode Reference to \a nvmlNvlinkSetBwMode_t + * + * @return + * - \ref NVML_SUCCESS if the Bandwidth mode was successfully set + * - \ref NVML_ERROR_INVALID_ARGUMENT if device is invalid or setBwMode is NULL + * - \ref NVML_ERROR_NO_PERMISSION if user does not have permission to change Bandwidth mode + * - \ref NVML_ERROR_NOT_SUPPORTED if this feature is not supported by the device + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the version specified is not supported + **/ +nvmlReturn_t DECLDIR nvmlDeviceSetNvlinkBwMode(nvmlDevice_t device, + nvmlNvlinkSetBwMode_t *setBwMode); + +/** + * Query NVLINK information associated with this device. + * + * @param[in] device The identifier of the target device + * @param[out] info Reference to \a nvmlNvLinkInfo_t + * + * @return + * - \ref NVML_SUCCESS if query is success + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a info is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the version is invalid/unsupported + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkInfo(nvmlDevice_t device, nvmlNvLinkInfo_t *info); + +/** @} */ // @defgroup NvLink NvLink Methods + +/***************************************************************************************************/ +/** @defgroup nvmlEvents Event Handling Methods + * This chapter describes methods that NVML can perform against each device to register and wait for + * some event to occur. + * @{ + */ +/***************************************************************************************************/ + +/** + * Create an empty set of events. + * Event set should be freed by \ref nvmlEventSetFree + * + * For Fermi &tm; or newer fully supported devices. + * @param set Reference in which to return the event handle + * + * @return + * - \ref NVML_SUCCESS if the event has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a set is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlEventSetFree + */ +nvmlReturn_t DECLDIR nvmlEventSetCreate(nvmlEventSet_t *set); + +/** + * Starts recording of events on a specified devices and add the events to specified \ref nvmlEventSet_t + * + * For Fermi &tm; or newer fully supported devices. + * ECC events are available only on ECC-enabled devices (see \ref nvmlDeviceGetTotalEccErrors) + * Power capping events are available only on Power Management enabled devices (see \ref nvmlDeviceGetPowerManagementMode) + * + * For Linux only. + * + * This call starts recording of events on specific device. + * All events that occurred before this call are not recorded. + * Checking if some event occurred can be done with \ref nvmlEventSetWait_v2 + * + * If function reports NVML_ERROR_UNKNOWN, event set is in undefined state and should be freed. + * If function reports NVML_ERROR_NOT_SUPPORTED, event set can still be used. None of the requested eventTypes + * are registered in that case. + * + * @param device The identifier of the target device + * @param eventTypes Bitmask of \ref nvmlEventType to record + * @param set Set to which add new event types + * + * @return + * - \ref NVML_SUCCESS if the event has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a eventTypes is invalid or \a set is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the platform does not support this feature or some of requested event types + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlEventType + * @see nvmlDeviceGetSupportedEventTypes + * @see nvmlEventSetWait + * @see nvmlEventSetFree + */ +nvmlReturn_t DECLDIR nvmlDeviceRegisterEvents(nvmlDevice_t device, unsigned long long eventTypes, nvmlEventSet_t set); + +/** + * Returns information about events supported on device + * + * For Fermi &tm; or newer fully supported devices. + * + * Events are not supported on Windows. So this function returns an empty mask in \a eventTypes on Windows. + * + * @param device The identifier of the target device + * @param eventTypes Reference in which to return bitmask of supported events + * + * @return + * - \ref NVML_SUCCESS if the eventTypes has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a eventType is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlEventType + * @see nvmlDeviceRegisterEvents + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedEventTypes(nvmlDevice_t device, unsigned long long *eventTypes); + +/** + * Waits on events and delivers events + * + * For Fermi &tm; or newer fully supported devices. + * + * If some events are ready to be delivered at the time of the call, function returns immediately. + * If there are no events ready to be delivered, function sleeps till event arrives + * but not longer than specified timeout. This function in certain conditions can return before + * specified timeout passes (e.g. when interrupt arrives) + * + * On Windows, in case of Xid error, the function returns the most recent Xid error type seen by the system. + * If there are multiple Xid errors generated before nvmlEventSetWait is invoked then the last seen Xid error + * type is returned for all Xid error events. + * + * On Linux, every Xid error event would return the associated event data and other information if applicable. + * + * In MIG mode, if device handle is provided, the API reports all the events for the available instances, + * only if the caller has appropriate privileges. In absence of required privileges, only the events which + * affect all the instances (i.e. whole device) are reported. + * + * This API does not currently support per-instance event reporting using MIG device handles. + * + * @param set Reference to set of events to wait on + * @param data Reference in which to return event data + * @param timeoutms Maximum amount of wait time in milliseconds for registered event + * + * @return + * - \ref NVML_SUCCESS if the data has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a data is NULL + * - \ref NVML_ERROR_TIMEOUT if no event arrived in specified timeout or interrupt arrived + * - \ref NVML_ERROR_GPU_IS_LOST if a GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlEventType + * @see nvmlDeviceRegisterEvents + */ +nvmlReturn_t DECLDIR nvmlEventSetWait_v2(nvmlEventSet_t set, nvmlEventData_t * data, unsigned int timeoutms); + +/** + * Releases events in the set + * + * For Fermi &tm; or newer fully supported devices. + * + * @param set Reference to events to be released + * + * @return + * - \ref NVML_SUCCESS if the event has been successfully released + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceRegisterEvents + */ +nvmlReturn_t DECLDIR nvmlEventSetFree(nvmlEventSet_t set); + +/** + * Create an empty set of system events. + * Event set should be freed by \ref nvmlSystemEventSetFree + * + * For Fermi &tm; or newer fully supported devices. + * @param request Reference to nvmlSystemEventSetCreateRequest_t + * + * @return + * - \ref NVML_SUCCESS if the event has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if request is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH for unsupported version + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlSystemEventSetFree + */ +nvmlReturn_t DECLDIR nvmlSystemEventSetCreate(nvmlSystemEventSetCreateRequest_t *request); + +/** + * Releases system event set + * + * For Fermi &tm; or newer fully supported devices. + * + * @param request Reference to nvmlSystemEventSetFreeRequest_t + * + * @return + * - \ref NVML_SUCCESS if the event has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if request is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH for unsupported version + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceRegisterEvents + */ +nvmlReturn_t DECLDIR nvmlSystemEventSetFree(nvmlSystemEventSetFreeRequest_t *request); + +/** + * Starts recording of events on system and add the events to specified \ref nvmlSystemEventSet_t + * + * For Linux only. + * + * This call starts recording of events on specific device. + * All events that occurred before this call are not recorded. + * Checking if some event occurred can be done with \ref nvmlSystemEventSetWait + * + * If function reports NVML_ERROR_UNKNOWN, event set is in undefined state and should be freed. + * If function reports NVML_ERROR_NOT_SUPPORTED, event set can still be used. None of the requested eventTypes + * are registered in that case. + * + * @param request Reference to the struct nvmlSystemRegisterEventRequest_t + * + * @return + * - \ref NVML_SUCCESS if the event has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if request is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH for unsupported version + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlSystemEventType + * @see nvmlSystemEventSetWait + * @see nvmlEventSetFree + */ +nvmlReturn_t DECLDIR nvmlSystemRegisterEvents(nvmlSystemRegisterEventRequest_t *request); + +/** + * Waits on system events and delivers events + * + * For Fermi &tm; or newer fully supported devices. + * + * If some events are ready to be delivered at the time of the call, function returns immediately. + * If there are no events ready to be delivered, function sleeps till event arrives + * but not longer than specified timeout. This function in certain conditions can return before + * specified timeout passes (e.g. when interrupt arrives) + * + * if the return request->numEvent equals to request->dataSize, there might be outstanding + * event, it is recommended to call nvmlSystemEventSetWait again to query all the events. + * + * @param request Reference in which to nvmlSystemEventSetWaitRequest_t + * + * @return + * - \ref NVML_SUCCESS if the event has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if request is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH for unsupported version + * - \ref NVML_ERROR_TIMEOUT if no event notification after timeoutms + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlSystemEventType + * @see nvmlSystemRegisterEvents + */ +nvmlReturn_t DECLDIR nvmlSystemEventSetWait(nvmlSystemEventSetWaitRequest_t *request); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlZPI Drain states + * This chapter describes methods that NVML can perform against each device to control their drain state + * and recognition by NVML and NVIDIA kernel driver. These methods can be used with out-of-band tools to + * power on/off GPUs, enable robust reset scenarios, etc. + * @{ + */ +/***************************************************************************************************/ + +/** + * Modify the drain state of a GPU. This method forces a GPU to no longer accept new incoming requests. + * Any new NVML process will no longer see this GPU. Persistence mode for this GPU must be turned off before + * this call is made. + * Must be called as administrator. + * For Linux only. + * + * For Pascal &tm; or newer fully supported devices. + * Some Kepler devices supported. + * + * @param pciInfo The PCI address of the GPU drain state to be modified + * @param newState The drain state that should be entered, see \ref nvmlEnableState_t + * + * @return + * - \ref NVML_SUCCESS if counters were successfully reset + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a nvmlIndex or \a newState is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the calling process has insufficient permissions to perform operation + * - \ref NVML_ERROR_IN_USE if the device has persistence mode turned on + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceModifyDrainState (nvmlPciInfo_t *pciInfo, nvmlEnableState_t newState); + +/** + * Query the drain state of a GPU. This method is used to check if a GPU is in a currently draining + * state. + * For Linux only. + * + * For Pascal &tm; or newer fully supported devices. + * Some Kepler devices supported. + * + * @param pciInfo The PCI address of the GPU drain state to be queried + * @param currentState The current drain state for this GPU, see \ref nvmlEnableState_t + * + * @return + * - \ref NVML_SUCCESS if counters were successfully reset + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a nvmlIndex or \a currentState is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceQueryDrainState (nvmlPciInfo_t *pciInfo, nvmlEnableState_t *currentState); + +/** + * This method will remove the specified GPU from the view of both NVML and the NVIDIA kernel driver + * as long as no other processes are attached. If other processes are attached, this call will return + * NVML_ERROR_IN_USE and the GPU will be returned to its original "draining" state. Note: the + * only situation where a process can still be attached after nvmlDeviceModifyDrainState() is called + * to initiate the draining state is if that process was using, and is still using, a GPU before the + * call was made. Also note, persistence mode counts as an attachment to the GPU thus it must be disabled + * prior to this call. + * + * For long-running NVML processes please note that this will change the enumeration of current GPUs. + * For example, if there are four GPUs present and GPU1 is removed, the new enumeration will be 0-2. + * Also, device handles after the removed GPU will not be valid and must be re-established. + * Must be run as administrator. + * For Linux only. + * + * For Pascal &tm; or newer fully supported devices. + * Some Kepler devices supported. + * + * @param pciInfo The PCI address of the GPU to be removed + * @param gpuState Whether the GPU is to be removed, from the OS + * see \ref nvmlDetachGpuState_t + * @param linkState Requested upstream PCIe link state, see \ref nvmlPcieLinkState_t + * + * @return + * - \ref NVML_SUCCESS if counters were successfully reset + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a nvmlIndex is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_IN_USE if the device is still in use and cannot be removed + */ +nvmlReturn_t DECLDIR nvmlDeviceRemoveGpu_v2(nvmlPciInfo_t *pciInfo, nvmlDetachGpuState_t gpuState, nvmlPcieLinkState_t linkState); + +/** + * Request the OS and the NVIDIA kernel driver to rediscover a portion of the PCI subsystem looking for GPUs that + * were previously removed. The portion of the PCI tree can be narrowed by specifying a domain, bus, and device. + * If all are zeroes then the entire PCI tree will be searched. Please note that for long-running NVML processes + * the enumeration will change based on how many GPUs are discovered and where they are inserted in bus order. + * + * In addition, all newly discovered GPUs will be initialized and their ECC scrubbed which may take several seconds + * per GPU. Also, all device handles are no longer guaranteed to be valid post discovery. + * + * Must be run as administrator. + * For Linux only. + * + * For Pascal &tm; or newer fully supported devices. + * Some Kepler devices supported. + * + * @param pciInfo The PCI tree to be searched. Only the domain, bus, and device + * fields are used in this call. + * + * @return + * - \ref NVML_SUCCESS if counters were successfully reset + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pciInfo is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the operating system does not support this feature + * - \ref NVML_ERROR_OPERATING_SYSTEM if the operating system is denying this feature + * - \ref NVML_ERROR_NO_PERMISSION if the calling process has insufficient permissions to perform operation + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceDiscoverGpus (nvmlPciInfo_t *pciInfo); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlFieldValueQueries Field Value Queries + * This chapter describes NVML operations that are associated with retrieving Field Values from NVML + * @{ + */ +/***************************************************************************************************/ + +/** + * Request values for a list of fields for a device. This API allows multiple fields to be queried at once. + * If any of the underlying fieldIds are populated by the same driver call, the results for those field IDs + * will be populated from a single call rather than making a driver call for each fieldId. + * + * @param device The device handle of the GPU to request field values for + * @param valuesCount Number of entries in values that should be retrieved + * @param values Array of \a valuesCount structures to hold field values. + * Each value's fieldId must be populated prior to this call + * + * @return + * - \ref NVML_SUCCESS if any values in \a values were populated. Note that you must + * check the nvmlReturn field of each value for each individual + * status + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a values is NULL + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFieldValues(nvmlDevice_t device, int valuesCount, nvmlFieldValue_t *values); + +/** + * Clear values for a list of fields for a device. This API allows multiple fields to be cleared at once. + * + * @param device The device handle of the GPU to request field values for + * @param valuesCount Number of entries in values that should be cleared + * @param values Array of \a valuesCount structures to hold field values. + * Each value's fieldId must be populated prior to this call + * + * @return + * - \ref NVML_SUCCESS if any values in \a values were cleared. Note that you must + * check the nvmlReturn field of each value for each individual + * status + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a values is NULL + */ +nvmlReturn_t DECLDIR nvmlDeviceClearFieldValues(nvmlDevice_t device, int valuesCount, nvmlFieldValue_t *values); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlVirtualGpuQueries vGPU APIs + * This chapter describes operations that are associated with NVIDIA vGPU Software products. + * @{ + */ +/***************************************************************************************************/ + +/** + * This method is used to get the virtualization mode corresponding to the GPU. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device Identifier of the target device + * @param pVirtualMode Reference to virtualization mode. One of NVML_GPU_VIRTUALIZATION_? + * + * @return + * - \ref NVML_SUCCESS if \a pVirtualMode is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pVirtualMode is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVirtualizationMode(nvmlDevice_t device, nvmlGpuVirtualizationMode_t *pVirtualMode); + +/** + * Queries if SR-IOV host operation is supported on a vGPU supported device. + * + * Checks whether SR-IOV host capability is supported by the device and the + * driver, and indicates device is in SR-IOV mode if both of these conditions + * are true. + * + * @param device The identifier of the target device + * @param pHostVgpuMode Reference in which to return the current vGPU mode + * + * @return + * - \ref NVML_SUCCESS if device's vGPU mode has been successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device handle is 0 or \a pVgpuMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device doesn't support this feature. + * - \ref NVML_ERROR_UNKNOWN if any unexpected error occurred + */ +nvmlReturn_t DECLDIR nvmlDeviceGetHostVgpuMode(nvmlDevice_t device, nvmlHostVgpuMode_t *pHostVgpuMode); + +/** + * This method is used to set the virtualization mode corresponding to the GPU. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device Identifier of the target device + * @param virtualMode virtualization mode. One of NVML_GPU_VIRTUALIZATION_? + * + * @return + * - \ref NVML_SUCCESS if \a virtualMode is set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a virtualMode is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if setting of virtualization mode is not supported. + * - \ref NVML_ERROR_NO_PERMISSION if setting of virtualization mode is not allowed for this client. + */ +nvmlReturn_t DECLDIR nvmlDeviceSetVirtualizationMode(nvmlDevice_t device, nvmlGpuVirtualizationMode_t virtualMode); + +/** + * Get the vGPU heterogeneous mode for the device. + * + * When in heterogeneous mode, a vGPU can concurrently host timesliced vGPUs with differing framebuffer sizes. + * + * On successful return, the function returns \a pHeterogeneousMode->mode with the current vGPU heterogeneous mode. + * \a pHeterogeneousMode->version is the version number of the structure nvmlVgpuHeterogeneousMode_t, the caller should + * set the correct version number to retrieve the vGPU heterogeneous mode. + * \a pHeterogeneousMode->mode can either be \ref NVML_FEATURE_ENABLED or \ref NVML_FEATURE_DISABLED. + * + * @param device The identifier of the target device + * @param pHeterogeneousMode Pointer to the caller-provided structure of nvmlVgpuHeterogeneousMode_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid or \a pHeterogeneousMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or \a device doesn't support this feature + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pHeterogeneousMode is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuHeterogeneousMode(nvmlDevice_t device, nvmlVgpuHeterogeneousMode_t *pHeterogeneousMode); + +/** + * Enable or disable vGPU heterogeneous mode for the device. + * + * When in heterogeneous mode, a vGPU can concurrently host timesliced vGPUs with differing framebuffer sizes. + * + * API would return an appropriate error code upon unsuccessful activation. For example, the heterogeneous mode + * set will fail with error \ref NVML_ERROR_IN_USE if any vGPU instance is active on the device. The caller of this API + * is expected to shutdown the vGPU VMs and retry setting the \a mode. + * On KVM platform, setting heterogeneous mode is allowed, if no MDEV device is created on the device, else will fail + * with same error \ref NVML_ERROR_IN_USE. + * On successful return, the function updates the vGPU heterogeneous mode with the user provided \a pHeterogeneousMode->mode. + * \a pHeterogeneousMode->version is the version number of the structure nvmlVgpuHeterogeneousMode_t, the caller should + * set the correct version number to set the vGPU heterogeneous mode. + * + * @param device Identifier of the target device + * @param pHeterogeneousMode Pointer to the caller-provided structure of nvmlVgpuHeterogeneousMode_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device or \a pHeterogeneousMode is NULL or \a pHeterogeneousMode->mode is invalid + * - \ref NVML_ERROR_IN_USE If the \a device is in use + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or \a device doesn't support this feature + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pHeterogeneousMode is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetVgpuHeterogeneousMode(nvmlDevice_t device, const nvmlVgpuHeterogeneousMode_t *pHeterogeneousMode); + +/** + * Query the placement ID of active vGPU instance. + * + * When in vGPU heterogeneous mode, this function returns a valid placement ID as \a pPlacement->placementId + * else NVML_INVALID_VGPU_PLACEMENT_ID is returned. + * \a pPlacement->version is the version number of the structure nvmlVgpuPlacementId_t, the caller should + * set the correct version number to get placement id of the vGPU instance \a vgpuInstance. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param pPlacement Pointer to vGPU placement ID structure \a nvmlVgpuPlacementId_t + * + * @return + * - \ref NVML_SUCCESS If information is successfully retrieved + * - \ref NVML_ERROR_NOT_FOUND If \a vgpuInstance does not match a valid active vGPU instance + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a vgpuInstance is invalid or \a pPlacement is NULL + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pPlacement is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetPlacementId(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuPlacementId_t *pPlacement); + +/** + * Query the supported vGPU placement ID of the vGPU type. + * + * The function returns an array of supported vGPU placement IDs for the specified vGPU type ID in the buffer provided + * by the caller at \a pPlacementList->placementIds. The required memory for the placementIds array must be allocated + * based on the maximum number of vGPU type instances, which is retrievable through \ref nvmlVgpuTypeGetMaxInstances(). + * If the provided count by the caller is insufficient, the function will return NVML_ERROR_INSUFFICIENT_SIZE along with + * the number of required entries in \a pPlacementList->count. The caller should then reallocate a buffer with the size + * of pPlacementList->count * sizeof(pPlacementList->placementIds) and invoke the function again. + * + * To obtain a list of homogeneous placement IDs, the caller needs to set \a pPlacementList->mode to NVML_VGPU_PGPU_HOMOGENEOUS_MODE. + * For heterogeneous placement IDs, \a pPlacementList->mode should be set to NVML_VGPU_PGPU_HETEROGENEOUS_MODE. + * By default, a list of heterogeneous placement IDs is returned. + * + * @param device Identifier of the target device + * @param vgpuTypeId Handle to vGPU type. The vGPU type ID + * @param pPlacementList Pointer to the vGPU placement structure \a nvmlVgpuPlacementList_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device or \a vgpuTypeId is invalid or \a pPlacementList is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device or \a vgpuTypeId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pPlacementList is invalid + * - \ref NVML_ERROR_INSUFFICIENT_SIZE If the buffer is small, element count is returned in \a pPlacementList->count + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuTypeSupportedPlacements(nvmlDevice_t device, nvmlVgpuTypeId_t vgpuTypeId, nvmlVgpuPlacementList_t *pPlacementList); + +/** + * Query the creatable vGPU placement ID of the vGPU type. + * + * An array of creatable vGPU placement IDs for the vGPU type ID indicated by \a vgpuTypeId is returned in the + * caller-supplied buffer of \a pPlacementList->placementIds. Memory needed for the placementIds array should be + * allocated based on maximum instances of a vGPU type which can be queried via \ref nvmlVgpuTypeGetMaxInstances(). + * If the provided count by the caller is insufficient, the function will return NVML_ERROR_INSUFFICIENT_SIZE along with + * the number of required entries in \a pPlacementList->count. The caller should then reallocate a buffer with the size + * of pPlacementList->count * sizeof(pPlacementList->placementIds) and invoke the function again. + * + * The creatable vGPU placement IDs may differ over time, as there may be restrictions on what type of vGPU the + * vGPU instance is running. + * + * @param device The identifier of the target device + * @param vgpuTypeId Handle to vGPU type. The vGPU type ID + * @param pPlacementList Pointer to the list of vGPU placement structure \a nvmlVgpuPlacementList_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device or \a vgpuTypeId is invalid or \a pPlacementList is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or \a device or \a vgpuTypeId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pPlacementList is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuTypeCreatablePlacements(nvmlDevice_t device, nvmlVgpuTypeId_t vgpuTypeId, nvmlVgpuPlacementList_t *pPlacementList); + +/** + * Retrieve the static GSP heap size of the vGPU type in bytes + * + * @param vgpuTypeId Handle to vGPU type + * @param gspHeapSize Reference to return the GSP heap size value + * @return + * - \ref NVML_SUCCESS Successful completion + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a vgpuTypeId is invalid, or \a gspHeapSize is NULL + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetGspHeapSize(nvmlVgpuTypeId_t vgpuTypeId, unsigned long long *gspHeapSize); + +/** + * Retrieve the static framebuffer reservation of the vGPU type in bytes + * + * @param vgpuTypeId Handle to vGPU type + * @param fbReservation Reference to return the framebuffer reservation + * @return + * - \ref NVML_SUCCESS Successful completion + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a vgpuTypeId is invalid, or \a fbReservation is NULL + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetFbReservation(nvmlVgpuTypeId_t vgpuTypeId, unsigned long long *fbReservation); + +/** + * Retrieve the currently used runtime state size of the vGPU instance + * + * This size represents the maximum in-memory data size utilized by a vGPU instance during standard operation. + * This measurement is exclusive of frame buffer (FB) data size assigned to the vGPU instance. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param pState Pointer to the vGPU runtime state's structure \a nvmlVgpuRuntimeState_t + * + * @return + * - \ref NVML_SUCCESS If information is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a vgpuInstance is invalid, or \a pState is NULL + * - \ref NVML_ERROR_NOT_FOUND If \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pState is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetRuntimeStateSize(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuRuntimeState_t *pState); + +/** + * Set the desirable vGPU capability of a device + * + * Refer to the \a nvmlDeviceVgpuCapability_t structure for the specific capabilities that can be set. + * See \ref nvmlEnableState_t for available state. + * + * @param device The identifier of the target device + * @param capability Specifies the \a nvmlDeviceVgpuCapability_t to be set + * @param state The target capability mode + * + * @return + * - \ref NVML_SUCCESS Successful completion + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid, or \a capability is invalid, or \a state is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported in current state, or \a device not in vGPU mode + * - \ref NVML_ERROR_UNKNOWN On any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceSetVgpuCapabilities(nvmlDevice_t device, nvmlDeviceVgpuCapability_t capability, nvmlEnableState_t state); + +/** + * Retrieve the vGPU Software licensable features. + * + * Identifies whether the system supports vGPU Software Licensing. If it does, return the list of licensable feature(s) + * and their current license status. + * + * @param device Identifier of the target device + * @param pGridLicensableFeatures Pointer to structure in which vGPU software licensable features are returned + * + * @return + * - \ref NVML_SUCCESS if licensable features are successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pGridLicensableFeatures is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures_v4(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlVgpu vGPU Management + * @{ + * + * This chapter describes APIs supporting NVIDIA vGPU. + */ +/***************************************************************************************************/ + +/** + * Retrieve the requested vGPU driver capability. + * + * Refer to the \a nvmlVgpuDriverCapability_t structure for the specific capabilities that can be queried. + * The return value in \a capResult should be treated as a boolean, with a non-zero value indicating that the capability + * is supported. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param capability Specifies the \a nvmlVgpuDriverCapability_t to be queried + * @param capResult A boolean for the queried capability indicating that feature is supported + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a capability is invalid, or \a capResult is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED the API is not supported in current state or \a devices not in vGPU mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlGetVgpuDriverCapabilities(nvmlVgpuDriverCapability_t capability, unsigned int *capResult); + +/** + * Retrieve the requested vGPU capability for GPU. + * + * Refer to the \a nvmlDeviceVgpuCapability_t structure for the specific capabilities that can be queried. + * The return value in \a capResult reports a non-zero value indicating that the capability + * is supported, and also reports the capability's data based on the queried capability. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param capability Specifies the \a nvmlDeviceVgpuCapability_t to be queried + * @param capResult Specifies that the queried capability is supported, and also returns capability's data + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a capability is invalid, or \a capResult is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED the API is not supported in current state or \a device not in vGPU mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuCapabilities(nvmlDevice_t device, nvmlDeviceVgpuCapability_t capability, unsigned int *capResult); + +/** + * Retrieve the supported vGPU types on a physical GPU (device). + * + * An array of supported vGPU types for the physical GPU indicated by \a device is returned in the caller-supplied buffer + * pointed at by \a vgpuTypeIds. The element count of nvmlVgpuTypeId_t array is passed in \a vgpuCount, and \a vgpuCount + * is used to return the number of vGPU types written to the buffer. + * + * If the supplied buffer is not large enough to accommodate the vGPU type array, the function returns + * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlVgpuTypeId_t array required in \a vgpuCount. + * To query the number of vGPU types supported for the GPU, call this function with *vgpuCount = 0. + * The code will return NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if no vGPU types are supported. + * + * @param device The identifier of the target device + * @param vgpuCount Pointer to caller-supplied array size, and returns number of vGPU types + * @param vgpuTypeIds Pointer to caller-supplied array in which to return list of vGPU types + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_INSUFFICIENT_SIZE \a vgpuTypeIds buffer is too small, array element count is returned in \a vgpuCount + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuCount is NULL or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedVgpus(nvmlDevice_t device, unsigned int *vgpuCount, nvmlVgpuTypeId_t *vgpuTypeIds); + +/** + * Retrieve the currently creatable vGPU types on a physical GPU (device). + * + * An array of creatable vGPU types for the physical GPU indicated by \a device is returned in the caller-supplied buffer + * pointed at by \a vgpuTypeIds. The element count of nvmlVgpuTypeId_t array is passed in \a vgpuCount, and \a vgpuCount + * is used to return the number of vGPU types written to the buffer. + * + * The creatable vGPU types for a device may differ over time, as there may be restrictions on what type of vGPU types + * can concurrently run on a device. For example, if only one vGPU type is allowed at a time on a device, then the creatable + * list will be restricted to whatever vGPU type is already running on the device. + * + * If the supplied buffer is not large enough to accommodate the vGPU type array, the function returns + * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlVgpuTypeId_t array required in \a vgpuCount. + * To query the number of vGPU types that can be created for the GPU, call this function with *vgpuCount = 0. + * The code will return NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if no vGPU types are creatable. + * + * @param device The identifier of the target device + * @param vgpuCount Pointer to caller-supplied array size, and returns number of vGPU types + * @param vgpuTypeIds Pointer to caller-supplied array in which to return list of vGPU types + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_INSUFFICIENT_SIZE \a vgpuTypeIds buffer is too small, array element count is returned in \a vgpuCount + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuCount is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCreatableVgpus(nvmlDevice_t device, unsigned int *vgpuCount, nvmlVgpuTypeId_t *vgpuTypeIds); + +/** + * Retrieve the class of a vGPU type. It will not exceed 64 characters in length (including the NUL terminator). + * See \ref nvmlConstants::NVML_DEVICE_NAME_BUFFER_SIZE. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param vgpuTypeClass Pointer to string array to return class in + * @param size Size of string + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a vgpuTypeClass is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetClass(nvmlVgpuTypeId_t vgpuTypeId, char *vgpuTypeClass, unsigned int *size); + +/** + * Retrieve the vGPU type name. + * + * The name is an alphanumeric string that denotes a particular vGPU, e.g. GRID M60-2Q. It will not + * exceed 64 characters in length (including the NUL terminator). See \ref + * nvmlConstants::NVML_DEVICE_NAME_BUFFER_SIZE. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param vgpuTypeName Pointer to buffer to return name + * @param size Size of buffer + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a name is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetName(nvmlVgpuTypeId_t vgpuTypeId, char *vgpuTypeName, unsigned int *size); + +/** + * Retrieve the GPU Instance Profile ID for the given vGPU type ID. + * The API will return a valid GPU Instance Profile ID for the MIG capable vGPU types, else INVALID_GPU_INSTANCE_PROFILE_ID is + * returned. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param gpuInstanceProfileId GPU Instance Profile ID + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device is not in vGPU Host virtualization mode + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a gpuInstanceProfileId is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetGpuInstanceProfileId(nvmlVgpuTypeId_t vgpuTypeId, unsigned int *gpuInstanceProfileId); + +/** + * Retrieve the device ID of a vGPU type. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param deviceID Device ID and vendor ID of the device contained in single 32 bit value + * @param subsystemID Subsystem ID and subsystem vendor ID of the device contained in single 32 bit value + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a deviceId or \a subsystemID are NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetDeviceID(nvmlVgpuTypeId_t vgpuTypeId, unsigned long long *deviceID, unsigned long long *subsystemID); + +/** + * Retrieve the vGPU framebuffer size in bytes. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param fbSize Pointer to framebuffer size in bytes + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a fbSize is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetFramebufferSize(nvmlVgpuTypeId_t vgpuTypeId, unsigned long long *fbSize); + +/** + * Retrieve count of vGPU's supported display heads. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param numDisplayHeads Pointer to number of display heads + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a numDisplayHeads is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetNumDisplayHeads(nvmlVgpuTypeId_t vgpuTypeId, unsigned int *numDisplayHeads); + +/** + * Retrieve vGPU display head's maximum supported resolution. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param displayIndex Zero-based index of display head + * @param xdim Pointer to maximum number of pixels in X dimension + * @param ydim Pointer to maximum number of pixels in Y dimension + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a xdim or \a ydim are NULL, or \a displayIndex + * is out of range. + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetResolution(nvmlVgpuTypeId_t vgpuTypeId, unsigned int displayIndex, unsigned int *xdim, unsigned int *ydim); + +/** + * Retrieve license requirements for a vGPU type + * + * The license type and version required to run the specified vGPU type is returned as an alphanumeric string, in the form + * ",", for example "GRID-Virtual-PC,2.0". If a vGPU is runnable with* more than one type of license, + * the licenses are delimited by a semicolon, for example "GRID-Virtual-PC,2.0;GRID-Virtual-WS,2.0;GRID-Virtual-WS-Ext,2.0". + * + * The total length of the returned string will not exceed 128 characters, including the NUL terminator. + * See \ref nvmlVgpuConstants::NVML_GRID_LICENSE_BUFFER_SIZE. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param vgpuTypeLicenseString Pointer to buffer to return license info + * @param size Size of \a vgpuTypeLicenseString buffer + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a vgpuTypeLicenseString is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetLicense(nvmlVgpuTypeId_t vgpuTypeId, char *vgpuTypeLicenseString, unsigned int size); + +/** + * Retrieve the static frame rate limit value of the vGPU type + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param frameRateLimit Reference to return the frame rate limit value + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_NOT_SUPPORTED if frame rate limiter is turned off for the vGPU type + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a frameRateLimit is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetFrameRateLimit(nvmlVgpuTypeId_t vgpuTypeId, unsigned int *frameRateLimit); + +/** + * Retrieve the maximum number of vGPU instances creatable on a device for given vGPU type + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param vgpuTypeId Handle to vGPU type + * @param vgpuInstanceCount Pointer to get the max number of vGPU instances + * that can be created on a deicve for given vgpuTypeId + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid or is not supported on target device, + * or \a vgpuInstanceCount is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetMaxInstances(nvmlDevice_t device, nvmlVgpuTypeId_t vgpuTypeId, unsigned int *vgpuInstanceCount); + +/** + * Retrieve the maximum number of vGPU instances supported per VM for given vGPU type + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param vgpuInstanceCountPerVm Pointer to get the max number of vGPU instances supported per VM for given \a vgpuTypeId + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a vgpuInstanceCountPerVm is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetMaxInstancesPerVm(nvmlVgpuTypeId_t vgpuTypeId, unsigned int *vgpuInstanceCountPerVm); + +/** + * Retrieve the BAR1 info for given vGPU type. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param bar1Info Pointer to the vGPU type BAR1 information structure \a nvmlVgpuTypeBar1Info_t + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a bar1Info is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetBAR1Info(nvmlVgpuTypeId_t vgpuTypeId, nvmlVgpuTypeBar1Info_t *bar1Info); + +/** + * Retrieve the active vGPU instances on a device. + * + * An array of active vGPU instances is returned in the caller-supplied buffer pointed at by \a vgpuInstances. The + * array element count is passed in \a vgpuCount, and \a vgpuCount is used to return the number of vGPU instances + * written to the buffer. + * + * If the supplied buffer is not large enough to accommodate the vGPU instance array, the function returns + * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlVgpuInstance_t array required in \a vgpuCount. + * To query the number of active vGPU instances, call this function with *vgpuCount = 0. The code will return + * NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if no vGPU Types are supported. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param vgpuCount Pointer which passes in the array size as well as get + * back the number of types + * @param vgpuInstances Pointer to array in which to return list of vGPU instances + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a vgpuCount is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetActiveVgpus(nvmlDevice_t device, unsigned int *vgpuCount, nvmlVgpuInstance_t *vgpuInstances); + +/** + * Retrieve the VM ID associated with a vGPU instance. + * + * The VM ID is returned as a string, not exceeding 80 characters in length (including the NUL terminator). + * See \ref nvmlConstants::NVML_DEVICE_UUID_BUFFER_SIZE. + * + * The format of the VM ID varies by platform, and is indicated by the type identifier returned in \a vmIdType. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param vmId Pointer to caller-supplied buffer to hold VM ID + * @param size Size of buffer in bytes + * @param vmIdType Pointer to hold VM ID type + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vmId or \a vmIdType is NULL, or \a vgpuInstance is 0 + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetVmID(nvmlVgpuInstance_t vgpuInstance, char *vmId, unsigned int size, nvmlVgpuVmIdType_t *vmIdType); + +/** + * Retrieve the UUID of a vGPU instance. + * + * The UUID is a globally unique identifier associated with the vGPU, and is returned as a 5-part hexadecimal string, + * not exceeding 80 characters in length (including the NULL terminator). + * See \ref nvmlConstants::NVML_DEVICE_UUID_BUFFER_SIZE. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param uuid Pointer to caller-supplied buffer to hold vGPU UUID + * @param size Size of buffer in bytes + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a uuid is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetUUID(nvmlVgpuInstance_t vgpuInstance, char *uuid, unsigned int size); + +/** + * Retrieve the NVIDIA driver version installed in the VM associated with a vGPU. + * + * The version is returned as an alphanumeric string in the caller-supplied buffer \a version. The length of the version + * string will not exceed 80 characters in length (including the NUL terminator). + * See \ref nvmlConstants::NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE. + * + * nvmlVgpuInstanceGetVmDriverVersion() may be called at any time for a vGPU instance. The guest VM driver version is + * returned as "Not Available" if no NVIDIA driver is installed in the VM, or the VM has not yet booted to the point where the + * NVIDIA driver is loaded and initialized. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param version Caller-supplied buffer to return driver version string + * @param length Size of \a version buffer + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0 + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetVmDriverVersion(nvmlVgpuInstance_t vgpuInstance, char* version, unsigned int length); + +/** + * Retrieve the framebuffer usage in bytes. + * + * Framebuffer usage is the amont of vGPU framebuffer memory that is currently in use by the VM. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuInstance The identifier of the target instance + * @param fbUsage Pointer to framebuffer usage in bytes + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a fbUsage is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFbUsage(nvmlVgpuInstance_t vgpuInstance, unsigned long long *fbUsage); + +/** + * @deprecated Use \ref nvmlVgpuInstanceGetLicenseInfo_v2. + * + * Retrieve the current licensing state of the vGPU instance. + * + * If the vGPU is currently licensed, \a licensed is set to 1, otherwise it is set to 0. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param licensed Reference to return the licensing status + * + * @return + * - \ref NVML_SUCCESS if \a licensed has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a licensed is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +DEPRECATED(13.0) nvmlReturn_t DECLDIR nvmlVgpuInstanceGetLicenseStatus(nvmlVgpuInstance_t vgpuInstance, unsigned int *licensed); + +/** + * Retrieve the vGPU type of a vGPU instance. + * + * Returns the vGPU type ID of vgpu assigned to the vGPU instance. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param vgpuTypeId Reference to return the vgpuTypeId + * + * @return + * - \ref NVML_SUCCESS if \a vgpuTypeId has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a vgpuTypeId is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetType(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuTypeId_t *vgpuTypeId); + +/** + * Retrieve the frame rate limit set for the vGPU instance. + * + * Returns the value of the frame rate limit set for the vGPU instance + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param frameRateLimit Reference to return the frame rate limit + * + * @return + * - \ref NVML_SUCCESS if \a frameRateLimit has been set + * - \ref NVML_ERROR_NOT_SUPPORTED if frame rate limiter is turned off for the vGPU type + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a frameRateLimit is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFrameRateLimit(nvmlVgpuInstance_t vgpuInstance, unsigned int *frameRateLimit); + +/** + * Retrieve the current ECC mode of vGPU instance. + * + * @param vgpuInstance The identifier of the target vGPU instance + * @param eccMode Reference in which to return the current ECC mode + * + * @return + * - \ref NVML_SUCCESS if the vgpuInstance's ECC mode has been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a mode is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEccMode(nvmlVgpuInstance_t vgpuInstance, nvmlEnableState_t *eccMode); + +/** + * Retrieve the encoder capacity of a vGPU instance, as a percentage of maximum encoder capacity with valid values in the range 0-100. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param encoderCapacity Reference to an unsigned int for the encoder capacity + * + * @return + * - \ref NVML_SUCCESS if \a encoderCapacity has been retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a encoderQueryType is invalid + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderCapacity(nvmlVgpuInstance_t vgpuInstance, unsigned int *encoderCapacity); + +/** + * Set the encoder capacity of a vGPU instance, as a percentage of maximum encoder capacity with valid values in the range 0-100. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param encoderCapacity Unsigned int for the encoder capacity value + * + * @return + * - \ref NVML_SUCCESS if \a encoderCapacity has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a encoderCapacity is out of range of 0-100. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceSetEncoderCapacity(nvmlVgpuInstance_t vgpuInstance, unsigned int encoderCapacity); + +/** + * Retrieves the current encoder statistics of a vGPU Instance + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param sessionCount Reference to an unsigned int for count of active encoder sessions + * @param averageFps Reference to an unsigned int for trailing average FPS of all active sessions + * @param averageLatency Reference to an unsigned int for encode latency in microseconds + * + * @return + * - \ref NVML_SUCCESS if \a sessionCount, \a averageFps and \a averageLatency is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount , or \a averageFps or \a averageLatency is NULL + * or \a vgpuInstance is 0. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderStats(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, + unsigned int *averageFps, unsigned int *averageLatency); + +/** + * Retrieves information about all active encoder sessions on a vGPU Instance. + * + * An array of active encoder sessions is returned in the caller-supplied buffer pointed at by \a sessionInfo. The + * array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions + * written to the buffer. + * + * If the supplied buffer is not large enough to accommodate the active session array, the function returns + * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlEncoderSessionInfo_t array required in \a sessionCount. + * To query the number of active encoder sessions, call this function with *sessionCount = 0. The code will return + * NVML_SUCCESS with number of active encoder sessions updated in *sessionCount. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param sessionCount Reference to caller supplied array size, and returns + * the number of sessions. + * @param sessionInfo Reference to caller supplied array in which the list + * of session information us returned. + * + * @return + * - \ref NVML_SUCCESS if \a sessionInfo is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is + returned in \a sessionCount + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL, or \a vgpuInstance is 0. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderSessions(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, nvmlEncoderSessionInfo_t *sessionInfo); + +/** +* Retrieves the active frame buffer capture sessions statistics of a vGPU Instance +* +* For Maxwell &tm; or newer fully supported devices. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param fbcStats Reference to nvmlFBCStats_t structure containing NvFBC stats +* +* @return +* - \ref NVML_SUCCESS if \a fbcStats is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a fbcStats is NULL +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFBCStats(nvmlVgpuInstance_t vgpuInstance, nvmlFBCStats_t *fbcStats); + +/** +* Retrieves information about active frame buffer capture sessions on a vGPU Instance. +* +* An array of active FBC sessions is returned in the caller-supplied buffer pointed at by \a sessionInfo. The +* array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions +* written to the buffer. +* +* If the supplied buffer is not large enough to accommodate the active session array, the function returns +* NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlFBCSessionInfo_t array required in \a sessionCount. +* To query the number of active FBC sessions, call this function with *sessionCount = 0. The code will return +* NVML_SUCCESS with number of active FBC sessions updated in *sessionCount. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @note hResolution, vResolution, averageFPS and averageLatency data for a FBC session returned in \a sessionInfo may +* be zero if there are no new frames captured since the session started. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param sessionCount Reference to caller supplied array size, and returns the number of sessions. +* @param sessionInfo Reference in which to return the session information +* +* @return +* - \ref NVML_SUCCESS if \a sessionInfo is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a sessionCount is NULL. +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is returned in \a sessionCount +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFBCSessions(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, nvmlFBCSessionInfo_t *sessionInfo); + +/** +* Retrieve the GPU Instance ID for the given vGPU Instance. +* The API will return a valid GPU Instance ID for MIG backed vGPU Instance, else INVALID_GPU_INSTANCE_ID is returned. +* +* For Kepler &tm; or newer fully supported devices. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param gpuInstanceId GPU Instance ID +* +* @return +* - \ref NVML_SUCCESS successful completion +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a gpuInstanceId is NULL. +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetGpuInstanceId(nvmlVgpuInstance_t vgpuInstance, unsigned int *gpuInstanceId); + +/** +* Retrieves the PCI Id of the given vGPU Instance i.e. the PCI Id of the GPU as seen inside the VM. +* +* The vGPU PCI id is returned as "00000000:00:00.0" if NVIDIA driver is not installed on the vGPU instance. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param vgpuPciId Caller-supplied buffer to return vGPU PCI Id string +* @param length Size of the vgpuPciId buffer +* +* @return +* - \ref NVML_SUCCESS if vGPU PCI Id is sucessfully retrieved +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a vgpuPciId is NULL +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running on the vGPU instance +* - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small, \a length is set to required length +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetGpuPciId(nvmlVgpuInstance_t vgpuInstance, char *vgpuPciId, unsigned int *length); + +/** +* Retrieve the requested capability for a given vGPU type. Refer to the \a nvmlVgpuCapability_t structure +* for the specific capabilities that can be queried. The return value in \a capResult should be treated as +* a boolean, with a non-zero value indicating that the capability is supported. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @param vgpuTypeId Handle to vGPU type +* @param capability Specifies the \a nvmlVgpuCapability_t to be queried +* @param capResult A boolean for the queried capability indicating that feature is supported +* +* @return +* - \ref NVML_SUCCESS successful completion +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a capability is invalid, or \a capResult is NULL +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetCapabilities(nvmlVgpuTypeId_t vgpuTypeId, nvmlVgpuCapability_t capability, unsigned int *capResult); + +/** + * Retrieve the MDEV UUID of a vGPU instance. + * + * The MDEV UUID is a globally unique identifier of the mdev device assigned to the VM, and is returned as a 5-part hexadecimal string, + * not exceeding 80 characters in length (including the NULL terminator). + * MDEV UUID is displayed only on KVM platform. + * See \ref nvmlConstants::NVML_DEVICE_UUID_BUFFER_SIZE. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param mdevUuid Pointer to caller-supplied buffer to hold MDEV UUID + * @param size Size of buffer in bytes + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NOT_SUPPORTED on any hypervisor other than KVM + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a mdevUuid is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetMdevUUID(nvmlVgpuInstance_t vgpuInstance, char *mdevUuid, unsigned int size); + +/** + * Query the currently creatable vGPU types on a specific GPU Instance. + * + * The function returns an array of vGPU types that can be created for a specified GPU instance. This array is stored + * in a caller-supplied buffer, with the buffer's element count passed through \a pVgpus->vgpuCount. The number of + * vGPU types written to the buffer is indicated by \a pVgpus->vgpuCount. If the buffer is too small to hold the vGPU + * type array, the function returns NVML_ERROR_INSUFFICIENT_SIZE and updates \a pVgpus->vgpuCount with the required + * element count. + * + * To determine the creatable vGPUs for a GPU Instance, invoke this function with \a pVgpus->vgpuCount set to 0 and + * \a pVgpus->vgpuTypeIds as NULL. This will result in NVML_ERROR_INSUFFICIENT_SIZE being returned, along with the + * count value in \a pVgpus->vgpuCount. + * + * The creatable vGPU types may differ over time, as there may be restrictions on what type of vGPUs can concurrently + * run on the device. + * + * @param gpuInstance The GPU instance handle + * @param pVgpus Pointer to the caller-provided structure of nvmlVgpuTypeIdInfo_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, or \a pVgpus is NULL + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU + * - \ref NVML_ERROR_INSUFFICIENT_SIZE If \a pVgpus->vgpuTypeIds buffer is small + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pVgpus is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetCreatableVgpus(nvmlGpuInstance_t gpuInstance, nvmlVgpuTypeIdInfo_t *pVgpus); + +/** + * Retrieve the maximum number of vGPU instances per GPU instance for given vGPU type + * + * @param pMaxInstance Pointer to the caller-provided structure of nvmlVgpuTypeMaxInstance_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a pMaxInstance is NULL or \a pMaxInstance->vgpuTypeId is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU or non-MIG vGPU type + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pMaxInstance is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetMaxInstancesPerGpuInstance(nvmlVgpuTypeMaxInstance_t *pMaxInstance); + +/** + * Retrieve the active vGPU instances within a GPU instance. + * + * An array of active vGPU instances is returned in the caller-supplied buffer pointed + * at by \a pVgpuInstanceInfo->vgpuInstances. The array element count is passed in + * \a pVgpuInstanceInfo->vgpuCount, and \a pVgpuInstanceInfo->vgpuCount is used to return + * the number of vGPU instances written to the buffer. + * + * If the supplied buffer is not large enough to accommodate the vGPU instance array, + * the function returns NVML_ERROR_INSUFFICIENT_SIZE, with the element count of + * nvmlVgpuInstance_t array required in \a pVgpuInstanceInfo->vgpuCount. To query the + * number of active vGPU instances, call this function with pVgpuInstanceInfo->vgpuCount = 0 + * and pVgpuInstanceInfo->vgpuTypeIds = NULL. The code will return NVML_ERROR_INSUFFICIENT_SIZE, + * or NVML_SUCCESS if no vGPU Types are active. + * + * @param gpuInstance The GPU instance handle + * @param pVgpuInstanceInfo Pointer to the vGPU instance information structure \a nvmlActiveVgpuInstanceInfo_t + * + * @return + * - \ref NVML_SUCCESS Successful completion + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, or \a pVgpuInstanceInfo is NULL + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_INSUFFICIENT_SIZE \a pVgpuInstanceInfo->vgpuTypeIds buffer is too small, + * array element count is returned in \a pVgpuInstanceInfo->vgpuCount + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pVgpuInstanceInfo is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetActiveVgpus(nvmlGpuInstance_t gpuInstance, nvmlActiveVgpuInstanceInfo_t *pVgpuInstanceInfo); + +/** + * Set vGPU scheduler state for the given GPU instance + * + * %GB20X_OR_NEWER% + * + * Scheduler state and params will be allowed to set only when no VM is running within the GPU instance. + * In \a nvmlVgpuSchedulerState_t, IFF enableARRMode is enabled then provide the avgFactor and frequency + * as input. If enableARRMode is disabled then provide timeslice as input. + * + * The scheduler state change won't persist across module load/unload and GPU Instance creation/deletion. + * + * @param gpuInstance The GPU instance handle + * @param pScheduler Pointer to the caller-provided structure of nvmlVgpuSchedulerState_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, or \a pScheduler is NULL + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_RESET_REQUIRED If setting the state failed with fatal error, reboot is required + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU or if any vGPU instance exists + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pScheduler is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceSetVgpuSchedulerState(nvmlGpuInstance_t gpuInstance, nvmlVgpuSchedulerState_t *pScheduler); + +/** + * Returns the vGPU scheduler state for the given GPU instance. + * The information returned in \a nvmlVgpuSchedulerStateInfo_t is not relevant if the BEST EFFORT policy is set. + * + * %GB20X_OR_NEWER% + * + * @param gpuInstance The GPU instance handle + * @param pSchedulerStateInfo Reference in which \a pSchedulerStateInfo is returned + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler state is successfully obtained + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, or \a pSchedulerStateInfo is NULL + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pSchedulerStateInfo is invalid + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetVgpuSchedulerState(nvmlGpuInstance_t gpuInstance, nvmlVgpuSchedulerStateInfo_t *pSchedulerStateInfo); + +/** + * Returns the vGPU scheduler logs for the given GPU instance. + * \a pSchedulerLogInfo points to a caller-allocated structure to contain the logs. The number of elements returned will + * never exceed \a NVML_SCHEDULER_SW_MAX_LOG_ENTRIES. + * + * To get the entire logs, call the function atleast 5 times a second. + * + * %GB20X_OR_NEWER% + * + * @param gpuInstance The GPU instance handle + * @param pSchedulerLogInfo Reference in which \a pSchedulerLogInfo is written + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler logs are successfully obtained + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, or \a pSchedulerLogInfo is NULL + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pSchedulerLogInfo is invalid + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetVgpuSchedulerLog(nvmlGpuInstance_t gpuInstance, nvmlVgpuSchedulerLogInfo_t *pSchedulerLogInfo); + +/** + * Query the creatable vGPU placement ID of the vGPU type within a GPU instance. + * + * %GB20X_OR_NEWER% + * + * An array of creatable vGPU placement IDs for the vGPU type ID indicated by \a pCreatablePlacementInfo->vgpuTypeId + * is returned in the caller-supplied buffer of \a pCreatablePlacementInfo->placementIds. Memory needed for the + * placementIds array should be allocated based on maximum instances of a vGPU type per GPU instance which can be + * queried via \ref nvmlVgpuTypeGetMaxInstancesPerGpuInstance(). + * If the provided count by the caller is insufficient, the function will return NVML_ERROR_INSUFFICIENT_SIZE along with + * the number of required entries in \a pCreatablePlacementInfo->count. The caller should then reallocate a buffer with the size + * of pCreatablePlacementInfo->count * sizeof(pCreatablePlacementInfo->placementIds) and invoke the function again. + * The creatable vGPU placement IDs may differ over time, as there may be restrictions on what type of vGPU the + * vGPU instance is running. + * + * @param gpuInstance The GPU instance handle + * @param pCreatablePlacementInfo Pointer to the list of vGPU creatable placement structure \a nvmlVgpuCreatablePlacementInfo_t + * + * @return + * - \ref NVML_SUCCESS Successful completion + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, or \a pCreatablePlacementInfo is NULL + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_INSUFFICIENT_SIZE If the buffer is small, element count is returned in \a pCreatablePlacementInfo->count + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pCreatablePlacementInfo is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU or vGPU heterogeneous mode is not enabled + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetVgpuTypeCreatablePlacements(nvmlGpuInstance_t gpuInstance, nvmlVgpuCreatablePlacementInfo_t *pCreatablePlacementInfo); + +/** + * Get the vGPU heterogeneous mode for the GPU instance. + * + * When in heterogeneous mode, a vGPU can concurrently host timesliced vGPUs with differing framebuffer sizes. + * + * On successful return, the function returns \a pHeterogeneousMode->mode with the current vGPU heterogeneous mode. + * \a pHeterogeneousMode->version is the version number of the structure nvmlVgpuHeterogeneousMode_t, the caller should + * set the correct version number to retrieve the vGPU heterogeneous mode. + * \a pHeterogeneousMode->mode can either be \ref NVML_FEATURE_ENABLED or \ref NVML_FEATURE_DISABLED. + * + * %GB20X_OR_NEWER% + * + * @param gpuInstance The GPU instance handle + * @param pHeterogeneousMode Pointer to the caller-provided structure of nvmlVgpuHeterogeneousMode_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, or \a pHeterogeneousMode is NULL + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU or not in MIG mode + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pHeterogeneousMode is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetVgpuHeterogeneousMode(nvmlGpuInstance_t gpuInstance, nvmlVgpuHeterogeneousMode_t *pHeterogeneousMode); + +/** + * Enable or disable vGPU heterogeneous mode for the GPU instance. + * + * When in heterogeneous mode, a vGPU can concurrently host timesliced vGPUs with differing framebuffer sizes. + * + * API would return an appropriate error code upon unsuccessful activation. For example, the heterogeneous mode + * set will fail with error \ref NVML_ERROR_IN_USE if any vGPU instance is active within the GPU instance. + * The caller of this API is expected to shutdown the vGPU VMs and retry setting the \a mode. + * On successful return, the function updates the vGPU heterogeneous mode with the user provided \a pHeterogeneousMode->mode. + * \a pHeterogeneousMode->version is the version number of the structure nvmlVgpuHeterogeneousMode_t, the caller should + * set the correct version number to set the vGPU heterogeneous mode. + * + * @param gpuInstance The GPU instance handle + * @param pHeterogeneousMode Pointer to the caller-provided structure of nvmlVgpuHeterogeneousMode_t + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is NULL or invalid, + * or \a pHeterogeneousMode is NULL or \a pHeterogeneousMode->mode is invalid + * or GPU Instance Id is invalid + * - \ref NVML_ERROR_IN_USE If the \a gpuInstance is in use + * - \ref NVML_ERROR_NOT_SUPPORTED If not on a vGPU host or an unsupported GPU + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a pHeterogeneousMode is invalid + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceSetVgpuHeterogeneousMode(nvmlGpuInstance_t gpuInstance, const nvmlVgpuHeterogeneousMode_t *pHeterogeneousMode); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlVgpuMigration vGPU Migration + * This chapter describes operations that are associated with vGPU Migration. + * @{ + */ +/***************************************************************************************************/ + +/** + * Structure representing range of vGPU versions. + */ +typedef struct nvmlVgpuVersion_st +{ + unsigned int minVersion; //!< Minimum vGPU version. + unsigned int maxVersion; //!< Maximum vGPU version. +} nvmlVgpuVersion_t; + +/** + * vGPU metadata structure. + */ +typedef struct nvmlVgpuMetadata_st +{ + unsigned int version; //!< Current version of the structure + unsigned int revision; //!< Current revision of the structure + nvmlVgpuGuestInfoState_t guestInfoState; //!< Current state of Guest-dependent fields + char guestDriverVersion[NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE]; //!< Version of driver installed in guest + char hostDriverVersion[NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE]; //!< Version of driver installed in host + unsigned int reserved[6]; //!< Reserved for internal use + unsigned int vgpuVirtualizationCaps; //!< vGPU virtualization capabilities bitfield + unsigned int guestVgpuVersion; //!< vGPU version of guest driver + unsigned int opaqueDataSize; //!< Size of opaque data field in bytes + char opaqueData[4]; //!< Opaque data +} nvmlVgpuMetadata_t; + +/** + * Physical GPU metadata structure + */ +typedef struct nvmlVgpuPgpuMetadata_st +{ + unsigned int version; //!< Current version of the structure + unsigned int revision; //!< Current revision of the structure + char hostDriverVersion[NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE]; //!< Host driver version + unsigned int pgpuVirtualizationCaps; //!< Pgpu virtualization capabilities bitfield + unsigned int reserved[5]; //!< Reserved for internal use + nvmlVgpuVersion_t hostSupportedVgpuRange; //!< vGPU version range supported by host driver + unsigned int opaqueDataSize; //!< Size of opaque data field in bytes + char opaqueData[4]; //!< Opaque data +} nvmlVgpuPgpuMetadata_t; + +/** + * vGPU VM compatibility codes + */ +typedef enum nvmlVgpuVmCompatibility_enum +{ + NVML_VGPU_VM_COMPATIBILITY_NONE = 0x0, //!< vGPU is not runnable + NVML_VGPU_VM_COMPATIBILITY_COLD = 0x1, //!< vGPU is runnable from a cold / powered-off state (ACPI S5) + NVML_VGPU_VM_COMPATIBILITY_HIBERNATE = 0x2, //!< vGPU is runnable from a hibernated state (ACPI S4) + NVML_VGPU_VM_COMPATIBILITY_SLEEP = 0x4, //!< vGPU is runnable from a sleeped state (ACPI S3) + NVML_VGPU_VM_COMPATIBILITY_LIVE = 0x8 //!< vGPU is runnable from a live/paused (ACPI S0) +} nvmlVgpuVmCompatibility_t; + +/** + * vGPU-pGPU compatibility limit codes + */ +typedef enum nvmlVgpuPgpuCompatibilityLimitCode_enum +{ + NVML_VGPU_COMPATIBILITY_LIMIT_NONE = 0x0, //!< Compatibility is not limited. + NVML_VGPU_COMPATIBILITY_LIMIT_HOST_DRIVER = 0x1, //!< ompatibility is limited by host driver version. + NVML_VGPU_COMPATIBILITY_LIMIT_GUEST_DRIVER = 0x2, //!< Compatibility is limited by guest driver version. + NVML_VGPU_COMPATIBILITY_LIMIT_GPU = 0x4, //!< Compatibility is limited by GPU hardware. + NVML_VGPU_COMPATIBILITY_LIMIT_OTHER = 0x80000000 //!< Compatibility is limited by an undefined factor. +} nvmlVgpuPgpuCompatibilityLimitCode_t; + +/** + * vGPU-pGPU compatibility structure + */ +typedef struct nvmlVgpuPgpuCompatibility_st +{ + nvmlVgpuVmCompatibility_t vgpuVmCompatibility; //!< Compatibility of vGPU VM. See \ref nvmlVgpuVmCompatibility_t + nvmlVgpuPgpuCompatibilityLimitCode_t compatibilityLimitCode; //!< Limiting factor for vGPU-pGPU compatibility. See \ref nvmlVgpuPgpuCompatibilityLimitCode_t +} nvmlVgpuPgpuCompatibility_t; + +/** + * Returns vGPU metadata structure for a running vGPU. The structure contains information about the vGPU and its associated VM + * such as the currently installed NVIDIA guest driver version, together with host driver version and an opaque data section + * containing internal state. + * + * nvmlVgpuInstanceGetMetadata() may be called at any time for a vGPU instance. Some fields in the returned structure are + * dependent on information obtained from the guest VM, which may not yet have reached a state where that information + * is available. The current state of these dependent fields is reflected in the info structure's \ref nvmlVgpuGuestInfoState_t field. + * + * The VMM may choose to read and save the vGPU's VM info as persistent metadata associated with the VM, and provide + * it to Virtual GPU Manager when creating a vGPU for subsequent instances of the VM. + * + * The caller passes in a buffer via \a vgpuMetadata, with the size of the buffer in \a bufferSize. If the vGPU Metadata structure + * is too large to fit in the supplied buffer, the function returns NVML_ERROR_INSUFFICIENT_SIZE with the size needed + * in \a bufferSize. + * + * @param vgpuInstance vGPU instance handle + * @param vgpuMetadata Pointer to caller-supplied buffer into which vGPU metadata is written + * @param bufferSize Size of vgpuMetadata buffer + * + * @return + * - \ref NVML_SUCCESS vGPU metadata structure was successfully returned + * - \ref NVML_ERROR_INSUFFICIENT_SIZE vgpuMetadata buffer is too small, required size is returned in \a bufferSize + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a bufferSize is NULL or \a vgpuInstance is 0; if \a vgpuMetadata is NULL and the value of \a bufferSize is not 0. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetMetadata(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuMetadata_t *vgpuMetadata, unsigned int *bufferSize); + +/** + * Returns a vGPU metadata structure for the physical GPU indicated by \a device. The structure contains information about + * the GPU and the currently installed NVIDIA host driver version that's controlling it, together with an opaque data section + * containing internal state. + * + * The caller passes in a buffer via \a pgpuMetadata, with the size of the buffer in \a bufferSize. If the \a pgpuMetadata + * structure is too large to fit in the supplied buffer, the function returns NVML_ERROR_INSUFFICIENT_SIZE with the size needed + * in \a bufferSize. + * + * @param device The identifier of the target device + * @param pgpuMetadata Pointer to caller-supplied buffer into which \a pgpuMetadata is written + * @param bufferSize Pointer to size of \a pgpuMetadata buffer + * + * @return + * - \ref NVML_SUCCESS GPU metadata structure was successfully returned + * - \ref NVML_ERROR_INSUFFICIENT_SIZE pgpuMetadata buffer is too small, required size is returned in \a bufferSize + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a bufferSize is NULL or \a device is invalid; if \a pgpuMetadata is NULL and the value of \a bufferSize is not 0. + * - \ref NVML_ERROR_NOT_SUPPORTED vGPU is not supported by the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuMetadata(nvmlDevice_t device, nvmlVgpuPgpuMetadata_t *pgpuMetadata, unsigned int *bufferSize); + +/** + * Takes a vGPU instance metadata structure read from \ref nvmlVgpuInstanceGetMetadata(), and a vGPU metadata structure for a + * physical GPU read from \ref nvmlDeviceGetVgpuMetadata(), and returns compatibility information of the vGPU instance and the + * physical GPU. + * + * The caller passes in a buffer via \a compatibilityInfo, into which a compatibility information structure is written. The + * structure defines the states in which the vGPU / VM may be booted on the physical GPU. If the vGPU / VM compatibility + * with the physical GPU is limited, a limit code indicates the factor limiting compatability. + * (see \ref nvmlVgpuPgpuCompatibilityLimitCode_t for details). + * + * Note: vGPU compatibility does not take into account dynamic capacity conditions that may limit a system's ability to + * boot a given vGPU or associated VM. + * + * @param vgpuMetadata Pointer to caller-supplied vGPU metadata structure + * @param pgpuMetadata Pointer to caller-supplied GPU metadata structure + * @param compatibilityInfo Pointer to caller-supplied buffer to hold compatibility info + * + * @return + * - \ref NVML_SUCCESS vGPU metadata structure was successfully returned + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a vgpuMetadata or \a pgpuMetadata or \a bufferSize are NULL + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGetVgpuCompatibility(nvmlVgpuMetadata_t *vgpuMetadata, nvmlVgpuPgpuMetadata_t *pgpuMetadata, nvmlVgpuPgpuCompatibility_t *compatibilityInfo); + +/** + * Returns the properties of the physical GPU indicated by the device in an ascii-encoded string format. + * + * The caller passes in a buffer via \a pgpuMetadata, with the size of the buffer in \a bufferSize. If the + * string is too large to fit in the supplied buffer, the function returns NVML_ERROR_INSUFFICIENT_SIZE with the size needed + * in \a bufferSize. + * + * @param device The identifier of the target device + * @param pgpuMetadata Pointer to caller-supplied buffer into which \a pgpuMetadata is written + * @param bufferSize Pointer to size of \a pgpuMetadata buffer + * + * @return + * - \ref NVML_SUCCESS GPU metadata structure was successfully returned + * - \ref NVML_ERROR_INSUFFICIENT_SIZE \a pgpuMetadata buffer is too small, required size is returned in \a bufferSize + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a bufferSize is NULL or \a device is invalid; if \a pgpuMetadata is NULL and the value of \a bufferSize is not 0. + * - \ref NVML_ERROR_NOT_SUPPORTED If vGPU is not supported by the system + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPgpuMetadataString(nvmlDevice_t device, char *pgpuMetadata, unsigned int *bufferSize); + +/** + * Returns the vGPU Software scheduler logs. + * \a pSchedulerLog points to a caller-allocated structure to contain the logs. The number of elements returned will + * never exceed \a NVML_SCHEDULER_SW_MAX_LOG_ENTRIES. + * + * To get the entire logs, call the function atleast 5 times a second. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target \a device + * @param pSchedulerLog Reference in which \a pSchedulerLog is written + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler logs were successfully obtained + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a pSchedulerLog is NULL or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or \a device not in vGPU host mode + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuSchedulerLog(nvmlDevice_t device, nvmlVgpuSchedulerLog_t *pSchedulerLog); + +/** + * Returns the vGPU scheduler state. + * The information returned in \a nvmlVgpuSchedulerGetState_t is not relevant if the BEST EFFORT policy is set. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target \a device + * @param pSchedulerState Reference in which \a pSchedulerState is returned + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler state is successfully obtained + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a pSchedulerState is NULL or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or \a device not in vGPU host mode + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuSchedulerState(nvmlDevice_t device, nvmlVgpuSchedulerGetState_t *pSchedulerState); + +/** + * Returns the vGPU scheduler capabilities. + * The list of supported vGPU schedulers returned in \a nvmlVgpuSchedulerCapabilities_t is from + * the NVML_VGPU_SCHEDULER_POLICY_*. This list enumerates the supported scheduler policies + * if the engine is Graphics type. + * The other values in \a nvmlVgpuSchedulerCapabilities_t are also applicable if the engine is + * Graphics type. For other engine types, it is BEST EFFORT policy. + * If ARR is supported and enabled, scheduling frequency and averaging factor are applicable + * else timeSlice is applicable. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target \a device + * @param pCapabilities Reference in which \a pCapabilities is written + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler capabilities were successfully obtained + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a pCapabilities is NULL or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported in current state or \a device not in vGPU host mode + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuSchedulerCapabilities(nvmlDevice_t device, nvmlVgpuSchedulerCapabilities_t *pCapabilities); + +/** + * Sets the vGPU scheduler state. + * + * For Pascal &tm; or newer fully supported devices. + * + * The scheduler state change won't persist across module load/unload. + * Scheduler state and params will be allowed to set only when no VM is running. + * In \a nvmlVgpuSchedulerSetState_t, IFF enableARRMode is enabled then + * provide avgFactorForARR and frequency as input. If enableARRMode is disabled + * then provide timeslice as input. + * + * @param device The identifier of the target \a device + * @param pSchedulerState vGPU \a pSchedulerState to set + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler state has been successfully set + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a pSchedulerState is NULL or \a device is invalid + * - \ref NVML_ERROR_RESET_REQUIRED If setting \a pSchedulerState failed with fatal error, + * reboot is required to overcome from this error. + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or \a device not in vGPU host mode + * or if any vGPU instance currently exists on the \a device + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetVgpuSchedulerState(nvmlDevice_t device, nvmlVgpuSchedulerSetState_t *pSchedulerState); + +/* + * Virtual GPU (vGPU) version + * + * The NVIDIA vGPU Manager and the guest drivers are tagged with a range of supported vGPU versions. This determines the range of NVIDIA guest driver versions that + * are compatible for vGPU feature support with a given NVIDIA vGPU Manager. For vGPU feature support, the range of supported versions for the NVIDIA vGPU Manager + * and the guest driver must overlap. Otherwise, the guest driver fails to load in the VM. + * + * When the NVIDIA guest driver loads, either when the VM is booted or when the driver is installed or upgraded, a negotiation occurs between the guest driver + * and the NVIDIA vGPU Manager to select the highest mutually compatible vGPU version. The negotiated vGPU version stays the same across VM migration. + */ + +/** + * Query the ranges of supported vGPU versions. + * + * This function gets the linear range of supported vGPU versions that is preset for the NVIDIA vGPU Manager and the range set by an administrator. + * If the preset range has not been overridden by \ref nvmlSetVgpuVersion, both ranges are the same. + * + * The caller passes pointers to the following \ref nvmlVgpuVersion_t structures, into which the NVIDIA vGPU Manager writes the ranges: + * 1. \a supported structure that represents the preset range of vGPU versions supported by the NVIDIA vGPU Manager. + * 2. \a current structure that represents the range of supported vGPU versions set by an administrator. By default, this range is the same as the preset range. + * + * @param supported Pointer to the structure in which the preset range of vGPU versions supported by the NVIDIA vGPU Manager is written + * @param current Pointer to the structure in which the range of supported vGPU versions set by an administrator is written + * + * @return + * - \ref NVML_SUCCESS The vGPU version range structures were successfully obtained. + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported. + * - \ref NVML_ERROR_INVALID_ARGUMENT The \a supported parameter or the \a current parameter is NULL. + * - \ref NVML_ERROR_UNKNOWN An error occurred while the data was being fetched. + */ +nvmlReturn_t DECLDIR nvmlGetVgpuVersion(nvmlVgpuVersion_t *supported, nvmlVgpuVersion_t *current); + +/** + * Override the preset range of vGPU versions supported by the NVIDIA vGPU Manager with a range set by an administrator. + * + * This function configures the NVIDIA vGPU Manager with a range of supported vGPU versions set by an administrator. This range must be a subset of the + * preset range that the NVIDIA vGPU Manager supports. The custom range set by an administrator takes precedence over the preset range and is advertised to + * the guest VM for negotiating the vGPU version. See \ref nvmlGetVgpuVersion for details of how to query the preset range of versions supported. + * + * This function takes a pointer to vGPU version range structure \ref nvmlVgpuVersion_t as input to override the preset vGPU version range that the NVIDIA vGPU Manager supports. + * + * After host system reboot or driver reload, the range of supported versions reverts to the range that is preset for the NVIDIA vGPU Manager. + * + * @note 1. The range set by the administrator must be a subset of the preset range that the NVIDIA vGPU Manager supports. Otherwise, an error is returned. + * 2. If the range of supported guest driver versions does not overlap the range set by the administrator, the guest driver fails to load. + * 3. If the range of supported guest driver versions overlaps the range set by the administrator, the guest driver will load with a negotiated + * vGPU version that is the maximum value in the overlapping range. + * 4. No VMs must be running on the host when this function is called. If a VM is running on the host, the call to this function fails. + * + * @param vgpuVersion Pointer to a caller-supplied range of supported vGPU versions. + * + * @return + * - \ref NVML_SUCCESS The preset range of supported vGPU versions was successfully overridden. + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported. + * - \ref NVML_ERROR_IN_USE The range was not overridden because a VM is running on the host. + * - \ref NVML_ERROR_INVALID_ARGUMENT The \a vgpuVersion parameter specifies a range that is outside the range supported by the NVIDIA vGPU Manager or if \a vgpuVersion is NULL. + */ +nvmlReturn_t DECLDIR nvmlSetVgpuVersion(nvmlVgpuVersion_t *vgpuVersion); + +/** @} */ // @defgroup nvmlVgpuMigration vGPU Migration + +/***************************************************************************************************/ +/** @defgroup nvmlUtil vGPU Utilization and Accounting + * This chapter describes operations that are associated with vGPU Utilization and Accounting. + * @{ + */ +/***************************************************************************************************/ + +/** + * Retrieves current utilization for vGPUs on a physical GPU (device). + * + * For Kepler &tm; or newer fully supported devices. + * + * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, and video decoder for vGPU instances running + * on a device. Utilization values are returned as an array of utilization sample structures in the caller-supplied buffer + * pointed at by \a utilizationSamples. One utilization sample structure is returned per vGPU instance, and includes the + * CPU timestamp at which the samples were recorded. Individual utilization values are returned as "unsigned int" values + * in nvmlValue_t unions. The function sets the caller-supplied \a sampleValType to NVML_VALUE_TYPE_UNSIGNED_INT to + * indicate the returned value type. + * + * To read utilization values, first determine the size of buffer required to hold the samples by invoking the function with + * \a utilizationSamples set to NULL. The function will return NVML_ERROR_INSUFFICIENT_SIZE, with the current vGPU instance + * count in \a vgpuInstanceSamplesCount, or NVML_SUCCESS if the current vGPU instance count is zero. The caller should allocate + * a buffer of size vgpuInstanceSamplesCount * sizeof(nvmlVgpuInstanceUtilizationSample_t). Invoke the function again with + * the allocated buffer passed in \a utilizationSamples, and \a vgpuInstanceSamplesCount set to the number of entries the + * buffer is sized for. + * + * On successful return, the function updates \a vgpuInstanceSampleCount with the number of vGPU utilization sample + * structures that were actually written. This may differ from a previously read value as vGPU instances are created or + * destroyed. + * + * lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 + * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set lastSeenTimeStamp + * to a timeStamp retrieved from a previous query to read utilization since the previous query. + * + * @param device The identifier for the target device + * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. + * @param sampleValType Pointer to caller-supplied buffer to hold the type of returned sample values + * @param vgpuInstanceSamplesCount Pointer to caller-supplied array size, and returns number of vGPU instances + * @param utilizationSamples Pointer to caller-supplied buffer in which vGPU utilization samples are returned + + * @return + * - \ref NVML_SUCCESS if utilization samples are successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a vgpuInstanceSamplesCount or \a sampleValType is + * NULL, or a sample count of 0 is passed with a non-NULL \a utilizationSamples + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if supplied \a vgpuInstanceSamplesCount is too small to return samples for all + * vGPU instances currently executing on the device + * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_FOUND if sample entries are not found + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuUtilization(nvmlDevice_t device, unsigned long long lastSeenTimeStamp, + nvmlValueType_t *sampleValType, unsigned int *vgpuInstanceSamplesCount, + nvmlVgpuInstanceUtilizationSample_t *utilizationSamples); + +/** + * Retrieves recent utilization for vGPU instances running on a physical GPU (device). + * + * For Kepler &tm; or newer fully supported devices. + * + * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, video decoder, jpeg decoder, and OFA for vGPU + * instances running on a device. Utilization values are returned as an array of utilization sample structures in the caller-supplied + * buffer pointed at by \a vgpuUtilInfo->vgpuUtilArray. One utilization sample structure is returned per vGPU instance, and includes the + * CPU timestamp at which the samples were recorded. Individual utilization values are returned as "unsigned int" values + * in nvmlValue_t unions. The function sets the caller-supplied \a vgpuUtilInfo->sampleValType to NVML_VALUE_TYPE_UNSIGNED_INT to + * indicate the returned value type. + * + * To read utilization values, first determine the size of buffer required to hold the samples by invoking the function with + * \a vgpuUtilInfo->vgpuUtilArray set to NULL. The function will return NVML_ERROR_INSUFFICIENT_SIZE, with the current vGPU instance + * count in \a vgpuUtilInfo->vgpuInstanceCount, or NVML_SUCCESS if the current vGPU instance count is zero. The caller should allocate + * a buffer of size vgpuUtilInfo->vgpuInstanceCount * sizeof(nvmlVgpuInstanceUtilizationInfo_t). Invoke the function again with + * the allocated buffer passed in \a vgpuUtilInfo->vgpuUtilArray, and \a vgpuUtilInfo->vgpuInstanceCount set to the number of entries the + * buffer is sized for. + * + * On successful return, the function updates \a vgpuUtilInfo->vgpuInstanceCount with the number of vGPU utilization sample + * structures that were actually written. This may differ from a previously read value as vGPU instances are created or + * destroyed. + * + * \a vgpuUtilInfo->lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 + * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set \a vgpuUtilInfo->lastSeenTimeStamp + * to a timeStamp retrieved from a previous query to read utilization since the previous query. + * + * @param device The identifier for the target device + * @param vgpuUtilInfo Pointer to the caller-provided structure of nvmlVgpuInstancesUtilizationInfo_t + + * @return + * - \ref NVML_SUCCESS If utilization samples are successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid, \a vgpuUtilInfo is NULL, or \a vgpuUtilInfo->vgpuInstanceCount is 0 + * - \ref NVML_ERROR_NOT_SUPPORTED If vGPU is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST If the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a vgpuUtilInfo is invalid + * - \ref NVML_ERROR_INSUFFICIENT_SIZE If \a vgpuUtilInfo->vgpuUtilArray is NULL, or the buffer size of vgpuUtilInfo->vgpuInstanceCount is too small. + * The caller should check the current vGPU instance count from the returned vgpuUtilInfo->vgpuInstanceCount, and call + * the function again with a buffer of size vgpuUtilInfo->vgpuInstanceCount * sizeof(nvmlVgpuInstanceUtilizationInfo_t) + * - \ref NVML_ERROR_NOT_FOUND If sample entries are not found + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuInstancesUtilizationInfo(nvmlDevice_t device, + nvmlVgpuInstancesUtilizationInfo_t *vgpuUtilInfo); + +/** + * Retrieves current utilization for processes running on vGPUs on a physical GPU (device). + * + * For Maxwell &tm; or newer fully supported devices. + * + * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, and video decoder for processes running on + * vGPU instances active on a device. Utilization values are returned as an array of utilization sample structures in the + * caller-supplied buffer pointed at by \a utilizationSamples. One utilization sample structure is returned per process running + * on vGPU instances, that had some non-zero utilization during the last sample period. It includes the CPU timestamp at which + * the samples were recorded. Individual utilization values are returned as "unsigned int" values. + * + * To read utilization values, first determine the size of buffer required to hold the samples by invoking the function with + * \a utilizationSamples set to NULL. The function will return NVML_ERROR_INSUFFICIENT_SIZE, with the current vGPU instance + * count in \a vgpuProcessSamplesCount. The caller should allocate a buffer of size + * vgpuProcessSamplesCount * sizeof(nvmlVgpuProcessUtilizationSample_t). Invoke the function again with + * the allocated buffer passed in \a utilizationSamples, and \a vgpuProcessSamplesCount set to the number of entries the + * buffer is sized for. + * + * On successful return, the function updates \a vgpuSubProcessSampleCount with the number of vGPU sub process utilization sample + * structures that were actually written. This may differ from a previously read value depending on the number of processes that are active + * in any given sample period. + * + * lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 + * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set lastSeenTimeStamp + * to a timeStamp retrieved from a previous query to read utilization since the previous query. + * + * @param device The identifier for the target device + * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. + * @param vgpuProcessSamplesCount Pointer to caller-supplied array size, and returns number of processes running on vGPU instances + * @param utilizationSamples Pointer to caller-supplied buffer in which vGPU sub process utilization samples are returned + + * @return + * - \ref NVML_SUCCESS if utilization samples are successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a vgpuProcessSamplesCount or a sample count of 0 is + * passed with a non-NULL \a utilizationSamples + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if supplied \a vgpuProcessSamplesCount is too small to return samples for all + * vGPU instances currently executing on the device + * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_FOUND if sample entries are not found + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuProcessUtilization(nvmlDevice_t device, unsigned long long lastSeenTimeStamp, + unsigned int *vgpuProcessSamplesCount, + nvmlVgpuProcessUtilizationSample_t *utilizationSamples); + +/** + * Retrieves recent utilization for processes running on vGPU instances on a physical GPU (device). + * + * For Maxwell &tm; or newer fully supported devices. + * + * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, video decoder, jpeg decoder, and OFA for processes running + * on vGPU instances active on a device. Utilization values are returned as an array of utilization sample structures in the caller-supplied + * buffer pointed at by \a vgpuProcUtilInfo->vgpuProcUtilArray. One utilization sample structure is returned per process running + * on vGPU instances, that had some non-zero utilization during the last sample period. It includes the CPU timestamp at which + * the samples were recorded. Individual utilization values are returned as "unsigned int" values. + * + * To read utilization values, first determine the size of buffer required to hold the samples by invoking the function with + * \a vgpuProcUtilInfo->vgpuProcUtilArray set to NULL. The function will return NVML_ERROR_INSUFFICIENT_SIZE, with the current processes' count + * running on vGPU instances in \a vgpuProcUtilInfo->vgpuProcessCount. The caller should allocate a buffer of size + * vgpuProcUtilInfo->vgpuProcessCount * sizeof(nvmlVgpuProcessUtilizationSample_t). Invoke the function again with the allocated buffer passed + * in \a vgpuProcUtilInfo->vgpuProcUtilArray, and \a vgpuProcUtilInfo->vgpuProcessCount set to the number of entries the buffer is sized for. + * + * On successful return, the function updates \a vgpuProcUtilInfo->vgpuProcessCount with the number of vGPU sub process utilization sample + * structures that were actually written. This may differ from a previously read value depending on the number of processes that are active + * in any given sample period. + * + * vgpuProcUtilInfo->lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 + * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set vgpuProcUtilInfo->lastSeenTimeStamp + * to a timeStamp retrieved from a previous query to read utilization since the previous query. + * + * @param device The identifier for the target device + * @param vgpuProcUtilInfo Pointer to the caller-provided structure of nvmlVgpuProcessesUtilizationInfo_t + + * @return + * - \ref NVML_SUCCESS If utilization samples are successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid, or \a vgpuProcUtilInfo is null + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the version of \a vgpuProcUtilInfo is invalid + * - \ref NVML_ERROR_INSUFFICIENT_SIZE If \a vgpuProcUtilInfo->vgpuProcUtilArray is null, or supplied \a vgpuProcUtilInfo->vgpuProcessCount + * is too small to return samples for all processes on vGPU instances currently executing on the device. + * The caller should check the current processes count from the returned \a vgpuProcUtilInfo->vgpuProcessCount, + * and call the function again with a buffer of size + * vgpuProcUtilInfo->vgpuProcessCount * sizeof(nvmlVgpuProcessUtilizationSample_t) + * - \ref NVML_ERROR_NOT_SUPPORTED If vGPU is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST If the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_FOUND If sample entries are not found + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuProcessesUtilizationInfo(nvmlDevice_t device, nvmlVgpuProcessesUtilizationInfo_t *vgpuProcUtilInfo); + +/** + * Queries the state of per process accounting mode on vGPU. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance The identifier of the target vGPU instance + * @param mode Reference in which to return the current accounting mode + * + * @return + * - \ref NVML_SUCCESS if the mode has been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a mode is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature + * - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running on the vGPU instance + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetAccountingMode(nvmlVgpuInstance_t vgpuInstance, nvmlEnableState_t *mode); + +/** + * Queries list of processes running on vGPU that can be queried for accounting stats. The list of processes + * returned can be in running or terminated state. + * + * For Maxwell &tm; or newer fully supported devices. + * + * To just query the maximum number of processes that can be queried, call this function with *count = 0 and + * pids=NULL. The return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if list is empty. + * + * For more details see \ref nvmlVgpuInstanceGetAccountingStats. + * + * @note In case of PID collision some processes might not be accessible before the circular buffer is full. + * + * @param vgpuInstance The identifier of the target vGPU instance + * @param count Reference in which to provide the \a pids array size, and + * to return the number of elements ready to be queried + * @param pids Reference in which to return list of process ids + * + * @return + * - \ref NVML_SUCCESS if pids were successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a count is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small (\a count is set to expected value) + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlVgpuInstanceGetAccountingPids + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetAccountingPids(nvmlVgpuInstance_t vgpuInstance, unsigned int *count, unsigned int *pids); + +/** + * Queries process's accounting stats. + * + * For Maxwell &tm; or newer fully supported devices. + * + * Accounting stats capture GPU utilization and other statistics across the lifetime of a process, and + * can be queried during life time of the process or after its termination. + * The time field in \ref nvmlAccountingStats_t is reported as 0 during the lifetime of the process and + * updated to actual running time after its termination. + * Accounting stats are kept in a circular buffer, newly created processes overwrite information about old + * processes. + * + * See \ref nvmlAccountingStats_t for description of each returned metric. + * List of processes that can be queried can be retrieved from \ref nvmlVgpuInstanceGetAccountingPids. + * + * @note Accounting Mode needs to be on. See \ref nvmlVgpuInstanceGetAccountingMode. + * @note Only compute and graphics applications stats can be queried. Monitoring applications stats can't be + * queried since they don't contribute to GPU utilization. + * @note In case of pid collision stats of only the latest process (that terminated last) will be reported + * + * @param vgpuInstance The identifier of the target vGPU instance + * @param pid Process Id of the target process to query stats for + * @param stats Reference in which to return the process's accounting stats + * + * @return + * - \ref NVML_SUCCESS if stats have been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a stats is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * or \a stats is not found + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetAccountingStats(nvmlVgpuInstance_t vgpuInstance, unsigned int pid, nvmlAccountingStats_t *stats); + +/** + * Clears accounting information of the vGPU instance that have already terminated. + * + * For Maxwell &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * @note Accounting Mode needs to be on. See \ref nvmlVgpuInstanceGetAccountingMode. + * @note Only compute and graphics applications stats are reported and can be cleared since monitoring applications + * stats don't contribute to GPU utilization. + * + * @param vgpuInstance The identifier of the target vGPU instance + * + * @return + * - \ref NVML_SUCCESS if accounting information has been cleared + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceClearAccountingPids(nvmlVgpuInstance_t vgpuInstance); + +/** + * Query the license information of the vGPU instance. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param licenseInfo Pointer to vGPU license information structure + * + * @return + * - \ref NVML_SUCCESS if information is successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a licenseInfo is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running on the vGPU instance + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetLicenseInfo_v2(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuLicenseInfo_t *licenseInfo); +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlExcludedGpuQueries Excluded GPU Queries + * This chapter describes NVML operations that are associated with excluded GPUs. + * @{ + */ +/***************************************************************************************************/ + +/** + * Excluded GPU device information + **/ +typedef struct nvmlExcludedDeviceInfo_st +{ + nvmlPciInfo_t pciInfo; //!< The PCI information for the excluded GPU + char uuid[NVML_DEVICE_UUID_BUFFER_SIZE]; //!< The ASCII string UUID for the excluded GPU +} nvmlExcludedDeviceInfo_t; + + /** + * Retrieves the number of excluded GPU devices in the system. + * + * For all products. + * + * @param deviceCount Reference in which to return the number of excluded devices + * + * @return + * - \ref NVML_SUCCESS if \a deviceCount has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a deviceCount is NULL + */ +nvmlReturn_t DECLDIR nvmlGetExcludedDeviceCount(unsigned int *deviceCount); + +/** + * Acquire the device information for an excluded GPU device, based on its index. + * + * For all products. + * + * Valid indices are derived from the \a deviceCount returned by + * \ref nvmlGetExcludedDeviceCount(). For example, if \a deviceCount is 2 the valid indices + * are 0 and 1, corresponding to GPU 0 and GPU 1. + * + * @param index The index of the target GPU, >= 0 and < \a deviceCount + * @param info Reference in which to return the device information + * + * @return + * - \ref NVML_SUCCESS if \a device has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a index is invalid or \a info is NULL + * + * @see nvmlGetExcludedDeviceCount + */ +nvmlReturn_t DECLDIR nvmlGetExcludedDeviceInfoByIndex(unsigned int index, nvmlExcludedDeviceInfo_t *info); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlGPUPRMAccess PRM Access + * This chapter describes NVML operations that are associated with PRM register reads + * @{ + */ +/***************************************************************************************************/ + +#define NVML_PRM_DATA_MAX_SIZE 496 +/** + * Main PRM input structure + */ +typedef struct +{ + /* I/O parameters */ + unsigned dataSize; //!< Size of the input TLV data. + unsigned status; //!< OUT: status of the PRM command + union { + /* Input data in TLV format */ + unsigned char inData[NVML_PRM_DATA_MAX_SIZE]; //!< IN: Input data in TLV format + /* Output data in TLV format */ + unsigned char outData[NVML_PRM_DATA_MAX_SIZE]; //!< OUT: Output PRM data in TLV format + }; +} nvmlPRMTLV_v1_t; + +/** + * Read or write a GPU PRM register. The input is assumed to be in TLV format in + * network byte order. + * + * %BLACKWELL_OR_NEWER% + * + * Supported on Linux only. + * + * @param device Identifer of target GPU device + * @param buffer Structure holding the input data in TLV format as well as + * the PRM register contents in TLV format (in the case of a successful + * read operation). + * Note: the input data and any returned data shall be in network byte order. + * + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if \p device or \p buffer are invalid + * - \ref NVML_ERROR_NO_PERMISSION if user does not have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if this feature is not supported by the device + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the version specified in \p buffer is not supported + */ +nvmlReturn_t DECLDIR nvmlDeviceReadWritePRM_v1(nvmlDevice_t device, nvmlPRMTLV_v1_t *buffer); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlMultiInstanceGPU Multi Instance GPU Management + * This chapter describes NVML operations that are associated with Multi Instance GPU management. + * @{ + */ +/***************************************************************************************************/ + +/** + * Disable Multi Instance GPU mode. + */ +#define NVML_DEVICE_MIG_DISABLE 0x0 + +/** + * Enable Multi Instance GPU mode. + */ +#define NVML_DEVICE_MIG_ENABLE 0x1 + +/** + * GPU instance profiles. + * + * These macros should be passed to \ref nvmlDeviceGetGpuInstanceProfileInfo to retrieve the + * detailed information about a GPU instance such as profile ID, engine counts. + */ +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE 0x0 +#define NVML_GPU_INSTANCE_PROFILE_2_SLICE 0x1 +#define NVML_GPU_INSTANCE_PROFILE_3_SLICE 0x2 +#define NVML_GPU_INSTANCE_PROFILE_4_SLICE 0x3 +#define NVML_GPU_INSTANCE_PROFILE_7_SLICE 0x4 +#define NVML_GPU_INSTANCE_PROFILE_8_SLICE 0x5 +#define NVML_GPU_INSTANCE_PROFILE_6_SLICE 0x6 +// 1_SLICE profile with at least one (if supported at all) of Decoder, Encoder, JPEG, OFA engines. +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE_REV1 0x7 +// 2_SLICE profile with at least one (if supported at all) of Decoder, Encoder, JPEG, OFA engines. +#define NVML_GPU_INSTANCE_PROFILE_2_SLICE_REV1 0x8 +// 1_SLICE profile with twice the amount of memory resources. +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE_REV2 0x9 +// 1_SLICE gfx capable profile +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE_GFX 0x0A +// 2_SLICE gfx capable profile +#define NVML_GPU_INSTANCE_PROFILE_2_SLICE_GFX 0x0B +// 4_SLICE gfx capable profile +#define NVML_GPU_INSTANCE_PROFILE_4_SLICE_GFX 0x0C +// 1_SLICE profile with none of Decode, Encoder, JPEG, OFA engines. +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE_NO_ME 0x0D +// 2_SLICE profile with none of Decode, Encoder, JPEG, OFA engines. +#define NVML_GPU_INSTANCE_PROFILE_2_SLICE_NO_ME 0x0E +// 1_SLICE profile with all of GPU Decode, Encoder, JPEG, OFA engines. +// Allocation of instance of this profile prevents allocation of +// all but _NO_ME profiles. +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE_ALL_ME 0x0F +// 2_SLICE profile with all of GPU Decode, Encoder, JPEG, OFA engines. +// Allocation of instance of this profile prevents allocation of +// all but _NO_ME profiles. +#define NVML_GPU_INSTANCE_PROFILE_2_SLICE_ALL_ME 0x10 +#define NVML_GPU_INSTANCE_PROFILE_COUNT 0x11 + +/** + * MIG GPU instance profile capability. + * + * Bit field values representing MIG profile capabilities + * \ref nvmlGpuInstanceProfileInfo_v3_t.capabilities + */ +#define NVML_GPU_INSTANCE_PROFILE_CAPS_P2P 0x1 +#define NVML_GPU_INTSTANCE_PROFILE_CAPS_P2P 0x1 //!< Deprecated, do not use +#define NVML_GPU_INSTANCE_PROFILE_CAPS_GFX 0x2 + +/** + * MIG compute instance profile capability. + * + * Bit field values representing MIG profile capabilities + * \ref nvmlComputeInstanceProfileInfo_v3_t.capabilities + */ +#define NVML_COMPUTE_INSTANCE_PROFILE_CAPS_GFX 0x1 + +typedef struct nvmlGpuInstancePlacement_st +{ + unsigned int start; //!< Index of first occupied memory slice + unsigned int size; //!< Number of memory slices occupied +} nvmlGpuInstancePlacement_t; + +/** + * GPU instance profile information. + */ +typedef struct nvmlGpuInstanceProfileInfo_st +{ + unsigned int id; //!< Unique profile ID within the device + unsigned int isP2pSupported; //!< Peer-to-Peer support + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< GPU instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int copyEngineCount; //!< Copy Engine count + unsigned int decoderCount; //!< Decoder Engine count + unsigned int encoderCount; //!< Encoder Engine count + unsigned int jpegCount; //!< JPEG Engine count + unsigned int ofaCount; //!< OFA Engine count + unsigned long long memorySizeMB; //!< Memory size in MBytes +} nvmlGpuInstanceProfileInfo_t; + +/** + * GPU instance profile information (v2). + * + * Version 2 adds the \ref nvmlGpuInstanceProfileInfo_v2_t.version field + * to the start of the structure, and the \ref nvmlGpuInstanceProfileInfo_v2_t.name + * field to the end. This structure is not backwards-compatible with + * \ref nvmlGpuInstanceProfileInfo_t. + */ +typedef struct nvmlGpuInstanceProfileInfo_v2_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlGpuInstanceProfileInfo_v2) + unsigned int id; //!< Unique profile ID within the device + unsigned int isP2pSupported; //!< Peer-to-Peer support + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< GPU instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int copyEngineCount; //!< Copy Engine count + unsigned int decoderCount; //!< Decoder Engine count + unsigned int encoderCount; //!< Encoder Engine count + unsigned int jpegCount; //!< JPEG Engine count + unsigned int ofaCount; //!< OFA Engine count + unsigned long long memorySizeMB; //!< Memory size in MBytes + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name +} nvmlGpuInstanceProfileInfo_v2_t; + +/** + * Version identifier value for \ref nvmlGpuInstanceProfileInfo_v2_t.version. + */ +#define nvmlGpuInstanceProfileInfo_v2 NVML_STRUCT_VERSION(GpuInstanceProfileInfo, 2) + +/** + * GPU instance profile information (v3). + * + * Version 3 removes isP2pSupported field and adds the \ref nvmlGpuInstanceProfileInfo_v3_t.capabilities + * field \ref nvmlGpuInstanceProfileInfo_t. + */ +typedef struct nvmlGpuInstanceProfileInfo_v3_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlGpuInstanceProfileInfo_v3) + unsigned int id; //!< Unique profile ID within the device + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< GPU instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int copyEngineCount; //!< Copy Engine count + unsigned int decoderCount; //!< Decoder Engine count + unsigned int encoderCount; //!< Encoder Engine count + unsigned int jpegCount; //!< JPEG Engine count + unsigned int ofaCount; //!< OFA Engine count + unsigned long long memorySizeMB; //!< Memory size in MBytes + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name + unsigned int capabilities; //!< Additional capabilities +} nvmlGpuInstanceProfileInfo_v3_t; + +/** + * Version identifier value for \ref nvmlGpuInstanceProfileInfo_v3_t.version. + */ +#define nvmlGpuInstanceProfileInfo_v3 NVML_STRUCT_VERSION(GpuInstanceProfileInfo, 3) + +typedef struct nvmlGpuInstanceInfo_st +{ + nvmlDevice_t device; //!< Parent device + unsigned int id; //!< Unique instance ID within the device + unsigned int profileId; //!< Unique profile ID within the device + nvmlGpuInstancePlacement_t placement; //!< Placement for this instance +} nvmlGpuInstanceInfo_t; + +/** + * Compute instance profiles. + * + * These macros should be passed to \ref nvmlGpuInstanceGetComputeInstanceProfileInfo to retrieve the + * detailed information about a compute instance such as profile ID, engine counts + */ +#define NVML_COMPUTE_INSTANCE_PROFILE_1_SLICE 0x0 +#define NVML_COMPUTE_INSTANCE_PROFILE_2_SLICE 0x1 +#define NVML_COMPUTE_INSTANCE_PROFILE_3_SLICE 0x2 +#define NVML_COMPUTE_INSTANCE_PROFILE_4_SLICE 0x3 +#define NVML_COMPUTE_INSTANCE_PROFILE_7_SLICE 0x4 +#define NVML_COMPUTE_INSTANCE_PROFILE_8_SLICE 0x5 +#define NVML_COMPUTE_INSTANCE_PROFILE_6_SLICE 0x6 +#define NVML_COMPUTE_INSTANCE_PROFILE_1_SLICE_REV1 0x7 +#define NVML_COMPUTE_INSTANCE_PROFILE_COUNT 0x8 + +#define NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_SHARED 0x0 //!< All the engines except multiprocessors would be shared +#define NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_COUNT 0x1 + +typedef struct nvmlComputeInstancePlacement_st +{ + unsigned int start; //!< Index of first occupied compute slice + unsigned int size; //!< Number of compute slices occupied +} nvmlComputeInstancePlacement_t; + +/** + * Compute instance profile information. + */ +typedef struct nvmlComputeInstanceProfileInfo_st +{ + unsigned int id; //!< Unique profile ID within the GPU instance + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< Compute instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count +} nvmlComputeInstanceProfileInfo_t; + +/** + * Compute instance profile information (v2). + * + * Version 2 adds the \ref nvmlComputeInstanceProfileInfo_v2_t.version field + * to the start of the structure, and the \ref nvmlComputeInstanceProfileInfo_v2_t.name + * field to the end. This structure is not backwards-compatible with + * \ref nvmlComputeInstanceProfileInfo_t. + */ +typedef struct nvmlComputeInstanceProfileInfo_v2_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlComputeInstanceProfileInfo_v2) + unsigned int id; //!< Unique profile ID within the GPU instance + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< Compute instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name +} nvmlComputeInstanceProfileInfo_v2_t; + +/** + * Version identifier value for \ref nvmlComputeInstanceProfileInfo_v2_t.version. + */ +#define nvmlComputeInstanceProfileInfo_v2 NVML_STRUCT_VERSION(ComputeInstanceProfileInfo, 2) + +/** + * Compute instance profile information (v3). + * + * Version 3 adds the \ref nvmlComputeInstanceProfileInfo_v3_t.capabilities field + * \ref nvmlComputeInstanceProfileInfo_t. + */ +typedef struct nvmlComputeInstanceProfileInfo_v3_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlComputeInstanceProfileInfo_v3) + unsigned int id; //!< Unique profile ID within the GPU instance + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< Compute instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name + unsigned int capabilities; //!< Additional capabilities +} nvmlComputeInstanceProfileInfo_v3_t; + +/** + * Version identifier value for \ref nvmlComputeInstanceProfileInfo_v3_t.version. + */ +#define nvmlComputeInstanceProfileInfo_v3 NVML_STRUCT_VERSION(ComputeInstanceProfileInfo, 3) + +typedef struct nvmlComputeInstanceInfo_st +{ + nvmlDevice_t device; //!< Parent device + nvmlGpuInstance_t gpuInstance; //!< Parent GPU instance + unsigned int id; //!< Unique instance ID within the GPU instance + unsigned int profileId; //!< Unique profile ID within the GPU instance + nvmlComputeInstancePlacement_t placement; //!< Placement for this instance within the GPU instance's compute slice range {0, sliceCount} +} nvmlComputeInstanceInfo_t; + +typedef struct +{ + struct nvmlComputeInstance_st* handle; +} nvmlComputeInstance_t; + +/** + * Set MIG mode for the device. + * + * For Ampere &tm; or newer fully supported devices. + * Requires root user. + * + * This mode determines whether a GPU instance can be created. + * + * This API may unbind or reset the device to activate the requested mode. Thus, the attributes associated with the + * device, such as minor number, might change. The caller of this API is expected to query such attributes again. + * + * On certain platforms like pass-through virtualization, where reset functionality may not be exposed directly, VM + * reboot is required. \a activationStatus would return \ref NVML_ERROR_RESET_REQUIRED for such cases. + * + * \a activationStatus would return the appropriate error code upon unsuccessful activation. For example, if device + * unbind fails because the device isn't idle, \ref NVML_ERROR_IN_USE would be returned. The caller of this API + * is expected to idle the device and retry setting the \a mode. + * + * @note On Windows, only disabling MIG mode is supported. \a activationStatus would return \ref + * NVML_ERROR_NOT_SUPPORTED as GPU reset is not supported on Windows through this API. + * + * @param device The identifier of the target device + * @param mode The mode to be set, \ref NVML_DEVICE_MIG_DISABLE or + * \ref NVML_DEVICE_MIG_ENABLE + * @param activationStatus The activationStatus status + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device,\a mode or \a activationStatus are invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support MIG mode + */ +nvmlReturn_t DECLDIR nvmlDeviceSetMigMode(nvmlDevice_t device, unsigned int mode, nvmlReturn_t *activationStatus); + +/** + * Get MIG mode for the device. + * + * For Ampere &tm; or newer fully supported devices. + * + * Changing MIG modes may require device unbind or reset. The "pending" MIG mode refers to the target mode following the + * next activation trigger. + * + * @param device The identifier of the target device + * @param currentMode Returns the current mode, \ref NVML_DEVICE_MIG_DISABLE or + * \ref NVML_DEVICE_MIG_ENABLE + * @param pendingMode Returns the pending mode, \ref NVML_DEVICE_MIG_DISABLE or + * \ref NVML_DEVICE_MIG_ENABLE + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a currentMode or \a pendingMode are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support MIG mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMigMode(nvmlDevice_t device, unsigned int *currentMode, unsigned int *pendingMode); + +/** + * Get GPU instance profile information + * + * Information provided by this API is immutable throughout the lifetime of a MIG mode. + * + * @note This API can be used to enumerate all MIG profiles supported by NVML in a forward compatible + * way by invoking it on \a profile values starting from 0, until the API returns \ref NVML_ERROR_INVALID_ARGUMENT. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param profile One of the NVML_GPU_INSTANCE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile or \a info are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support MIG or \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceProfileInfo(nvmlDevice_t device, unsigned int profile, + nvmlGpuInstanceProfileInfo_t *info); + +/** + * Versioned wrapper around \ref nvmlDeviceGetGpuInstanceProfileInfo that accepts a versioned + * \ref nvmlGpuInstanceProfileInfo_v2_t or later output structure. + * + * @note The caller must set the \ref nvmlGpuInstanceProfileInfo_v2_t.version field to the + * appropriate version prior to calling this function. For example: + * \code + * nvmlGpuInstanceProfileInfo_v2_t profileInfo = + * { .version = nvmlGpuInstanceProfileInfo_v2 }; + * nvmlReturn_t result = nvmlDeviceGetGpuInstanceProfileInfoV(device, + * profile, + * &profileInfo); + * \endcode + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param profile One of the NVML_GPU_INSTANCE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile, \a info, or \a info->version are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceProfileInfoV(nvmlDevice_t device, unsigned int profile, + nvmlGpuInstanceProfileInfo_v2_t *info); + +/** + * GPU instance profile query function that accepts profile ID, instead of profile name. + * It accepts a versioned \ref nvmlGpuInstanceProfileInfo_v2_t or later output structure. + * + * @note The caller must set the \ref nvmlGpuInstanceProfileInfo_v2_t.version field to the + * appropriate version prior to calling this function. For example: + * \code + * nvmlGpuInstanceProfileInfo_v2_t profileInfo = + * { .version = nvmlGpuInstanceProfileInfo_v2 }; + * nvmlReturn_t result = nvmlDeviceGetGpuInstanceProfileInfoV(device, + * profile, + * &profileInfo); + * \endcode + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param profileId One of the profile IDs. + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profileId, \a info, or \a info->version are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceProfileInfoByIdV(nvmlDevice_t device, unsigned int profileId, + nvmlGpuInstanceProfileInfo_v2_t *info); + +/** + * Get GPU instance placements. + * + * A placement represents the location of a GPU instance within a device. This API only returns all the possible + * placements for the given profile regardless of whether MIG is enabled or not. + * A created GPU instance occupies memory slices described by its placement. Creation of new GPU instance will + * fail if there is overlap with the already occupied memory slices. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param placements Returns placements allowed for the profile. Can be NULL to discover number + * of allowed placements for this profile. If non-NULL must be large enough + * to accommodate the placements supported by the profile. + * @param count Returns number of allowed placemenets for the profile. + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profileId or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support MIG or \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstancePossiblePlacements_v2(nvmlDevice_t device, unsigned int profileId, + nvmlGpuInstancePlacement_t *placements, + unsigned int *count); + +/** + * Get GPU instance profile capacity. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param count Returns remaining instance count for the profile ID + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profileId or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceRemainingCapacity(nvmlDevice_t device, unsigned int profileId, + unsigned int *count); + +/** + * Create GPU instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the GPU instance is destroyed explicitly, the GPU instance handle would + * become invalid. The GPU instance must be recreated to acquire a valid handle. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param gpuInstance Returns the GPU instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile, \a profileId or \a gpuInstance are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or in vGPU guest + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested GPU instance could not be created + */ +nvmlReturn_t DECLDIR nvmlDeviceCreateGpuInstance(nvmlDevice_t device, unsigned int profileId, + nvmlGpuInstance_t *gpuInstance); + +/** + * Create GPU instance with the specified placement. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the GPU instance is destroyed explicitly, the GPU instance handle would + * become invalid. The GPU instance must be recreated to acquire a valid handle. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param placement The requested placement. See \ref nvmlDeviceGetGpuInstancePossiblePlacements_v2 + * @param gpuInstance Returns the GPU instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile, \a profileId, \a placement or \a gpuInstance + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or in vGPU guest + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested GPU instance could not be created + */ +nvmlReturn_t DECLDIR nvmlDeviceCreateGpuInstanceWithPlacement(nvmlDevice_t device, unsigned int profileId, + const nvmlGpuInstancePlacement_t *placement, + nvmlGpuInstance_t *gpuInstance); +/** + * Destroy GPU instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The GPU instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or in vGPU guest + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_IN_USE If the GPU instance is in use. This error would be returned if processes + * (e.g. CUDA application) or compute instances are active on the + * GPU instance. + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceDestroy(nvmlGpuInstance_t gpuInstance); + +/** + * Get GPU instances for given profile ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param gpuInstances Returns pre-exiting GPU instances, the buffer must be large enough to + * accommodate the instances supported by the profile. + * See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param count The count of returned GPU instances + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profileId, \a gpuInstances or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstances(nvmlDevice_t device, unsigned int profileId, + nvmlGpuInstance_t *gpuInstances, unsigned int *count); + +/** + * Get GPU instances for given instance ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param id The GPU instance ID + * @param gpuInstance Returns GPU instance + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a id or \a gpuInstance are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_NOT_FOUND If the GPU instance is not found. + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceById(nvmlDevice_t device, unsigned int id, nvmlGpuInstance_t *gpuInstance); + +/** + * Get GPU instance information. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param gpuInstance The GPU instance handle + * @param info Return GPU instance information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance or \a info are invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetInfo(nvmlGpuInstance_t gpuInstance, nvmlGpuInstanceInfo_t *info); + +/** + * Get compute instance profile information. + * + * Information provided by this API is immutable throughout the lifetime of a MIG mode. + * + * @note This API can be used to enumerate all MIG profiles supported by NVML in a forward compatible + * way by invoking it on \a profile values starting from 0, until the API returns \ref NVML_ERROR_INVALID_ARGUMENT. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profile One of the NVML_COMPUTE_INSTANCE_PROFILE_* + * @param engProfile One of the NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a engProfile or \a info are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceProfileInfo(nvmlGpuInstance_t gpuInstance, unsigned int profile, + unsigned int engProfile, + nvmlComputeInstanceProfileInfo_t *info); + +/** + * Versioned wrapper around \ref nvmlGpuInstanceGetComputeInstanceProfileInfo that accepts a versioned + * \ref nvmlComputeInstanceProfileInfo_v2_t or later output structure. + * + * @note The caller must set the \ref nvmlGpuInstanceProfileInfo_v2_t.version field to the + * appropriate version prior to calling this function. For example: + * \code + * nvmlComputeInstanceProfileInfo_v2_t profileInfo = + * { .version = nvmlComputeInstanceProfileInfo_v2 }; + * nvmlReturn_t result = nvmlGpuInstanceGetComputeInstanceProfileInfoV(gpuInstance, + * profile, + * engProfile, + * &profileInfo); + * \endcode + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profile One of the NVML_COMPUTE_INSTANCE_PROFILE_* + * @param engProfile One of the NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a engProfile, \a info, or \a info->version are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceProfileInfoV(nvmlGpuInstance_t gpuInstance, unsigned int profile, + unsigned int engProfile, + nvmlComputeInstanceProfileInfo_v2_t *info); + +/** + * Get compute instance profile capacity. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param count Returns remaining instance count for the profile ID + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profileId or \a availableCount are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceRemainingCapacity(nvmlGpuInstance_t gpuInstance, + unsigned int profileId, unsigned int *count); + +/** + * Get compute instance placements. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * A placement represents the location of a compute instance within a GPU instance. This API only returns all the possible + * placements for the given profile. + * A created compute instance occupies compute slices described by its placement. Creation of new compute instance will + * fail if there is overlap with the already occupied compute slices. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param placements Returns placements allowed for the profile. Can be NULL to discover number + * of allowed placements for this profile. If non-NULL must be large enough + * to accommodate the placements supported by the profile. + * @param count Returns number of allowed placemenets for the profile. + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profileId or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstancePossiblePlacements(nvmlGpuInstance_t gpuInstance, + unsigned int profileId, + nvmlComputeInstancePlacement_t *placements, + unsigned int *count); + +/** + * Create compute instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the parent GPU instance is destroyed or the compute instance is destroyed + * explicitly, the compute instance handle would become invalid. The compute instance must be recreated to acquire + * a valid handle. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param computeInstance Returns the compute instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a profileId or \a computeInstance + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested compute instance could not be created + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceCreateComputeInstance(nvmlGpuInstance_t gpuInstance, unsigned int profileId, + nvmlComputeInstance_t *computeInstance); + +/** + * Create compute instance with the specified placement. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the parent GPU instance is destroyed or the compute instance is destroyed + * explicitly, the compute instance handle would become invalid. The compute instance must be recreated to acquire + * a valid handle. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param placement The requested placement. See \ref nvmlGpuInstanceGetComputeInstancePossiblePlacements + * @param computeInstance Returns the compute instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a profileId or \a computeInstance + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested compute instance could not be created + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceCreateComputeInstanceWithPlacement(nvmlGpuInstance_t gpuInstance, unsigned int profileId, + const nvmlComputeInstancePlacement_t *placement, + nvmlComputeInstance_t *computeInstance); + +/** + * Destroy compute instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param computeInstance The compute instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a computeInstance is invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_IN_USE If the compute instance is in use. This error would be returned if + * processes (e.g. CUDA application) are active on the compute instance. + */ +nvmlReturn_t DECLDIR nvmlComputeInstanceDestroy(nvmlComputeInstance_t computeInstance); + +/** + * Get compute instances for given profile ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param computeInstances Returns pre-exiting compute instances, the buffer must be large enough to + * accommodate the instances supported by the profile. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param count The count of returned compute instances + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profileId, \a computeInstances or \a count + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstances(nvmlGpuInstance_t gpuInstance, unsigned int profileId, + nvmlComputeInstance_t *computeInstances, unsigned int *count); + +/** + * Get compute instance for given instance ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The identifier of the target GPU instance + * @param id The compute instance ID + * @param computeInstance Returns compute instance + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a ID or \a computeInstance are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_NOT_FOUND If the compute instance is not found. + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceById(nvmlGpuInstance_t gpuInstance, unsigned int id, + nvmlComputeInstance_t *computeInstance); + +/** + * Get compute instance information. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param computeInstance The compute instance handle + * @param info Return compute instance information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a computeInstance or \a info are invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlComputeInstanceGetInfo_v2(nvmlComputeInstance_t computeInstance, nvmlComputeInstanceInfo_t *info); + +/** + * Test if the given handle refers to a MIG device. + * + * A MIG device handle is an NVML abstraction which maps to a MIG compute instance. + * These overloaded references can be used (with some restrictions) interchangeably + * with a GPU device handle to execute queries at a per-compute instance granularity. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device NVML handle to test + * @param isMigDevice True when handle refers to a MIG device + * + * @return + * - \ref NVML_SUCCESS if \a device status was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device handle or \a isMigDevice reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this check is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceIsMigDeviceHandle(nvmlDevice_t device, unsigned int *isMigDevice); + +/** + * Get GPU instance ID for the given MIG device handle. + * + * GPU instance IDs are unique per device and remain valid until the GPU instance is destroyed. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Target MIG device handle + * @param id GPU instance ID + * + * @return + * - \ref NVML_SUCCESS if instance ID was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a id reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceId(nvmlDevice_t device, unsigned int *id); + +/** + * Get compute instance ID for the given MIG device handle. + * + * Compute instance IDs are unique per GPU instance and remain valid until the compute instance + * is destroyed. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Target MIG device handle + * @param id Compute instance ID + * + * @return + * - \ref NVML_SUCCESS if instance ID was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a id reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetComputeInstanceId(nvmlDevice_t device, unsigned int *id); + +/** + * Get the maximum number of MIG devices that can exist under a given parent NVML device. + * + * Returns zero if MIG is not supported or enabled. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Target device handle + * @param count Count of MIG devices + * + * @return + * - \ref NVML_SUCCESS if \a count was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a count reference is invalid + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMaxMigDeviceCount(nvmlDevice_t device, unsigned int *count); + +/** + * Get MIG device handle for the given index under its parent NVML device. + * + * If the compute instance is destroyed either explicitly or by destroying, + * resetting or unbinding the parent GPU instance or the GPU device itself + * the MIG device handle would remain invalid and must be requested again + * using this API. Handles may be reused and their properties can change in + * the process. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Reference to the parent GPU device handle + * @param index Index of the MIG device + * @param migDevice Reference to the MIG device handle + * + * @return + * - \ref NVML_SUCCESS if \a migDevice handle was successfully created + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a index or \a migDevice reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_NOT_FOUND if no valid MIG device was found at \a index + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMigDeviceHandleByIndex(nvmlDevice_t device, unsigned int index, + nvmlDevice_t *migDevice); + +/** + * Get parent device handle from a MIG device handle. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param migDevice MIG device handle + * @param device Device handle + * + * @return + * - \ref NVML_SUCCESS if \a device handle was successfully created + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a migDevice or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDeviceHandleFromMigDeviceHandle(nvmlDevice_t migDevice, nvmlDevice_t *device); + +/** @} */ // @defgroup nvmlMultiInstanceGPU + + +/***************************************************************************************************/ +/** @defgroup GPM NVML GPM + * @{ + */ +/***************************************************************************************************/ +/** @defgroup nvmlGpmEnums GPM Enums + * @{ + */ +/***************************************************************************************************/ + +/** + * GPM Metric Identifiers + */ +typedef enum +{ + NVML_GPM_METRIC_GRAPHICS_UTIL = 1, //!< Percentage of time any compute/graphics app was active on the GPU. 0.0 - 100.0 + NVML_GPM_METRIC_SM_UTIL = 2, //!< Percentage of SMs that were busy. 0.0 - 100.0 + NVML_GPM_METRIC_SM_OCCUPANCY = 3, //!< Percentage of warps that were active vs theoretical maximum. 0.0 - 100.0 + NVML_GPM_METRIC_INTEGER_UTIL = 4, //!< Percentage of time the GPU's SMs were doing integer operations. 0.0 - 100.0 + NVML_GPM_METRIC_ANY_TENSOR_UTIL = 5, //!< Percentage of time the GPU's SMs were doing ANY tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_DFMA_TENSOR_UTIL = 6, //!< Percentage of time the GPU's SMs were doing DFMA tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_HMMA_TENSOR_UTIL = 7, //!< Percentage of time the GPU's SMs were doing HMMA tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_IMMA_TENSOR_UTIL = 9, //!< Percentage of time the GPU's SMs were doing IMMA tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_DRAM_BW_UTIL = 10, //!< Percentage of DRAM bw used vs theoretical maximum. 0.0 - 100.0 */ + NVML_GPM_METRIC_FP64_UTIL = 11, //!< Percentage of time the GPU's SMs were doing non-tensor FP64 math. 0.0 - 100.0 + NVML_GPM_METRIC_FP32_UTIL = 12, //!< Percentage of time the GPU's SMs were doing non-tensor FP32 math. 0.0 - 100.0 + NVML_GPM_METRIC_FP16_UTIL = 13, //!< Percentage of time the GPU's SMs were doing non-tensor FP16 math. 0.0 - 100.0 + NVML_GPM_METRIC_PCIE_TX_PER_SEC = 20, //!< PCIe traffic from this GPU in MiB/sec + NVML_GPM_METRIC_PCIE_RX_PER_SEC = 21, //!< PCIe traffic to this GPU in MiB/sec + NVML_GPM_METRIC_NVDEC_0_UTIL = 30, //!< Percent utilization of NVDEC 0. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_1_UTIL = 31, //!< Percent utilization of NVDEC 1. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_2_UTIL = 32, //!< Percent utilization of NVDEC 2. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_3_UTIL = 33, //!< Percent utilization of NVDEC 3. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_4_UTIL = 34, //!< Percent utilization of NVDEC 4. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_5_UTIL = 35, //!< Percent utilization of NVDEC 5. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_6_UTIL = 36, //!< Percent utilization of NVDEC 6. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_7_UTIL = 37, //!< Percent utilization of NVDEC 7. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_0_UTIL = 40, //!< Percent utilization of NVJPG 0. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_1_UTIL = 41, //!< Percent utilization of NVJPG 1. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_2_UTIL = 42, //!< Percent utilization of NVJPG 2. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_3_UTIL = 43, //!< Percent utilization of NVJPG 3. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_4_UTIL = 44, //!< Percent utilization of NVJPG 4. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_5_UTIL = 45, //!< Percent utilization of NVJPG 5. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_6_UTIL = 46, //!< Percent utilization of NVJPG 6. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_7_UTIL = 47, //!< Percent utilization of NVJPG 7. 0.0 - 100.0 + NVML_GPM_METRIC_NVOFA_0_UTIL = 50, //!< Percent utilization of NVOFA 0. 0.0 - 100.0 + NVML_GPM_METRIC_NVOFA_1_UTIL = 51, //!< Percent utilization of NVOFA 1. 0.0 - 100.0 + NVML_GPM_METRIC_NVLINK_TOTAL_RX_PER_SEC = 60, //!< NvLink read bandwidth for all links in MiB/sec + NVML_GPM_METRIC_NVLINK_TOTAL_TX_PER_SEC = 61, //!< NvLink write bandwidth for all links in MiB/sec + NVML_GPM_METRIC_NVLINK_L0_RX_PER_SEC = 62, //!< NvLink read bandwidth for link 0 in MiB/sec + NVML_GPM_METRIC_NVLINK_L0_TX_PER_SEC = 63, //!< NvLink write bandwidth for link 0 in MiB/sec + NVML_GPM_METRIC_NVLINK_L1_RX_PER_SEC = 64, //!< NvLink read bandwidth for link 1 in MiB/sec + NVML_GPM_METRIC_NVLINK_L1_TX_PER_SEC = 65, //!< NvLink write bandwidth for link 1 in MiB/sec + NVML_GPM_METRIC_NVLINK_L2_RX_PER_SEC = 66, //!< NvLink read bandwidth for link 2 in MiB/sec + NVML_GPM_METRIC_NVLINK_L2_TX_PER_SEC = 67, //!< NvLink write bandwidth for link 2 in MiB/sec + NVML_GPM_METRIC_NVLINK_L3_RX_PER_SEC = 68, //!< NvLink read bandwidth for link 3 in MiB/sec + NVML_GPM_METRIC_NVLINK_L3_TX_PER_SEC = 69, //!< NvLink write bandwidth for link 3 in MiB/sec + NVML_GPM_METRIC_NVLINK_L4_RX_PER_SEC = 70, //!< NvLink read bandwidth for link 4 in MiB/sec + NVML_GPM_METRIC_NVLINK_L4_TX_PER_SEC = 71, //!< NvLink write bandwidth for link 4 in MiB/sec + NVML_GPM_METRIC_NVLINK_L5_RX_PER_SEC = 72, //!< NvLink read bandwidth for link 5 in MiB/sec + NVML_GPM_METRIC_NVLINK_L5_TX_PER_SEC = 73, //!< NvLink write bandwidth for link 5 in MiB/sec + NVML_GPM_METRIC_NVLINK_L6_RX_PER_SEC = 74, //!< NvLink read bandwidth for link 6 in MiB/sec + NVML_GPM_METRIC_NVLINK_L6_TX_PER_SEC = 75, //!< NvLink write bandwidth for link 6 in MiB/sec + NVML_GPM_METRIC_NVLINK_L7_RX_PER_SEC = 76, //!< NvLink read bandwidth for link 7 in MiB/sec + NVML_GPM_METRIC_NVLINK_L7_TX_PER_SEC = 77, //!< NvLink write bandwidth for link 7 in MiB/sec + NVML_GPM_METRIC_NVLINK_L8_RX_PER_SEC = 78, //!< NvLink read bandwidth for link 8 in MiB/sec + NVML_GPM_METRIC_NVLINK_L8_TX_PER_SEC = 79, //!< NvLink write bandwidth for link 8 in MiB/sec + NVML_GPM_METRIC_NVLINK_L9_RX_PER_SEC = 80, //!< NvLink read bandwidth for link 9 in MiB/sec + NVML_GPM_METRIC_NVLINK_L9_TX_PER_SEC = 81, //!< NvLink write bandwidth for link 9 in MiB/sec + NVML_GPM_METRIC_NVLINK_L10_RX_PER_SEC = 82, //!< NvLink read bandwidth for link 10 in MiB/sec + NVML_GPM_METRIC_NVLINK_L10_TX_PER_SEC = 83, //!< NvLink write bandwidth for link 10 in MiB/sec + NVML_GPM_METRIC_NVLINK_L11_RX_PER_SEC = 84, //!< NvLink read bandwidth for link 11 in MiB/sec + NVML_GPM_METRIC_NVLINK_L11_TX_PER_SEC = 85, //!< NvLink write bandwidth for link 11 in MiB/sec + NVML_GPM_METRIC_NVLINK_L12_RX_PER_SEC = 86, //!< NvLink read bandwidth for link 12 in MiB/sec + NVML_GPM_METRIC_NVLINK_L12_TX_PER_SEC = 87, //!< NvLink write bandwidth for link 12 in MiB/sec + NVML_GPM_METRIC_NVLINK_L13_RX_PER_SEC = 88, //!< NvLink read bandwidth for link 13 in MiB/sec + NVML_GPM_METRIC_NVLINK_L13_TX_PER_SEC = 89, //!< NvLink write bandwidth for link 13 in MiB/sec + NVML_GPM_METRIC_NVLINK_L14_RX_PER_SEC = 90, //!< NvLink read bandwidth for link 14 in MiB/sec + NVML_GPM_METRIC_NVLINK_L14_TX_PER_SEC = 91, //!< NvLink write bandwidth for link 14 in MiB/sec + NVML_GPM_METRIC_NVLINK_L15_RX_PER_SEC = 92, //!< NvLink read bandwidth for link 15 in MiB/sec + NVML_GPM_METRIC_NVLINK_L15_TX_PER_SEC = 93, //!< NvLink write bandwidth for link 15 in MiB/sec + NVML_GPM_METRIC_NVLINK_L16_RX_PER_SEC = 94, //!< NvLink read bandwidth for link 16 in MiB/sec + NVML_GPM_METRIC_NVLINK_L16_TX_PER_SEC = 95, //!< NvLink write bandwidth for link 16 in MiB/sec + NVML_GPM_METRIC_NVLINK_L17_RX_PER_SEC = 96, //!< NvLink read bandwidth for link 17 in MiB/sec + NVML_GPM_METRIC_NVLINK_L17_TX_PER_SEC = 97, //!< NvLink write bandwidth for link 17 in MiB/sec + //Put new metrics for BLACKWELL here... + NVML_GPM_METRIC_C2C_TOTAL_TX_PER_SEC = 100, + NVML_GPM_METRIC_C2C_TOTAL_RX_PER_SEC = 101, + NVML_GPM_METRIC_C2C_DATA_TX_PER_SEC = 102, + NVML_GPM_METRIC_C2C_DATA_RX_PER_SEC = 103, + NVML_GPM_METRIC_C2C_LINK0_TOTAL_TX_PER_SEC = 104, + NVML_GPM_METRIC_C2C_LINK0_TOTAL_RX_PER_SEC = 105, + NVML_GPM_METRIC_C2C_LINK0_DATA_TX_PER_SEC = 106, + NVML_GPM_METRIC_C2C_LINK0_DATA_RX_PER_SEC = 107, + NVML_GPM_METRIC_C2C_LINK1_TOTAL_TX_PER_SEC = 108, + NVML_GPM_METRIC_C2C_LINK1_TOTAL_RX_PER_SEC = 109, + NVML_GPM_METRIC_C2C_LINK1_DATA_TX_PER_SEC = 110, + NVML_GPM_METRIC_C2C_LINK1_DATA_RX_PER_SEC = 111, + NVML_GPM_METRIC_C2C_LINK2_TOTAL_TX_PER_SEC = 112, + NVML_GPM_METRIC_C2C_LINK2_TOTAL_RX_PER_SEC = 113, + NVML_GPM_METRIC_C2C_LINK2_DATA_TX_PER_SEC = 114, + NVML_GPM_METRIC_C2C_LINK2_DATA_RX_PER_SEC = 115, + NVML_GPM_METRIC_C2C_LINK3_TOTAL_TX_PER_SEC = 116, + NVML_GPM_METRIC_C2C_LINK3_TOTAL_RX_PER_SEC = 117, + NVML_GPM_METRIC_C2C_LINK3_DATA_TX_PER_SEC = 118, + NVML_GPM_METRIC_C2C_LINK3_DATA_RX_PER_SEC = 119, + NVML_GPM_METRIC_C2C_LINK4_TOTAL_TX_PER_SEC = 120, + NVML_GPM_METRIC_C2C_LINK4_TOTAL_RX_PER_SEC = 121, + NVML_GPM_METRIC_C2C_LINK4_DATA_TX_PER_SEC = 122, + NVML_GPM_METRIC_C2C_LINK4_DATA_RX_PER_SEC = 123, + NVML_GPM_METRIC_C2C_LINK5_TOTAL_TX_PER_SEC = 124, + NVML_GPM_METRIC_C2C_LINK5_TOTAL_RX_PER_SEC = 125, + NVML_GPM_METRIC_C2C_LINK5_DATA_TX_PER_SEC = 126, + NVML_GPM_METRIC_C2C_LINK5_DATA_RX_PER_SEC = 127, + NVML_GPM_METRIC_C2C_LINK6_TOTAL_TX_PER_SEC = 128, + NVML_GPM_METRIC_C2C_LINK6_TOTAL_RX_PER_SEC = 129, + NVML_GPM_METRIC_C2C_LINK6_DATA_TX_PER_SEC = 130, + NVML_GPM_METRIC_C2C_LINK6_DATA_RX_PER_SEC = 131, + NVML_GPM_METRIC_C2C_LINK7_TOTAL_TX_PER_SEC = 132, + NVML_GPM_METRIC_C2C_LINK7_TOTAL_RX_PER_SEC = 133, + NVML_GPM_METRIC_C2C_LINK7_DATA_TX_PER_SEC = 134, + NVML_GPM_METRIC_C2C_LINK7_DATA_RX_PER_SEC = 135, + NVML_GPM_METRIC_C2C_LINK8_TOTAL_TX_PER_SEC = 136, + NVML_GPM_METRIC_C2C_LINK8_TOTAL_RX_PER_SEC = 137, + NVML_GPM_METRIC_C2C_LINK8_DATA_TX_PER_SEC = 138, + NVML_GPM_METRIC_C2C_LINK8_DATA_RX_PER_SEC = 139, + NVML_GPM_METRIC_C2C_LINK9_TOTAL_TX_PER_SEC = 140, + NVML_GPM_METRIC_C2C_LINK9_TOTAL_RX_PER_SEC = 141, + NVML_GPM_METRIC_C2C_LINK9_DATA_TX_PER_SEC = 142, + NVML_GPM_METRIC_C2C_LINK9_DATA_RX_PER_SEC = 143, + NVML_GPM_METRIC_C2C_LINK10_TOTAL_TX_PER_SEC = 144, + NVML_GPM_METRIC_C2C_LINK10_TOTAL_RX_PER_SEC = 145, + NVML_GPM_METRIC_C2C_LINK10_DATA_TX_PER_SEC = 146, + NVML_GPM_METRIC_C2C_LINK10_DATA_RX_PER_SEC = 147, + NVML_GPM_METRIC_C2C_LINK11_TOTAL_TX_PER_SEC = 148, + NVML_GPM_METRIC_C2C_LINK11_TOTAL_RX_PER_SEC = 149, + NVML_GPM_METRIC_C2C_LINK11_DATA_TX_PER_SEC = 150, + NVML_GPM_METRIC_C2C_LINK11_DATA_RX_PER_SEC = 151, + NVML_GPM_METRIC_C2C_LINK12_TOTAL_TX_PER_SEC = 152, + NVML_GPM_METRIC_C2C_LINK12_TOTAL_RX_PER_SEC = 153, + NVML_GPM_METRIC_C2C_LINK12_DATA_TX_PER_SEC = 154, + NVML_GPM_METRIC_C2C_LINK12_DATA_RX_PER_SEC = 155, + NVML_GPM_METRIC_C2C_LINK13_TOTAL_TX_PER_SEC = 156, + NVML_GPM_METRIC_C2C_LINK13_TOTAL_RX_PER_SEC = 157, + NVML_GPM_METRIC_C2C_LINK13_DATA_TX_PER_SEC = 158, + NVML_GPM_METRIC_C2C_LINK13_DATA_RX_PER_SEC = 159, + NVML_GPM_METRIC_HOSTMEM_CACHE_HIT = 160, + NVML_GPM_METRIC_HOSTMEM_CACHE_MISS = 161, + NVML_GPM_METRIC_PEERMEM_CACHE_HIT = 162, + NVML_GPM_METRIC_PEERMEM_CACHE_MISS = 163, + NVML_GPM_METRIC_DRAM_CACHE_HIT = 164, + NVML_GPM_METRIC_DRAM_CACHE_MISS = 165, + NVML_GPM_METRIC_NVENC_0_UTIL = 166, + NVML_GPM_METRIC_NVENC_1_UTIL = 167, + NVML_GPM_METRIC_NVENC_2_UTIL = 168, + NVML_GPM_METRIC_NVENC_3_UTIL = 169, + NVML_GPM_METRIC_GR0_CTXSW_CYCLES_ELAPSED = 170, + NVML_GPM_METRIC_GR0_CTXSW_CYCLES_ACTIVE = 171, + NVML_GPM_METRIC_GR0_CTXSW_REQUESTS = 172, + NVML_GPM_METRIC_GR0_CTXSW_CYCLES_PER_REQ = 173, + NVML_GPM_METRIC_GR0_CTXSW_ACTIVE_PCT = 174, + NVML_GPM_METRIC_GR1_CTXSW_CYCLES_ELAPSED = 175, + NVML_GPM_METRIC_GR1_CTXSW_CYCLES_ACTIVE = 176, + NVML_GPM_METRIC_GR1_CTXSW_REQUESTS = 177, + NVML_GPM_METRIC_GR1_CTXSW_CYCLES_PER_REQ = 178, + NVML_GPM_METRIC_GR1_CTXSW_ACTIVE_PCT = 179, + NVML_GPM_METRIC_GR2_CTXSW_CYCLES_ELAPSED = 180, + NVML_GPM_METRIC_GR2_CTXSW_CYCLES_ACTIVE = 181, + NVML_GPM_METRIC_GR2_CTXSW_REQUESTS = 182, + NVML_GPM_METRIC_GR2_CTXSW_CYCLES_PER_REQ = 183, + NVML_GPM_METRIC_GR2_CTXSW_ACTIVE_PCT = 184, + NVML_GPM_METRIC_GR3_CTXSW_CYCLES_ELAPSED = 185, + NVML_GPM_METRIC_GR3_CTXSW_CYCLES_ACTIVE = 186, + NVML_GPM_METRIC_GR3_CTXSW_REQUESTS = 187, + NVML_GPM_METRIC_GR3_CTXSW_CYCLES_PER_REQ = 188, + NVML_GPM_METRIC_GR3_CTXSW_ACTIVE_PCT = 189, + NVML_GPM_METRIC_GR4_CTXSW_CYCLES_ELAPSED = 190, + NVML_GPM_METRIC_GR4_CTXSW_CYCLES_ACTIVE = 191, + NVML_GPM_METRIC_GR4_CTXSW_REQUESTS = 192, + NVML_GPM_METRIC_GR4_CTXSW_CYCLES_PER_REQ = 193, + NVML_GPM_METRIC_GR4_CTXSW_ACTIVE_PCT = 194, + NVML_GPM_METRIC_GR5_CTXSW_CYCLES_ELAPSED = 195, + NVML_GPM_METRIC_GR5_CTXSW_CYCLES_ACTIVE = 196, + NVML_GPM_METRIC_GR5_CTXSW_REQUESTS = 197, + NVML_GPM_METRIC_GR5_CTXSW_CYCLES_PER_REQ = 198, + NVML_GPM_METRIC_GR5_CTXSW_ACTIVE_PCT = 199, + NVML_GPM_METRIC_GR6_CTXSW_CYCLES_ELAPSED = 200, + NVML_GPM_METRIC_GR6_CTXSW_CYCLES_ACTIVE = 201, + NVML_GPM_METRIC_GR6_CTXSW_REQUESTS = 202, + NVML_GPM_METRIC_GR6_CTXSW_CYCLES_PER_REQ = 203, + NVML_GPM_METRIC_GR6_CTXSW_ACTIVE_PCT = 204, + NVML_GPM_METRIC_GR7_CTXSW_CYCLES_ELAPSED = 205, + NVML_GPM_METRIC_GR7_CTXSW_CYCLES_ACTIVE = 206, + NVML_GPM_METRIC_GR7_CTXSW_REQUESTS = 207, + NVML_GPM_METRIC_GR7_CTXSW_CYCLES_PER_REQ = 208, + NVML_GPM_METRIC_GR7_CTXSW_ACTIVE_PCT = 209, + NVML_GPM_METRIC_MAX = 210, //!< Maximum value above +1. Note that changing this should also change NVML_GPM_METRICS_GET_VERSION due to struct size change +} nvmlGpmMetricId_t; + +/** @} */ // @defgroup nvmlGpmEnums + + +/***************************************************************************************************/ +/** @defgroup nvmlGpmStructs GPM Structs + * @{ + */ +/***************************************************************************************************/ + +/** + * Handle to an allocated GPM sample allocated with nvmlGpmSampleAlloc(). Free this with nvmlGpmSampleFree(). + */ +typedef struct +{ + struct nvmlGpmSample_st* handle; +} nvmlGpmSample_t; + +/** + * GPM metric information. + */ +typedef struct +{ + unsigned int metricId; //!< IN: NVML_GPM_METRIC_? define of which metric to retrieve + nvmlReturn_t nvmlReturn; //!< OUT: Status of this metric. If this is nonzero, then value is not valid + double value; //!< OUT: Value of this metric. Is only valid if nvmlReturn is 0 (NVML_SUCCESS) + struct + { + char *shortName; + char *longName; + char *unit; + } metricInfo; //!< OUT: Metric name and unit. Those can be NULL if not defined +} nvmlGpmMetric_t; + +/** + * GPM buffer information. + */ +typedef struct +{ + unsigned int version; //!< IN: Set to NVML_GPM_METRICS_GET_VERSION + unsigned int numMetrics; //!< IN: How many metrics to retrieve in metrics[] + nvmlGpmSample_t sample1; //!< IN: Sample buffer + nvmlGpmSample_t sample2; //!< IN: Sample buffer + nvmlGpmMetric_t metrics[NVML_GPM_METRIC_MAX]; //!< IN/OUT: Array of metrics. Set metricId on call. See nvmlReturn and value on return +} nvmlGpmMetricsGet_t; + +#define NVML_GPM_METRICS_GET_VERSION 1 + +/** + * GPM device information. + */ +typedef struct +{ + unsigned int version; //!< IN: Set to NVML_GPM_SUPPORT_VERSION + unsigned int isSupportedDevice; //!< OUT: Indicates device support +} nvmlGpmSupport_t; + +#define NVML_GPM_SUPPORT_VERSION 1 + +/** @} */ // @defgroup nvmlGPMStructs + +/***************************************************************************************************/ +/** @defgroup nvmlGpmFunctions GPM Functions + * @{ + */ +/***************************************************************************************************/ + +/** + * Calculate GPM metrics from two samples. + * + * For Hopper &tm; or newer fully supported devices. + * + * To retrieve metrics, the user must first allocate the two sample buffers at \a metricsGet->sample1 + * and \a metricsGet->sample2 by calling \a nvmlGpmSampleAlloc(). Next, the user should fill in the ID of each metric + * in \a metricsGet->metrics[i].metricId and specify the total number of metrics to retrieve in \a metricsGet->numMetrics, + * The version should be set to NVML_GPM_METRICS_GET_VERSION in \a metricsGet->version. The user then calls the + * \a nvmlGpmSampleGet() API twice to obtain 2 samples of counters. \note that the interval between these + * two \a nvmlGpmSampleGet() calls should be greater than 100ms due to the internal sample refresh rate. + * Finally, the user calls \a nvmlGpmMetricsGet to retrieve the metrics, which will be stored at \a metricsGet->metrics + * + * @param metricsGet IN/OUT: populated \a nvmlGpmMetricsGet_t struct + * + * @return + * - \ref NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum on error + */ +nvmlReturn_t DECLDIR nvmlGpmMetricsGet(nvmlGpmMetricsGet_t *metricsGet); + + +/** + * Free an allocated sample buffer that was allocated with \ref nvmlGpmSampleAlloc() + * + * For Hopper &tm; or newer fully supported devices. + * + * @param gpmSample Sample to free + * + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid pointer is provided + */ +nvmlReturn_t DECLDIR nvmlGpmSampleFree(nvmlGpmSample_t gpmSample); + + +/** + * Allocate a sample buffer to be used with NVML GPM . You will need to allocate + * at least two of these buffers to use with the NVML GPM feature + * + * For Hopper &tm; or newer fully supported devices. + * + * @param gpmSample Where the allocated sample will be stored + * + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid pointer is provided + * - \ref NVML_ERROR_MEMORY if system memory is insufficient + */ +nvmlReturn_t DECLDIR nvmlGpmSampleAlloc(nvmlGpmSample_t *gpmSample); + +/** + * Read a sample of GPM metrics into the provided \a gpmSample buffer. After + * two samples are gathered, you can call nvmlGpmMetricGet on those samples to + * retrive metrics + * + * For Hopper &tm; or newer fully supported devices. + * + * @note The interval between two \a nvmlGpmSampleGet() calls should be greater than 100ms due to + * the internal sample refresh rate. + * + * @param device Device to get samples for + * @param gpmSample Buffer to read samples into + * + * @return + * - \ref NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum on error + */ +nvmlReturn_t DECLDIR nvmlGpmSampleGet(nvmlDevice_t device, nvmlGpmSample_t gpmSample); + +/** + * Read a sample of GPM metrics into the provided \a gpmSample buffer for a MIG GPU Instance. + * + * After two samples are gathered, you can call nvmlGpmMetricGet on those + * samples to retrive metrics + * + * For Hopper &tm; or newer fully supported devices. + * + * @note The interval between two \a nvmlGpmMigSampleGet() calls should be greater than 100ms due to + * the internal sample refresh rate. + * + * @param device Device to get samples for + * @param gpuInstanceId MIG GPU Instance ID + * @param gpmSample Buffer to read samples into + * + * @return + * - \ref NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum on error + */ +nvmlReturn_t DECLDIR nvmlGpmMigSampleGet(nvmlDevice_t device, unsigned int gpuInstanceId, nvmlGpmSample_t gpmSample); + +/** + * Indicate whether the supplied device supports GPM + * + * For Hopper &tm; or newer fully supported devices. + * + * @param device NVML device to query for + * @param gpmSupport Structure to indicate GPM support \a nvmlGpmSupport_t. Indicates + * GPM support per system for the supplied device + * + * @return + * - NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum if there is an error in processing the query + */ +nvmlReturn_t DECLDIR nvmlGpmQueryDeviceSupport(nvmlDevice_t device, nvmlGpmSupport_t *gpmSupport); + +/* GPM Stream State */ +/** + * Get GPM stream state. + * + * For Hopper &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param state Returns GPM stream state + * NVML_FEATURE_DISABLED or NVML_FEATURE_ENABLED + * + * @return + * - \ref NVML_SUCCESS if \a current GPM stream state were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a state is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlGpmQueryIfStreamingEnabled(nvmlDevice_t device, unsigned int *state); + +/** + * Set GPM stream state. + * + * For Hopper &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param state GPM stream state, + * NVML_FEATURE_DISABLED or NVML_FEATURE_ENABLED + * + * @return + * - \ref NVML_SUCCESS if \a current GPM stream state is successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlGpmSetStreamingEnabled(nvmlDevice_t device, unsigned int state); + +/** @} */ // @defgroup nvmlGpmFunctions +/** @} */ // @defgroup GPM + +#define NVML_DEV_CAP_EGM (1 << 0) // Extended GPU memory +/** + * Device capabilities + */ +typedef struct +{ + unsigned int version; //!< the API version number + unsigned int capMask; //!< OUT: Bit mask of capabilities. +} nvmlDeviceCapabilities_v1_t; +typedef nvmlDeviceCapabilities_v1_t nvmlDeviceCapabilities_t; +#define nvmlDeviceCapabilities_v1 NVML_STRUCT_VERSION(DeviceCapabilities, 1) + +/** + * Get device capabilities + * + * See \ref nvmlDeviceCapabilities_v1_t for more information on the struct. + * + * @param device The identifier of the target device + * @param caps Returns GPU's capabilities + * + * @return + * - \ref NVML_SUCCESS If the query is success + * - \ref NVML_ERROR_UNINITIALIZED If the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device is invalid or \a counters is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED If the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST If the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH If the provided version is invalid/unsupported + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCapabilities(nvmlDevice_t device, + nvmlDeviceCapabilities_t *caps); + + +/* + * Generic bitmask to hold 255 bits, represented by 8 elements of 32 bits + */ +#define NVML_255_MASK_BITS_PER_ELEM 32 +#define NVML_255_MASK_NUM_ELEMS 8 +#define NVML_255_MASK_BIT_SET(index, nvmlMask) \ + nvmlMask.mask[index / NVML_255_MASK_BITS_PER_ELEM] |= (1 << (index % NVML_255_MASK_BITS_PER_ELEM)) + +#define NVML_255_MASK_BIT_GET(index, nvmlMask) \ + nvmlMask.mask[index / NVML_255_MASK_BITS_PER_ELEM] & (1 << (index % NVML_255_MASK_BITS_PER_ELEM)) + +#define NVML_255_MASK_BIT_SET_PTR(index, nvmlMask) \ + nvmlMask->mask[index / NVML_255_MASK_BITS_PER_ELEM] |= (1 << (index % NVML_255_MASK_BITS_PER_ELEM)) + +#define NVML_255_MASK_BIT_GET_PTR(index, nvmlMask) \ + nvmlMask->mask[index / NVML_255_MASK_BITS_PER_ELEM] & (1 << (index % NVML_255_MASK_BITS_PER_ELEM)) + +typedef struct +{ + unsigned int mask[NVML_255_MASK_NUM_ELEMS]; //profileId is used and + * the rest of the structure is ignored. + * + * @return + * - \ref NVML_SUCCESS if the Desired Profile was successfully set + * - \ref NVML_ERROR_INVALID_ARGUMENT if device is invalid or structure was NULL + * - \ref NVML_ERROR_NO_PERMISSION if user does not have permission to change the profile number + * - \ref NVML_ERROR_NOT_SUPPORTED if this feature is not supported by the device + * + **/ +nvmlReturn_t DECLDIR nvmlDevicePowerSmoothingActivatePresetProfile(nvmlDevice_t device, + nvmlPowerSmoothingProfile_t *profile); + +/** + * Update the value of a specific profile parameter contained within \ref nvmlPowerSmoothingProfile_v1_t. + * Requires root/admin permissions. + * + * %BLACKWELL_OR_NEWER% + * + * NVML_POWER_SMOOTHING_PROFILE_PARAM_PERCENT_TMP_FLOOR expects a value as a percentage from 00.00-100.00% + * NVML_POWER_SMOOTHING_PROFILE_PARAM_RAMP_UP_RATE expects a value in W/s + * NVML_POWER_SMOOTHING_PROFILE_PARAM_RAMP_DOWN_RATE expects a value in W/s + * NVML_POWER_SMOOTHING_PROFILE_PARAM_RAMP_DOWN_HYSTERESIS expects a value in ms + * + * @param device The identifier of the target device + * @param profile Reference to \ref nvmlPowerSmoothingProfile_v1_t struct + * + * @return + * - \ref NVML_SUCCESS if the Active Profile was successfully set + * - \ref NVML_ERROR_INVALID_ARGUMENT if device is invalid or profile parameter/value was invalid + * - \ref NVML_ERROR_NO_PERMISSION if user does not have permission to change any profile parameters + * - \ref NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the structure version is not supported + * + **/ +nvmlReturn_t DECLDIR nvmlDevicePowerSmoothingUpdatePresetProfileParam(nvmlDevice_t device, + nvmlPowerSmoothingProfile_t *profile); +/** + * Enable or disable the Power Smoothing Feature. + * Requires root/admin permissions. + * + * %BLACKWELL_OR_NEWER% + * + * See \ref nvmlEnableState_t for details on allowed states + * + * @param device The identifier of the target device + * @param state Reference to \ref nvmlPowerSmoothingState_v1_t + * + * @return + * - \ref NVML_SUCCESS if the feature state was successfully set + * - \ref NVML_ERROR_INVALID_ARGUMENT if device is invalid or state is NULL + * - \ref NVML_ERROR_NO_PERMISSION if user does not have permission to change feature state + * - \ref NVML_ERROR_NOT_SUPPORTED if this feature is not supported by the device + * + **/ +nvmlReturn_t DECLDIR nvmlDevicePowerSmoothingSetState(nvmlDevice_t device, + nvmlPowerSmoothingState_t *state); +/** @} */ // @defgroup + +/** + * Retrieves the counts of SRAM unique uncorrected ECC errors + * + * %BLACKWELL_OR_NEWER% + * + * Reads SRAM unique uncorrected ECC error counts. The total number of unique errors is returned by + * \a errorCounts->entryCount. Error counts are returned as an array of in the caller-supplied buffer pointed at by + * \a errorCounts->entries. Each error count entry holds the location/address of the unique error, the error count and + * whether the error is parity or not. + * + * To read SRAM unique uncorrected ECC error counts, first determine the size of buffer required to hold the error + * counts by invoking the function with \a errorCounts->entries set to NULL. The required array size is returned in + * \a errorCounts->entryCount. The caller should allocate a buffer of size "errorCounts->entryCount * + * sizeof(nvmlEccSramUniqueUncorrectedErrorCounts_t)". Invoke the function again with the allocated buffer passed in + * \a errorCounts->entries. This time \a errorCounts->entryCount will be taken as the entry array size that caller + * allocates for \a errorCounts->entries. + * + * On successful return of the second query, the function updates \a errorCounts->entries with all unique errors. This + * may fail if \a errorCounts->entryCount is smaller than the actual number of unique errors. This can happen in cases + * like new errors occur since the previous query of \a errorCounts->entryCount. No matter the query succeeds or not, + * the latest number of unique errors will be returned in \a errorCounts->entryCount. + * + * @note The query is only supported when ECC mode is enabled. + * + * @param device The identifier of the target device + * @param errorCounts Pointer to caller-supplied array which returns the unique error count entries + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a errorCounts->entryCount is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature or ECC mods is not enabled + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if the allocated error entry array is not big enough + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSramUniqueUncorrectedEccErrorCounts(nvmlDevice_t device, + nvmlEccSramUniqueUncorrectedErrorCounts_t *errorCounts); + +/** + * NVML API versioning support + */ + +#ifdef NVML_NO_UNVERSIONED_FUNC_DEFS +nvmlReturn_t DECLDIR nvmlInit(void); +nvmlReturn_t DECLDIR nvmlDeviceGetCount(unsigned int *deviceCount); +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByIndex(unsigned int index, nvmlDevice_t *device); +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByPciBusId(const char *pciBusId, nvmlDevice_t *device); +nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo(nvmlDevice_t device, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo_v2(nvmlDevice_t device, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemotePciInfo(nvmlDevice_t device, unsigned int link, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures_v2(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures_v3(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); +nvmlReturn_t DECLDIR nvmlDeviceRemoveGpu(nvmlPciInfo_t *pciInfo); +nvmlReturn_t DECLDIR nvmlEventSetWait(nvmlEventSet_t set, nvmlEventData_t * data, unsigned int timeoutms); +nvmlReturn_t DECLDIR nvmlDeviceGetAttributes(nvmlDevice_t device, nvmlDeviceAttributes_t *attributes); +nvmlReturn_t DECLDIR nvmlComputeInstanceGetInfo(nvmlComputeInstance_t computeInstance, nvmlComputeInstanceInfo_t *info); +nvmlReturn_t DECLDIR nvmlDeviceGetComputeRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v1_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetComputeRunningProcesses_v2(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v2_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetGraphicsRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v1_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetGraphicsRunningProcesses_v2(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v2_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetMPSComputeRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v1_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetMPSComputeRunningProcesses_v2(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v2_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstancePossiblePlacements(nvmlDevice_t device, unsigned int profileId, nvmlGpuInstancePlacement_t *placements, unsigned int *count); +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetLicenseInfo(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuLicenseInfo_t *licenseInfo); +nvmlReturn_t DECLDIR nvmlDeviceGetDriverModel(nvmlDevice_t device, nvmlDriverModel_t *current, nvmlDriverModel_t *pending); +#endif // #ifdef NVML_NO_UNVERSIONED_FUNC_DEFS + +#if defined(NVML_NO_UNVERSIONED_FUNC_DEFS) +// We don't define APIs to run new versions if this guard is present so there is +// no need to undef +#elif defined(__NVML_API_VERSION_INTERNAL) +#undef nvmlDeviceGetGraphicsRunningProcesses +#undef nvmlDeviceGetComputeRunningProcesses +#undef nvmlDeviceGetMPSComputeRunningProcesses +#undef nvmlDeviceGetAttributes +#undef nvmlComputeInstanceGetInfo +#undef nvmlEventSetWait +#undef nvmlDeviceGetGridLicensableFeatures +#undef nvmlDeviceRemoveGpu +#undef nvmlDeviceGetNvLinkRemotePciInfo +#undef nvmlDeviceGetPciInfo +#undef nvmlDeviceGetCount +#undef nvmlDeviceGetHandleByIndex +#undef nvmlDeviceGetHandleByPciBusId +#undef nvmlInit +#undef nvmlBlacklistDeviceInfo_t +#undef nvmlGetBlacklistDeviceCount +#undef nvmlGetBlacklistDeviceInfoByIndex +#undef nvmlDeviceGetGpuInstancePossiblePlacements +#undef nvmlVgpuInstanceGetLicenseInfo +#undef nvmlDeviceGetDriverModel +#undef nvmlDeviceSetPowerManagementLimit + +#endif + +#ifdef __cplusplus +} +#endif + +#endif diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/refcount.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/refcount.go new file mode 100644 index 00000000..4d1e212e --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/refcount.go @@ -0,0 +1,31 @@ +/** +# Copyright 2024 NVIDIA CORPORATION +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +**/ + +package nvml + +type refcount int + +func (r *refcount) IncOnNoError(err error) { + if err == nil { + (*r)++ + } +} + +func (r *refcount) DecOnNoError(err error) { + if err == nil && (*r) > 0 { + (*r)-- + } +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/return.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/return.go new file mode 100644 index 00000000..1aec3ecc --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/return.go @@ -0,0 +1,103 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +import ( + "fmt" +) + +// nvml.ErrorString() +func (l *library) ErrorString(r Return) string { + return r.Error() +} + +// String returns the string representation of a Return. +func (r Return) String() string { + return r.Error() +} + +// Error returns the string representation of a Return. +func (r Return) Error() string { + return errorStringFunc(r) +} + +// Assigned to nvml.ErrorString if the system nvml library is in use. +var errorStringFunc = defaultErrorStringFunc + +// defaultErrorStringFunc provides a basic nvmlErrorString implementation. +// This allows the nvml.ErrorString function to be used even if the NVML library +// is not loaded. +var defaultErrorStringFunc = func(r Return) string { + switch r { + case SUCCESS: + return "SUCCESS" + case ERROR_UNINITIALIZED: + return "ERROR_UNINITIALIZED" + case ERROR_INVALID_ARGUMENT: + return "ERROR_INVALID_ARGUMENT" + case ERROR_NOT_SUPPORTED: + return "ERROR_NOT_SUPPORTED" + case ERROR_NO_PERMISSION: + return "ERROR_NO_PERMISSION" + case ERROR_ALREADY_INITIALIZED: + return "ERROR_ALREADY_INITIALIZED" + case ERROR_NOT_FOUND: + return "ERROR_NOT_FOUND" + case ERROR_INSUFFICIENT_SIZE: + return "ERROR_INSUFFICIENT_SIZE" + case ERROR_INSUFFICIENT_POWER: + return "ERROR_INSUFFICIENT_POWER" + case ERROR_DRIVER_NOT_LOADED: + return "ERROR_DRIVER_NOT_LOADED" + case ERROR_TIMEOUT: + return "ERROR_TIMEOUT" + case ERROR_IRQ_ISSUE: + return "ERROR_IRQ_ISSUE" + case ERROR_LIBRARY_NOT_FOUND: + return "ERROR_LIBRARY_NOT_FOUND" + case ERROR_FUNCTION_NOT_FOUND: + return "ERROR_FUNCTION_NOT_FOUND" + case ERROR_CORRUPTED_INFOROM: + return "ERROR_CORRUPTED_INFOROM" + case ERROR_GPU_IS_LOST: + return "ERROR_GPU_IS_LOST" + case ERROR_RESET_REQUIRED: + return "ERROR_RESET_REQUIRED" + case ERROR_OPERATING_SYSTEM: + return "ERROR_OPERATING_SYSTEM" + case ERROR_LIB_RM_VERSION_MISMATCH: + return "ERROR_LIB_RM_VERSION_MISMATCH" + case ERROR_IN_USE: + return "ERROR_IN_USE" + case ERROR_MEMORY: + return "ERROR_MEMORY" + case ERROR_NO_DATA: + return "ERROR_NO_DATA" + case ERROR_VGPU_ECC_NOT_SUPPORTED: + return "ERROR_VGPU_ECC_NOT_SUPPORTED" + case ERROR_INSUFFICIENT_RESOURCES: + return "ERROR_INSUFFICIENT_RESOURCES" + case ERROR_FREQ_NOT_SUPPORTED: + return "ERROR_FREQ_NOT_SUPPORTED" + case ERROR_ARGUMENT_VERSION_MISMATCH: + return "ERROR_ARGUMENT_VERSION_MISMATCH" + case ERROR_DEPRECATED: + return "ERROR_DEPRECATED" + case ERROR_UNKNOWN: + return "ERROR_UNKNOWN" + default: + return fmt.Sprintf("unknown return value: %d", r) + } +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/system.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/system.go new file mode 100644 index 00000000..f8247102 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/system.go @@ -0,0 +1,148 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +// nvml.SystemGetDriverVersion() +func (l *library) SystemGetDriverVersion() (string, Return) { + Version := make([]byte, SYSTEM_DRIVER_VERSION_BUFFER_SIZE) + ret := nvmlSystemGetDriverVersion(&Version[0], SYSTEM_DRIVER_VERSION_BUFFER_SIZE) + return string(Version[:clen(Version)]), ret +} + +// nvml.SystemGetNVMLVersion() +func (l *library) SystemGetNVMLVersion() (string, Return) { + Version := make([]byte, SYSTEM_NVML_VERSION_BUFFER_SIZE) + ret := nvmlSystemGetNVMLVersion(&Version[0], SYSTEM_NVML_VERSION_BUFFER_SIZE) + return string(Version[:clen(Version)]), ret +} + +// nvml.SystemGetCudaDriverVersion() +func (l *library) SystemGetCudaDriverVersion() (int, Return) { + var CudaDriverVersion int32 + ret := nvmlSystemGetCudaDriverVersion(&CudaDriverVersion) + return int(CudaDriverVersion), ret +} + +// nvml.SystemGetCudaDriverVersion_v2() +func (l *library) SystemGetCudaDriverVersion_v2() (int, Return) { + var CudaDriverVersion int32 + ret := nvmlSystemGetCudaDriverVersion_v2(&CudaDriverVersion) + return int(CudaDriverVersion), ret +} + +// nvml.SystemGetProcessName() +func (l *library) SystemGetProcessName(pid int) (string, Return) { + name := make([]byte, SYSTEM_PROCESS_NAME_BUFFER_SIZE) + ret := nvmlSystemGetProcessName(uint32(pid), &name[0], SYSTEM_PROCESS_NAME_BUFFER_SIZE) + return string(name[:clen(name)]), ret +} + +// nvml.SystemGetHicVersion() +func (l *library) SystemGetHicVersion() ([]HwbcEntry, Return) { + var hwbcCount uint32 = 1 // Will be reduced upon returning + for { + hwbcEntries := make([]HwbcEntry, hwbcCount) + ret := nvmlSystemGetHicVersion(&hwbcCount, &hwbcEntries[0]) + if ret == SUCCESS { + return hwbcEntries[:hwbcCount], ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + hwbcCount *= 2 + } +} + +// nvml.SystemGetTopologyGpuSet() +func (l *library) SystemGetTopologyGpuSet(cpuNumber int) ([]Device, Return) { + var count uint32 + ret := nvmlSystemGetTopologyGpuSet(uint32(cpuNumber), &count, nil) + if ret != SUCCESS { + return nil, ret + } + if count == 0 { + return []Device{}, ret + } + deviceArray := make([]nvmlDevice, count) + ret = nvmlSystemGetTopologyGpuSet(uint32(cpuNumber), &count, &deviceArray[0]) + return convertSlice[nvmlDevice, Device](deviceArray), ret +} + +// nvml.SystemGetConfComputeCapabilities() +func (l *library) SystemGetConfComputeCapabilities() (ConfComputeSystemCaps, Return) { + var capabilities ConfComputeSystemCaps + ret := nvmlSystemGetConfComputeCapabilities(&capabilities) + return capabilities, ret +} + +// nvml.SystemGetConfComputeState() +func (l *library) SystemGetConfComputeState() (ConfComputeSystemState, Return) { + var state ConfComputeSystemState + ret := nvmlSystemGetConfComputeState(&state) + return state, ret +} + +// nvml.SystemGetConfComputeGpusReadyState() +func (l *library) SystemGetConfComputeGpusReadyState() (uint32, Return) { + var isAcceptingWork uint32 + ret := nvmlSystemGetConfComputeGpusReadyState(&isAcceptingWork) + return isAcceptingWork, ret +} + +// nvml.SystemSetConfComputeGpusReadyState() +func (l *library) SystemSetConfComputeGpusReadyState(isAcceptingWork uint32) Return { + return nvmlSystemSetConfComputeGpusReadyState(isAcceptingWork) +} + +// nvml.SystemSetNvlinkBwMode() +func (l *library) SystemSetNvlinkBwMode(nvlinkBwMode uint32) Return { + return nvmlSystemSetNvlinkBwMode(nvlinkBwMode) +} + +// nvml.SystemGetNvlinkBwMode() +func (l *library) SystemGetNvlinkBwMode() (uint32, Return) { + var nvlinkBwMode uint32 + ret := nvmlSystemGetNvlinkBwMode(&nvlinkBwMode) + return nvlinkBwMode, ret +} + +// nvml.SystemGetConfComputeKeyRotationThresholdInfo() +func (l *library) SystemGetConfComputeKeyRotationThresholdInfo() (ConfComputeGetKeyRotationThresholdInfo, Return) { + var keyRotationThresholdInfo ConfComputeGetKeyRotationThresholdInfo + keyRotationThresholdInfo.Version = STRUCT_VERSION(keyRotationThresholdInfo, 1) + ret := nvmlSystemGetConfComputeKeyRotationThresholdInfo(&keyRotationThresholdInfo) + return keyRotationThresholdInfo, ret +} + +// nvml.SystemGetConfComputeSettings() +func (l *library) SystemGetConfComputeSettings() (SystemConfComputeSettings, Return) { + var settings SystemConfComputeSettings + settings.Version = STRUCT_VERSION(settings, 1) + ret := nvmlSystemGetConfComputeSettings(&settings) + return settings, ret +} + +// nvml.SystemSetConfComputeKeyRotationThresholdInfo() +func (l *library) SystemSetConfComputeKeyRotationThresholdInfo(keyRotationThresholdInfo ConfComputeSetKeyRotationThresholdInfo) Return { + return nvmlSystemSetConfComputeKeyRotationThresholdInfo(&keyRotationThresholdInfo) +} + +// nvml.SystemGetDriverBranch() +func (l *library) SystemGetDriverBranch() (SystemDriverBranchInfo, Return) { + var branchInfo SystemDriverBranchInfo + branchInfo.Version = STRUCT_VERSION(branchInfo, 1) + ret := nvmlSystemGetDriverBranch(&branchInfo, SYSTEM_DRIVER_VERSION_BUFFER_SIZE) + return branchInfo, ret +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/types_gen.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/types_gen.go new file mode 100644 index 00000000..efa58637 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/types_gen.go @@ -0,0 +1,1462 @@ +// Code generated by cmd/cgo -godefs; DO NOT EDIT. +// cgo -godefs types.go + +package nvml + +import "unsafe" + +type nvmlDevice struct { + Handle *_Ctype_struct_nvmlDevice_st +} + +type nvmlGpuInstance struct { + Handle *_Ctype_struct_nvmlGpuInstance_st +} + +type PciInfoExt_v1 struct { + Version uint32 + Domain uint32 + Bus uint32 + Device uint32 + PciDeviceId uint32 + PciSubSystemId uint32 + BaseClass uint32 + SubClass uint32 + BusId [32]uint8 +} + +type PciInfoExt struct { + Version uint32 + Domain uint32 + Bus uint32 + Device uint32 + PciDeviceId uint32 + PciSubSystemId uint32 + BaseClass uint32 + SubClass uint32 + BusId [32]uint8 +} + +type PciInfo struct { + BusIdLegacy [16]uint8 + Domain uint32 + Bus uint32 + Device uint32 + PciDeviceId uint32 + PciSubSystemId uint32 + BusId [32]uint8 +} + +type EccErrorCounts struct { + L1Cache uint64 + L2Cache uint64 + DeviceMemory uint64 + RegisterFile uint64 +} + +type Utilization struct { + Gpu uint32 + Memory uint32 +} + +type Memory struct { + Total uint64 + Free uint64 + Used uint64 +} + +type Memory_v2 struct { + Version uint32 + Total uint64 + Reserved uint64 + Free uint64 + Used uint64 +} + +type BAR1Memory struct { + Bar1Total uint64 + Bar1Free uint64 + Bar1Used uint64 +} + +type ProcessInfo_v1 struct { + Pid uint32 + UsedGpuMemory uint64 +} + +type ProcessInfo_v2 struct { + Pid uint32 + UsedGpuMemory uint64 + GpuInstanceId uint32 + ComputeInstanceId uint32 +} + +type ProcessInfo struct { + Pid uint32 + UsedGpuMemory uint64 + GpuInstanceId uint32 + ComputeInstanceId uint32 +} + +type ProcessDetail_v1 struct { + Pid uint32 + UsedGpuMemory uint64 + GpuInstanceId uint32 + ComputeInstanceId uint32 + UsedGpuCcProtectedMemory uint64 +} + +type ProcessDetailList_v1 struct { + Version uint32 + Mode uint32 + NumProcArrayEntries uint32 + ProcArray *ProcessDetail_v1 +} + +type ProcessDetailList struct { + Version uint32 + Mode uint32 + NumProcArrayEntries uint32 + ProcArray *ProcessDetail_v1 +} + +type DeviceAttributes struct { + MultiprocessorCount uint32 + SharedCopyEngineCount uint32 + SharedDecoderCount uint32 + SharedEncoderCount uint32 + SharedJpegCount uint32 + SharedOfaCount uint32 + GpuInstanceSliceCount uint32 + ComputeInstanceSliceCount uint32 + MemorySizeMB uint64 +} + +type C2cModeInfo_v1 struct { + IsC2cEnabled uint32 +} + +type DeviceAddressingMode_v1 struct { + Version uint32 + Value uint32 +} + +type DeviceAddressingMode struct { + Version uint32 + Value uint32 +} + +type RepairStatus_v1 struct { + Version uint32 + BChannelRepairPending uint32 + BTpcRepairPending uint32 +} + +type RepairStatus struct { + Version uint32 + BChannelRepairPending uint32 + BTpcRepairPending uint32 +} + +type RowRemapperHistogramValues struct { + Max uint32 + High uint32 + Partial uint32 + Low uint32 + None uint32 +} + +type NvLinkUtilizationControl struct { + Units uint32 + Pktfilter uint32 +} + +type BridgeChipInfo struct { + Type uint32 + FwVersion uint32 +} + +type BridgeChipHierarchy struct { + BridgeCount uint8 + BridgeChipInfo [128]BridgeChipInfo +} + +const sizeofValue = unsafe.Sizeof([8]byte{}) + +type Value [sizeofValue]byte + +type Sample struct { + TimeStamp uint64 + SampleValue [8]byte +} + +type ViolationTime struct { + ReferenceTime uint64 + ViolationTime uint64 +} + +type GpuThermalSettings struct { + Count uint32 + Sensor [3]_Ctype_struct___28 +} + +type CoolerInfo_v1 struct { + Version uint32 + Index uint32 + SignalType uint32 + Target uint32 +} + +type CoolerInfo struct { + Version uint32 + Index uint32 + SignalType uint32 + Target uint32 +} + +const sizeofUUIDValue = unsafe.Sizeof([41]byte{}) + +type UUIDValue [sizeofUUIDValue]byte + +type UUID_v1 struct { + Version uint32 + Type uint32 + Value [41]byte + Pad_cgo_0 [3]byte +} + +type UUID struct { + Version uint32 + Type uint32 + Value [41]byte + Pad_cgo_0 [3]byte +} + +type Pdi_v1 struct { + Version uint32 + Value uint64 +} + +type Pdi struct { + Version uint32 + Value uint64 +} + +type DramEncryptionInfo_v1 struct { + Version uint32 + EncryptionState uint32 +} + +type DramEncryptionInfo struct { + Version uint32 + EncryptionState uint32 +} + +type MarginTemperature_v1 struct { + Version uint32 + MarginTemperature int32 +} + +type MarginTemperature struct { + Version uint32 + MarginTemperature int32 +} + +type ClkMonFaultInfo struct { + ClkApiDomain uint32 + ClkDomainFaultMask uint32 +} + +type ClkMonStatus struct { + BGlobalStatus uint32 + ClkMonListSize uint32 + ClkMonList [32]ClkMonFaultInfo +} + +type ClockOffset_v1 struct { + Version uint32 + Type uint32 + Pstate uint32 + ClockOffsetMHz int32 + MinClockOffsetMHz int32 + MaxClockOffsetMHz int32 +} + +type ClockOffset struct { + Version uint32 + Type uint32 + Pstate uint32 + ClockOffsetMHz int32 + MinClockOffsetMHz int32 + MaxClockOffsetMHz int32 +} + +type FanSpeedInfo_v1 struct { + Version uint32 + Fan uint32 + Speed uint32 +} + +type FanSpeedInfo struct { + Version uint32 + Fan uint32 + Speed uint32 +} + +type DevicePerfModes_v1 struct { + Version uint32 + Str [2048]uint8 +} + +type DevicePerfModes struct { + Version uint32 + Str [2048]uint8 +} + +type DeviceCurrentClockFreqs_v1 struct { + Version uint32 + Str [2048]uint8 +} + +type DeviceCurrentClockFreqs struct { + Version uint32 + Str [2048]uint8 +} + +type DevicePowerMizerModes_v1 struct { + CurrentMode uint32 + Mode uint32 + SupportedPowerMizerModes uint32 +} + +type ProcessUtilizationSample struct { + Pid uint32 + TimeStamp uint64 + SmUtil uint32 + MemUtil uint32 + EncUtil uint32 + DecUtil uint32 +} + +type ProcessUtilizationInfo_v1 struct { + TimeStamp uint64 + Pid uint32 + SmUtil uint32 + MemUtil uint32 + EncUtil uint32 + DecUtil uint32 + JpgUtil uint32 + OfaUtil uint32 + Pad_cgo_0 [4]byte +} + +type ProcessesUtilizationInfo_v1 struct { + Version uint32 + ProcessSamplesCount uint32 + LastSeenTimeStamp uint64 + ProcUtilArray *ProcessUtilizationInfo_v1 +} + +type ProcessesUtilizationInfo struct { + Version uint32 + ProcessSamplesCount uint32 + LastSeenTimeStamp uint64 + ProcUtilArray *ProcessUtilizationInfo_v1 +} + +type EccSramErrorStatus_v1 struct { + Version uint32 + AggregateUncParity uint64 + AggregateUncSecDed uint64 + AggregateCor uint64 + VolatileUncParity uint64 + VolatileUncSecDed uint64 + VolatileCor uint64 + AggregateUncBucketL2 uint64 + AggregateUncBucketSm uint64 + AggregateUncBucketPcie uint64 + AggregateUncBucketMcu uint64 + AggregateUncBucketOther uint64 + BThresholdExceeded uint32 + Pad_cgo_0 [4]byte +} + +type EccSramErrorStatus struct { + Version uint32 + AggregateUncParity uint64 + AggregateUncSecDed uint64 + AggregateCor uint64 + VolatileUncParity uint64 + VolatileUncSecDed uint64 + VolatileCor uint64 + AggregateUncBucketL2 uint64 + AggregateUncBucketSm uint64 + AggregateUncBucketPcie uint64 + AggregateUncBucketMcu uint64 + AggregateUncBucketOther uint64 + BThresholdExceeded uint32 + Pad_cgo_0 [4]byte +} + +type PlatformInfo_v1 struct { + Version uint32 + IbGuid [16]uint8 + RackGuid [16]uint8 + ChassisPhysicalSlotNumber uint8 + ComputeSlotIndex uint8 + NodeIndex uint8 + PeerType uint8 + ModuleId uint8 + Pad_cgo_0 [3]byte +} + +type PlatformInfo_v2 struct { + Version uint32 + IbGuid [16]uint8 + ChassisSerialNumber [16]uint8 + SlotNumber uint8 + TrayIndex uint8 + HostId uint8 + PeerType uint8 + ModuleId uint8 + Pad_cgo_0 [3]byte +} + +type PlatformInfo struct { + Version uint32 + IbGuid [16]uint8 + ChassisSerialNumber [16]uint8 + SlotNumber uint8 + TrayIndex uint8 + HostId uint8 + PeerType uint8 + ModuleId uint8 + Pad_cgo_0 [3]byte +} + +type EccSramUniqueUncorrectedErrorEntry_v1 struct { + Unit uint32 + Location uint32 + Sublocation uint32 + Extlocation uint32 + Address uint32 + IsParity uint32 + Count uint32 +} + +type EccSramUniqueUncorrectedErrorCounts_v1 struct { + Version uint32 + EntryCount uint32 + Entries *EccSramUniqueUncorrectedErrorEntry_v1 +} + +type EccSramUniqueUncorrectedErrorCounts struct { + Version uint32 + EntryCount uint32 + Entries *EccSramUniqueUncorrectedErrorEntry_v1 +} + +type DeviceArchitecture uint32 + +type BusType uint32 + +type FanControlPolicy uint32 + +type PowerSource uint32 + +type GpuDynamicPstatesInfo struct { + Flags uint32 + Utilization [8]_Ctype_struct___23 +} + +type PowerScopeType byte + +type PowerValue_v2 struct { + Version uint32 + PowerScope uint8 + PowerValueMw uint32 +} + +type nvmlVgpuTypeId uint32 + +type nvmlVgpuInstance uint32 + +type VgpuHeterogeneousMode_v1 struct { + Version uint32 + Mode uint32 +} + +type VgpuHeterogeneousMode struct { + Version uint32 + Mode uint32 +} + +type VgpuPlacementId_v1 struct { + Version uint32 + PlacementId uint32 +} + +type VgpuPlacementId struct { + Version uint32 + PlacementId uint32 +} + +type VgpuPlacementList_v1 struct { + Version uint32 + PlacementSize uint32 + Count uint32 + PlacementIds *uint32 +} + +type VgpuPlacementList_v2 struct { + Version uint32 + PlacementSize uint32 + Count uint32 + PlacementIds *uint32 + Mode uint32 + Pad_cgo_0 [4]byte +} + +type VgpuPlacementList struct { + Version uint32 + PlacementSize uint32 + Count uint32 + PlacementIds *uint32 + Mode uint32 + Pad_cgo_0 [4]byte +} + +type VgpuTypeBar1Info_v1 struct { + Version uint32 + Bar1Size uint64 +} + +type VgpuTypeBar1Info struct { + Version uint32 + Bar1Size uint64 +} + +type VgpuInstanceUtilizationSample struct { + VgpuInstance uint32 + TimeStamp uint64 + SmUtil [8]byte + MemUtil [8]byte + EncUtil [8]byte + DecUtil [8]byte +} + +type VgpuInstanceUtilizationInfo_v1 struct { + TimeStamp uint64 + VgpuInstance uint32 + Pad_cgo_0 [4]byte + SmUtil [8]byte + MemUtil [8]byte + EncUtil [8]byte + DecUtil [8]byte + JpgUtil [8]byte + OfaUtil [8]byte +} + +type VgpuInstancesUtilizationInfo_v1 struct { + Version uint32 + SampleValType uint32 + VgpuInstanceCount uint32 + LastSeenTimeStamp uint64 + VgpuUtilArray *VgpuInstanceUtilizationInfo_v1 +} + +type VgpuInstancesUtilizationInfo struct { + Version uint32 + SampleValType uint32 + VgpuInstanceCount uint32 + LastSeenTimeStamp uint64 + VgpuUtilArray *VgpuInstanceUtilizationInfo_v1 +} + +type VgpuProcessUtilizationSample struct { + VgpuInstance uint32 + Pid uint32 + ProcessName [64]uint8 + TimeStamp uint64 + SmUtil uint32 + MemUtil uint32 + EncUtil uint32 + DecUtil uint32 +} + +type VgpuProcessUtilizationInfo_v1 struct { + ProcessName [64]uint8 + TimeStamp uint64 + VgpuInstance uint32 + Pid uint32 + SmUtil uint32 + MemUtil uint32 + EncUtil uint32 + DecUtil uint32 + JpgUtil uint32 + OfaUtil uint32 +} + +type VgpuProcessesUtilizationInfo_v1 struct { + Version uint32 + VgpuProcessCount uint32 + LastSeenTimeStamp uint64 + VgpuProcUtilArray *VgpuProcessUtilizationInfo_v1 +} + +type VgpuProcessesUtilizationInfo struct { + Version uint32 + VgpuProcessCount uint32 + LastSeenTimeStamp uint64 + VgpuProcUtilArray *VgpuProcessUtilizationInfo_v1 +} + +type VgpuRuntimeState_v1 struct { + Version uint32 + Size uint64 +} + +type VgpuRuntimeState struct { + Version uint32 + Size uint64 +} + +const sizeofVgpuSchedulerParams = unsafe.Sizeof([8]byte{}) + +type VgpuSchedulerParams [sizeofVgpuSchedulerParams]byte + +type VgpuSchedulerLogEntry struct { + Timestamp uint64 + TimeRunTotal uint64 + TimeRun uint64 + SwRunlistId uint32 + TargetTimeSlice uint64 + CumulativePreemptionTime uint64 +} + +type VgpuSchedulerLog struct { + EngineId uint32 + SchedulerPolicy uint32 + ArrMode uint32 + SchedulerParams [8]byte + EntriesCount uint32 + LogEntries [200]VgpuSchedulerLogEntry +} + +type VgpuSchedulerGetState struct { + SchedulerPolicy uint32 + ArrMode uint32 + SchedulerParams [8]byte +} + +const sizeofVgpuSchedulerSetParams = unsafe.Sizeof([8]byte{}) + +type VgpuSchedulerSetParams [sizeofVgpuSchedulerSetParams]byte + +type VgpuSchedulerSetState struct { + SchedulerPolicy uint32 + EnableARRMode uint32 + SchedulerParams [8]byte +} + +type VgpuSchedulerCapabilities struct { + SupportedSchedulers [3]uint32 + MaxTimeslice uint32 + MinTimeslice uint32 + IsArrModeSupported uint32 + MaxFrequencyForARR uint32 + MinFrequencyForARR uint32 + MaxAvgFactorForARR uint32 + MinAvgFactorForARR uint32 +} + +type VgpuLicenseExpiry struct { + Year uint32 + Month uint16 + Day uint16 + Hour uint16 + Min uint16 + Sec uint16 + Status uint8 + Pad_cgo_0 [1]byte +} + +type VgpuLicenseInfo struct { + IsLicensed uint8 + LicenseExpiry VgpuLicenseExpiry + CurrentState uint32 +} + +type GridLicenseExpiry struct { + Year uint32 + Month uint16 + Day uint16 + Hour uint16 + Min uint16 + Sec uint16 + Status uint8 + Pad_cgo_0 [1]byte +} + +type GridLicensableFeature struct { + FeatureCode uint32 + FeatureState uint32 + LicenseInfo [128]uint8 + ProductName [128]uint8 + FeatureEnabled uint32 + LicenseExpiry GridLicenseExpiry +} + +type GridLicensableFeatures struct { + IsGridLicenseSupported int32 + LicensableFeaturesCount uint32 + GridLicensableFeatures [3]GridLicensableFeature +} + +type VgpuTypeIdInfo_v1 struct { + Version uint32 + VgpuCount uint32 + VgpuTypeIds *uint32 +} + +type VgpuTypeIdInfo struct { + Version uint32 + VgpuCount uint32 + VgpuTypeIds *uint32 +} + +type VgpuTypeMaxInstance_v1 struct { + Version uint32 + VgpuTypeId uint32 + MaxInstancePerGI uint32 +} + +type VgpuTypeMaxInstance struct { + Version uint32 + VgpuTypeId uint32 + MaxInstancePerGI uint32 +} + +type ActiveVgpuInstanceInfo_v1 struct { + Version uint32 + VgpuCount uint32 + VgpuInstances *uint32 +} + +type ActiveVgpuInstanceInfo struct { + Version uint32 + VgpuCount uint32 + VgpuInstances *uint32 +} + +type VgpuSchedulerState_v1 struct { + Version uint32 + EngineId uint32 + SchedulerPolicy uint32 + EnableARRMode uint32 + SchedulerParams [8]byte +} + +type VgpuSchedulerState struct { + Version uint32 + EngineId uint32 + SchedulerPolicy uint32 + EnableARRMode uint32 + SchedulerParams [8]byte +} + +type VgpuSchedulerStateInfo_v1 struct { + Version uint32 + EngineId uint32 + SchedulerPolicy uint32 + ArrMode uint32 + SchedulerParams [8]byte +} + +type VgpuSchedulerStateInfo struct { + Version uint32 + EngineId uint32 + SchedulerPolicy uint32 + ArrMode uint32 + SchedulerParams [8]byte +} + +type VgpuSchedulerLogInfo_v1 struct { + Version uint32 + EngineId uint32 + SchedulerPolicy uint32 + ArrMode uint32 + SchedulerParams [8]byte + EntriesCount uint32 + LogEntries [200]VgpuSchedulerLogEntry +} + +type VgpuSchedulerLogInfo struct { + Version uint32 + EngineId uint32 + SchedulerPolicy uint32 + ArrMode uint32 + SchedulerParams [8]byte + EntriesCount uint32 + LogEntries [200]VgpuSchedulerLogEntry +} + +type VgpuCreatablePlacementInfo_v1 struct { + Version uint32 + VgpuTypeId uint32 + Count uint32 + PlacementIds *uint32 + PlacementSize uint32 + Pad_cgo_0 [4]byte +} + +type VgpuCreatablePlacementInfo struct { + Version uint32 + VgpuTypeId uint32 + Count uint32 + PlacementIds *uint32 + PlacementSize uint32 + Pad_cgo_0 [4]byte +} + +type NvLinkPowerThres struct { + LowPwrThreshold uint32 +} + +type FieldValue struct { + FieldId uint32 + ScopeId uint32 + Timestamp int64 + LatencyUsec int64 + ValueType uint32 + NvmlReturn uint32 + Value [8]byte +} + +type nvmlUnit struct { + Handle *_Ctype_struct_nvmlUnit_st +} + +type HwbcEntry struct { + HwbcId uint32 + FirmwareVersion [32]uint8 +} + +type LedState struct { + Cause [256]uint8 + Color uint32 +} + +type UnitInfo struct { + Name [96]uint8 + Id [96]uint8 + Serial [96]uint8 + FirmwareVersion [96]uint8 +} + +type PSUInfo struct { + State [256]uint8 + Current uint32 + Voltage uint32 + Power uint32 +} + +type UnitFanInfo struct { + Speed uint32 + State uint32 +} + +type UnitFanSpeeds struct { + Fans [24]UnitFanInfo + Count uint32 +} + +type nvmlEventSet struct { + Handle *_Ctype_struct_nvmlEventSet_st +} + +type nvmlEventData struct { + Device nvmlDevice + EventType uint64 + EventData uint64 + GpuInstanceId uint32 + ComputeInstanceId uint32 +} + +type SystemEventSet struct { + Handle *_Ctype_struct_nvmlSystemEventSet_st +} + +type SystemEventSetCreateRequest_v1 struct { + Version uint32 + Set SystemEventSet +} + +type SystemEventSetCreateRequest struct { + Version uint32 + Set SystemEventSet +} + +type SystemEventSetFreeRequest_v1 struct { + Version uint32 + Set SystemEventSet +} + +type SystemEventSetFreeRequest struct { + Version uint32 + Set SystemEventSet +} + +type SystemRegisterEventRequest_v1 struct { + Version uint32 + EventTypes uint64 + Set SystemEventSet +} + +type SystemRegisterEventRequest struct { + Version uint32 + EventTypes uint64 + Set SystemEventSet +} + +type SystemEventData_v1 struct { + EventType uint64 + GpuId uint32 + Pad_cgo_0 [4]byte +} + +type SystemEventSetWaitRequest_v1 struct { + Version uint32 + Timeoutms uint32 + Set SystemEventSet + Data *SystemEventData_v1 + DataSize uint32 + NumEvent uint32 +} + +type SystemEventSetWaitRequest struct { + Version uint32 + Timeoutms uint32 + Set SystemEventSet + Data *SystemEventData_v1 + DataSize uint32 + NumEvent uint32 +} + +type AccountingStats struct { + GpuUtilization uint32 + MemoryUtilization uint32 + MaxMemoryUsage uint64 + Time uint64 + StartTime uint64 + IsRunning uint32 + Reserved [5]uint32 +} + +type EncoderSessionInfo struct { + SessionId uint32 + Pid uint32 + VgpuInstance uint32 + CodecType uint32 + HResolution uint32 + VResolution uint32 + AverageFps uint32 + AverageLatency uint32 +} + +type FBCStats struct { + SessionsCount uint32 + AverageFPS uint32 + AverageLatency uint32 +} + +type FBCSessionInfo struct { + SessionId uint32 + Pid uint32 + VgpuInstance uint32 + DisplayOrdinal uint32 + SessionType uint32 + SessionFlags uint32 + HMaxResolution uint32 + VMaxResolution uint32 + HResolution uint32 + VResolution uint32 + AverageFPS uint32 + AverageLatency uint32 +} + +type ConfComputeSystemCaps struct { + CpuCaps uint32 + GpusCaps uint32 +} + +type ConfComputeSystemState struct { + Environment uint32 + CcFeature uint32 + DevToolsMode uint32 +} + +type SystemConfComputeSettings_v1 struct { + Version uint32 + Environment uint32 + CcFeature uint32 + DevToolsMode uint32 + MultiGpuMode uint32 +} + +type SystemConfComputeSettings struct { + Version uint32 + Environment uint32 + CcFeature uint32 + DevToolsMode uint32 + MultiGpuMode uint32 +} + +type ConfComputeMemSizeInfo struct { + ProtectedMemSizeKib uint64 + UnprotectedMemSizeKib uint64 +} + +type ConfComputeGpuCertificate struct { + CertChainSize uint32 + AttestationCertChainSize uint32 + CertChain [4096]uint8 + AttestationCertChain [5120]uint8 +} + +type ConfComputeGpuAttestationReport struct { + IsCecAttestationReportPresent uint32 + AttestationReportSize uint32 + CecAttestationReportSize uint32 + Nonce [32]uint8 + AttestationReport [8192]uint8 + CecAttestationReport [4096]uint8 +} + +type ConfComputeSetKeyRotationThresholdInfo_v1 struct { + Version uint32 + MaxAttackerAdvantage uint64 +} + +type ConfComputeSetKeyRotationThresholdInfo struct { + Version uint32 + MaxAttackerAdvantage uint64 +} + +type ConfComputeGetKeyRotationThresholdInfo_v1 struct { + Version uint32 + AttackerAdvantage uint64 +} + +type ConfComputeGetKeyRotationThresholdInfo struct { + Version uint32 + AttackerAdvantage uint64 +} + +type GpuFabricState byte + +type GpuFabricInfo struct { + ClusterUuid [16]uint8 + Status uint32 + CliqueId uint32 + State uint8 + Pad_cgo_0 [3]byte +} + +type GpuFabricInfo_v2 struct { + Version uint32 + ClusterUuid [16]uint8 + Status uint32 + CliqueId uint32 + State uint8 + HealthMask uint32 +} + +type GpuFabricInfo_v3 struct { + Version uint32 + ClusterUuid [16]uint8 + Status uint32 + CliqueId uint32 + State uint8 + HealthMask uint32 + HealthSummary uint8 + Pad_cgo_0 [3]byte +} + +type GpuFabricInfoV struct { + Version uint32 + ClusterUuid [16]uint8 + Status uint32 + CliqueId uint32 + State uint8 + HealthMask uint32 + HealthSummary uint8 + Pad_cgo_0 [3]byte +} + +type SystemDriverBranchInfo_v1 struct { + Version uint32 + Branch [80]uint8 +} + +type SystemDriverBranchInfo struct { + Version uint32 + Branch [80]uint8 +} + +type AffinityScope uint32 + +type Temperature_v1 struct { + Version uint32 + SensorType uint32 + Temperature int32 +} + +type Temperature struct { + Version uint32 + SensorType uint32 + Temperature int32 +} + +type NvlinkSupportedBwModes_v1 struct { + Version uint32 + BwModes [23]uint8 + TotalBwModes uint8 +} + +type NvlinkSupportedBwModes struct { + Version uint32 + BwModes [23]uint8 + TotalBwModes uint8 +} + +type NvlinkGetBwMode_v1 struct { + Version uint32 + BIsBest uint32 + BwMode uint8 + Pad_cgo_0 [3]byte +} + +type NvlinkGetBwMode struct { + Version uint32 + BIsBest uint32 + BwMode uint8 + Pad_cgo_0 [3]byte +} + +type NvlinkSetBwMode_v1 struct { + Version uint32 + BSetBest uint32 + BwMode uint8 + Pad_cgo_0 [3]byte +} + +type NvlinkSetBwMode struct { + Version uint32 + BSetBest uint32 + BwMode uint8 + Pad_cgo_0 [3]byte +} + +type NvLinkInfo_v1 struct { + Version uint32 + IsNvleEnabled uint32 +} + +type NvlinkFirmwareVersion struct { + UcodeType uint8 + Major uint32 + Minor uint32 + SubMinor uint32 +} + +type NvlinkFirmwareInfo struct { + FirmwareVersion [100]NvlinkFirmwareVersion + NumValidEntries uint32 +} + +type NvLinkInfo_v2 struct { + Version uint32 + IsNvleEnabled uint32 + FirmwareInfo NvlinkFirmwareInfo +} + +type NvLinkInfo struct { + Version uint32 + IsNvleEnabled uint32 + FirmwareInfo NvlinkFirmwareInfo +} + +type VgpuVersion struct { + MinVersion uint32 + MaxVersion uint32 +} + +type nvmlVgpuMetadata struct { + Version uint32 + Revision uint32 + GuestInfoState uint32 + GuestDriverVersion [80]uint8 + HostDriverVersion [80]uint8 + Reserved [6]uint32 + VgpuVirtualizationCaps uint32 + GuestVgpuVersion uint32 + OpaqueDataSize uint32 + OpaqueData [4]uint8 +} + +type nvmlVgpuPgpuMetadata struct { + Version uint32 + Revision uint32 + HostDriverVersion [80]uint8 + PgpuVirtualizationCaps uint32 + Reserved [5]uint32 + HostSupportedVgpuRange VgpuVersion + OpaqueDataSize uint32 + OpaqueData [4]uint8 +} + +type VgpuPgpuCompatibility struct { + VgpuVmCompatibility uint32 + CompatibilityLimitCode uint32 +} + +type ExcludedDeviceInfo struct { + PciInfo PciInfo + Uuid [80]uint8 +} + +type PRMTLV_v1 struct { + DataSize uint32 + Status uint32 + InData [496]uint8 +} + +type GpuInstancePlacement struct { + Start uint32 + Size uint32 +} + +type GpuInstanceProfileInfo struct { + Id uint32 + IsP2pSupported uint32 + SliceCount uint32 + InstanceCount uint32 + MultiprocessorCount uint32 + CopyEngineCount uint32 + DecoderCount uint32 + EncoderCount uint32 + JpegCount uint32 + OfaCount uint32 + MemorySizeMB uint64 +} + +type GpuInstanceProfileInfo_v2 struct { + Version uint32 + Id uint32 + IsP2pSupported uint32 + SliceCount uint32 + InstanceCount uint32 + MultiprocessorCount uint32 + CopyEngineCount uint32 + DecoderCount uint32 + EncoderCount uint32 + JpegCount uint32 + OfaCount uint32 + MemorySizeMB uint64 + Name [96]uint8 +} + +type GpuInstanceProfileInfo_v3 struct { + Version uint32 + Id uint32 + SliceCount uint32 + InstanceCount uint32 + MultiprocessorCount uint32 + CopyEngineCount uint32 + DecoderCount uint32 + EncoderCount uint32 + JpegCount uint32 + OfaCount uint32 + MemorySizeMB uint64 + Name [96]uint8 + Capabilities uint32 + Pad_cgo_0 [4]byte +} + +type nvmlGpuInstanceInfo struct { + Device nvmlDevice + Id uint32 + ProfileId uint32 + Placement GpuInstancePlacement +} + +type ComputeInstancePlacement struct { + Start uint32 + Size uint32 +} + +type ComputeInstanceProfileInfo struct { + Id uint32 + SliceCount uint32 + InstanceCount uint32 + MultiprocessorCount uint32 + SharedCopyEngineCount uint32 + SharedDecoderCount uint32 + SharedEncoderCount uint32 + SharedJpegCount uint32 + SharedOfaCount uint32 +} + +type ComputeInstanceProfileInfo_v2 struct { + Version uint32 + Id uint32 + SliceCount uint32 + InstanceCount uint32 + MultiprocessorCount uint32 + SharedCopyEngineCount uint32 + SharedDecoderCount uint32 + SharedEncoderCount uint32 + SharedJpegCount uint32 + SharedOfaCount uint32 + Name [96]uint8 +} + +type ComputeInstanceProfileInfo_v3 struct { + Version uint32 + Id uint32 + SliceCount uint32 + InstanceCount uint32 + MultiprocessorCount uint32 + SharedCopyEngineCount uint32 + SharedDecoderCount uint32 + SharedEncoderCount uint32 + SharedJpegCount uint32 + SharedOfaCount uint32 + Name [96]uint8 + Capabilities uint32 +} + +type nvmlComputeInstanceInfo struct { + Device nvmlDevice + GpuInstance nvmlGpuInstance + Id uint32 + ProfileId uint32 + Placement ComputeInstancePlacement +} + +type nvmlComputeInstance struct { + Handle *_Ctype_struct_nvmlComputeInstance_st +} + +type nvmlGpmSample struct { + Handle *_Ctype_struct_nvmlGpmSample_st +} + +type GpmMetric struct { + MetricId uint32 + NvmlReturn uint32 + Value float64 + MetricInfo _Ctype_struct___19 +} + +type nvmlGpmMetricsGetType struct { + Version uint32 + NumMetrics uint32 + Sample1 nvmlGpmSample + Sample2 nvmlGpmSample + Metrics [210]GpmMetric +} + +type GpmSupport struct { + Version uint32 + IsSupportedDevice uint32 +} + +type DeviceCapabilities_v1 struct { + Version uint32 + CapMask uint32 +} + +type DeviceCapabilities struct { + Version uint32 + CapMask uint32 +} + +type Mask255 struct { + Mask [8]uint32 +} + +type WorkloadPowerProfileInfo_v1 struct { + Version uint32 + ProfileId uint32 + Priority uint32 + ConflictingMask Mask255 +} + +type WorkloadPowerProfileInfo struct { + Version uint32 + ProfileId uint32 + Priority uint32 + ConflictingMask Mask255 +} + +type WorkloadPowerProfileProfilesInfo_v1 struct { + Version uint32 + PerfProfilesMask Mask255 + PerfProfile [255]WorkloadPowerProfileInfo +} + +type WorkloadPowerProfileProfilesInfo struct { + Version uint32 + PerfProfilesMask Mask255 + PerfProfile [255]WorkloadPowerProfileInfo +} + +type WorkloadPowerProfileCurrentProfiles_v1 struct { + Version uint32 + PerfProfilesMask Mask255 + RequestedProfilesMask Mask255 + EnforcedProfilesMask Mask255 +} + +type WorkloadPowerProfileCurrentProfiles struct { + Version uint32 + PerfProfilesMask Mask255 + RequestedProfilesMask Mask255 + EnforcedProfilesMask Mask255 +} + +type WorkloadPowerProfileRequestedProfiles_v1 struct { + Version uint32 + RequestedProfilesMask Mask255 +} + +type WorkloadPowerProfileRequestedProfiles struct { + Version uint32 + RequestedProfilesMask Mask255 +} + +type PowerSmoothingProfile_v1 struct { + Version uint32 + ProfileId uint32 + ParamId uint32 + Value float64 +} + +type PowerSmoothingProfile struct { + Version uint32 + ProfileId uint32 + ParamId uint32 + Value float64 +} + +type PowerSmoothingState_v1 struct { + Version uint32 + State uint32 +} + +type PowerSmoothingState struct { + Version uint32 + State uint32 +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/unit.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/unit.go new file mode 100644 index 00000000..617ad546 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/unit.go @@ -0,0 +1,113 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +// nvml.UnitGetCount() +func (l *library) UnitGetCount() (int, Return) { + var UnitCount uint32 + ret := nvmlUnitGetCount(&UnitCount) + return int(UnitCount), ret +} + +// nvml.UnitGetHandleByIndex() +func (l *library) UnitGetHandleByIndex(index int) (Unit, Return) { + var unit nvmlUnit + ret := nvmlUnitGetHandleByIndex(uint32(index), &unit) + return unit, ret +} + +// nvml.UnitGetUnitInfo() +func (l *library) UnitGetUnitInfo(unit Unit) (UnitInfo, Return) { + return unit.GetUnitInfo() +} + +func (unit nvmlUnit) GetUnitInfo() (UnitInfo, Return) { + var info UnitInfo + ret := nvmlUnitGetUnitInfo(unit, &info) + return info, ret +} + +// nvml.UnitGetLedState() +func (l *library) UnitGetLedState(unit Unit) (LedState, Return) { + return unit.GetLedState() +} + +func (unit nvmlUnit) GetLedState() (LedState, Return) { + var state LedState + ret := nvmlUnitGetLedState(unit, &state) + return state, ret +} + +// nvml.UnitGetPsuInfo() +func (l *library) UnitGetPsuInfo(unit Unit) (PSUInfo, Return) { + return unit.GetPsuInfo() +} + +func (unit nvmlUnit) GetPsuInfo() (PSUInfo, Return) { + var psu PSUInfo + ret := nvmlUnitGetPsuInfo(unit, &psu) + return psu, ret +} + +// nvml.UnitGetTemperature() +func (l *library) UnitGetTemperature(unit Unit, ttype int) (uint32, Return) { + return unit.GetTemperature(ttype) +} + +func (unit nvmlUnit) GetTemperature(ttype int) (uint32, Return) { + var temp uint32 + ret := nvmlUnitGetTemperature(unit, uint32(ttype), &temp) + return temp, ret +} + +// nvml.UnitGetFanSpeedInfo() +func (l *library) UnitGetFanSpeedInfo(unit Unit) (UnitFanSpeeds, Return) { + return unit.GetFanSpeedInfo() +} + +func (unit nvmlUnit) GetFanSpeedInfo() (UnitFanSpeeds, Return) { + var fanSpeeds UnitFanSpeeds + ret := nvmlUnitGetFanSpeedInfo(unit, &fanSpeeds) + return fanSpeeds, ret +} + +// nvml.UnitGetDevices() +func (l *library) UnitGetDevices(unit Unit) ([]Device, Return) { + return unit.GetDevices() +} + +func (unit nvmlUnit) GetDevices() ([]Device, Return) { + var deviceCount uint32 = 1 // Will be reduced upon returning + for { + devices := make([]nvmlDevice, deviceCount) + ret := nvmlUnitGetDevices(unit, &deviceCount, &devices[0]) + if ret == SUCCESS { + return convertSlice[nvmlDevice, Device](devices[:deviceCount]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + deviceCount *= 2 + } +} + +// nvml.UnitSetLedState() +func (l *library) UnitSetLedState(unit Unit, color LedColor) Return { + return unit.SetLedState(color) +} + +func (unit nvmlUnit) SetLedState(color LedColor) Return { + return nvmlUnitSetLedState(unit, color) +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/vgpu.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/vgpu.go new file mode 100644 index 00000000..9ab649a4 --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/vgpu.go @@ -0,0 +1,511 @@ +// Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package nvml + +import ( + "unsafe" +) + +// nvml.VgpuMetadata +type VgpuMetadata struct { + nvmlVgpuMetadata + OpaqueData []byte +} + +// nvml.VgpuPgpuMetadata +type VgpuPgpuMetadata struct { + nvmlVgpuPgpuMetadata + OpaqueData []byte +} + +// nvml.VgpuTypeGetClass() +func (l *library) VgpuTypeGetClass(vgpuTypeId VgpuTypeId) (string, Return) { + return vgpuTypeId.GetClass() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetClass() (string, Return) { + var size uint32 = DEVICE_NAME_BUFFER_SIZE + vgpuTypeClass := make([]byte, DEVICE_NAME_BUFFER_SIZE) + ret := nvmlVgpuTypeGetClass(vgpuTypeId, &vgpuTypeClass[0], &size) + return string(vgpuTypeClass[:clen(vgpuTypeClass)]), ret +} + +// nvml.VgpuTypeGetName() +func (l *library) VgpuTypeGetName(vgpuTypeId VgpuTypeId) (string, Return) { + return vgpuTypeId.GetName() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetName() (string, Return) { + var size uint32 = DEVICE_NAME_BUFFER_SIZE + vgpuTypeName := make([]byte, DEVICE_NAME_BUFFER_SIZE) + ret := nvmlVgpuTypeGetName(vgpuTypeId, &vgpuTypeName[0], &size) + return string(vgpuTypeName[:clen(vgpuTypeName)]), ret +} + +// nvml.VgpuTypeGetGpuInstanceProfileId() +func (l *library) VgpuTypeGetGpuInstanceProfileId(vgpuTypeId VgpuTypeId) (uint32, Return) { + return vgpuTypeId.GetGpuInstanceProfileId() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetGpuInstanceProfileId() (uint32, Return) { + var size uint32 + ret := nvmlVgpuTypeGetGpuInstanceProfileId(vgpuTypeId, &size) + return size, ret +} + +// nvml.VgpuTypeGetDeviceID() +func (l *library) VgpuTypeGetDeviceID(vgpuTypeId VgpuTypeId) (uint64, uint64, Return) { + return vgpuTypeId.GetDeviceID() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetDeviceID() (uint64, uint64, Return) { + var deviceID, subsystemID uint64 + ret := nvmlVgpuTypeGetDeviceID(vgpuTypeId, &deviceID, &subsystemID) + return deviceID, subsystemID, ret +} + +// nvml.VgpuTypeGetFramebufferSize() +func (l *library) VgpuTypeGetFramebufferSize(vgpuTypeId VgpuTypeId) (uint64, Return) { + return vgpuTypeId.GetFramebufferSize() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetFramebufferSize() (uint64, Return) { + var fbSize uint64 + ret := nvmlVgpuTypeGetFramebufferSize(vgpuTypeId, &fbSize) + return fbSize, ret +} + +// nvml.VgpuTypeGetNumDisplayHeads() +func (l *library) VgpuTypeGetNumDisplayHeads(vgpuTypeId VgpuTypeId) (int, Return) { + return vgpuTypeId.GetNumDisplayHeads() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetNumDisplayHeads() (int, Return) { + var numDisplayHeads uint32 + ret := nvmlVgpuTypeGetNumDisplayHeads(vgpuTypeId, &numDisplayHeads) + return int(numDisplayHeads), ret +} + +// nvml.VgpuTypeGetResolution() +func (l *library) VgpuTypeGetResolution(vgpuTypeId VgpuTypeId, displayIndex int) (uint32, uint32, Return) { + return vgpuTypeId.GetResolution(displayIndex) +} + +func (vgpuTypeId nvmlVgpuTypeId) GetResolution(displayIndex int) (uint32, uint32, Return) { + var xdim, ydim uint32 + ret := nvmlVgpuTypeGetResolution(vgpuTypeId, uint32(displayIndex), &xdim, &ydim) + return xdim, ydim, ret +} + +// nvml.VgpuTypeGetLicense() +func (l *library) VgpuTypeGetLicense(vgpuTypeId VgpuTypeId) (string, Return) { + return vgpuTypeId.GetLicense() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetLicense() (string, Return) { + vgpuTypeLicenseString := make([]byte, GRID_LICENSE_BUFFER_SIZE) + ret := nvmlVgpuTypeGetLicense(vgpuTypeId, &vgpuTypeLicenseString[0], GRID_LICENSE_BUFFER_SIZE) + return string(vgpuTypeLicenseString[:clen(vgpuTypeLicenseString)]), ret +} + +// nvml.VgpuTypeGetFrameRateLimit() +func (l *library) VgpuTypeGetFrameRateLimit(vgpuTypeId VgpuTypeId) (uint32, Return) { + return vgpuTypeId.GetFrameRateLimit() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetFrameRateLimit() (uint32, Return) { + var frameRateLimit uint32 + ret := nvmlVgpuTypeGetFrameRateLimit(vgpuTypeId, &frameRateLimit) + return frameRateLimit, ret +} + +// nvml.VgpuTypeGetMaxInstances() +func (l *library) VgpuTypeGetMaxInstances(device Device, vgpuTypeId VgpuTypeId) (int, Return) { + return vgpuTypeId.GetMaxInstances(device) +} + +func (device nvmlDevice) VgpuTypeGetMaxInstances(vgpuTypeId VgpuTypeId) (int, Return) { + return vgpuTypeId.GetMaxInstances(device) +} + +func (vgpuTypeId nvmlVgpuTypeId) GetMaxInstances(device Device) (int, Return) { + var vgpuInstanceCount uint32 + ret := nvmlVgpuTypeGetMaxInstances(nvmlDeviceHandle(device), vgpuTypeId, &vgpuInstanceCount) + return int(vgpuInstanceCount), ret +} + +// nvml.VgpuTypeGetMaxInstancesPerVm() +func (l *library) VgpuTypeGetMaxInstancesPerVm(vgpuTypeId VgpuTypeId) (int, Return) { + return vgpuTypeId.GetMaxInstancesPerVm() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetMaxInstancesPerVm() (int, Return) { + var vgpuInstanceCountPerVm uint32 + ret := nvmlVgpuTypeGetMaxInstancesPerVm(vgpuTypeId, &vgpuInstanceCountPerVm) + return int(vgpuInstanceCountPerVm), ret +} + +// nvml.VgpuInstanceGetVmID() +func (l *library) VgpuInstanceGetVmID(vgpuInstance VgpuInstance) (string, VgpuVmIdType, Return) { + return vgpuInstance.GetVmID() +} + +func (vgpuInstance nvmlVgpuInstance) GetVmID() (string, VgpuVmIdType, Return) { + var vmIdType VgpuVmIdType + vmId := make([]byte, DEVICE_UUID_BUFFER_SIZE) + ret := nvmlVgpuInstanceGetVmID(vgpuInstance, &vmId[0], DEVICE_UUID_BUFFER_SIZE, &vmIdType) + return string(vmId[:clen(vmId)]), vmIdType, ret +} + +// nvml.VgpuInstanceGetUUID() +func (l *library) VgpuInstanceGetUUID(vgpuInstance VgpuInstance) (string, Return) { + return vgpuInstance.GetUUID() +} + +func (vgpuInstance nvmlVgpuInstance) GetUUID() (string, Return) { + uuid := make([]byte, DEVICE_UUID_BUFFER_SIZE) + ret := nvmlVgpuInstanceGetUUID(vgpuInstance, &uuid[0], DEVICE_UUID_BUFFER_SIZE) + return string(uuid[:clen(uuid)]), ret +} + +// nvml.VgpuInstanceGetVmDriverVersion() +func (l *library) VgpuInstanceGetVmDriverVersion(vgpuInstance VgpuInstance) (string, Return) { + return vgpuInstance.GetVmDriverVersion() +} + +func (vgpuInstance nvmlVgpuInstance) GetVmDriverVersion() (string, Return) { + version := make([]byte, SYSTEM_DRIVER_VERSION_BUFFER_SIZE) + ret := nvmlVgpuInstanceGetVmDriverVersion(vgpuInstance, &version[0], SYSTEM_DRIVER_VERSION_BUFFER_SIZE) + return string(version[:clen(version)]), ret +} + +// nvml.VgpuInstanceGetFbUsage() +func (l *library) VgpuInstanceGetFbUsage(vgpuInstance VgpuInstance) (uint64, Return) { + return vgpuInstance.GetFbUsage() +} + +func (vgpuInstance nvmlVgpuInstance) GetFbUsage() (uint64, Return) { + var fbUsage uint64 + ret := nvmlVgpuInstanceGetFbUsage(vgpuInstance, &fbUsage) + return fbUsage, ret +} + +// nvml.VgpuInstanceGetLicenseInfo() +func (l *library) VgpuInstanceGetLicenseInfo(vgpuInstance VgpuInstance) (VgpuLicenseInfo, Return) { + return vgpuInstance.GetLicenseInfo() +} + +func (vgpuInstance nvmlVgpuInstance) GetLicenseInfo() (VgpuLicenseInfo, Return) { + var licenseInfo VgpuLicenseInfo + ret := nvmlVgpuInstanceGetLicenseInfo(vgpuInstance, &licenseInfo) + return licenseInfo, ret +} + +// nvml.VgpuInstanceGetLicenseStatus() +// +// Deprecated: Use VgpuInstanceGetLicenseInfo instead. +func (l *library) VgpuInstanceGetLicenseStatus(vgpuInstance VgpuInstance) (int, Return) { + return vgpuInstance.GetLicenseStatus() +} + +func (vgpuInstance nvmlVgpuInstance) GetLicenseStatus() (int, Return) { + var licensed uint32 + ret := nvmlVgpuInstanceGetLicenseStatus(vgpuInstance, &licensed) + return int(licensed), ret +} + +// nvml.VgpuInstanceGetType() +func (l *library) VgpuInstanceGetType(vgpuInstance VgpuInstance) (VgpuTypeId, Return) { + return vgpuInstance.GetType() +} + +func (vgpuInstance nvmlVgpuInstance) GetType() (VgpuTypeId, Return) { + var vgpuTypeId nvmlVgpuTypeId + ret := nvmlVgpuInstanceGetType(vgpuInstance, &vgpuTypeId) + return vgpuTypeId, ret +} + +// nvml.VgpuInstanceGetFrameRateLimit() +func (l *library) VgpuInstanceGetFrameRateLimit(vgpuInstance VgpuInstance) (uint32, Return) { + return vgpuInstance.GetFrameRateLimit() +} + +func (vgpuInstance nvmlVgpuInstance) GetFrameRateLimit() (uint32, Return) { + var frameRateLimit uint32 + ret := nvmlVgpuInstanceGetFrameRateLimit(vgpuInstance, &frameRateLimit) + return frameRateLimit, ret +} + +// nvml.VgpuInstanceGetEccMode() +func (l *library) VgpuInstanceGetEccMode(vgpuInstance VgpuInstance) (EnableState, Return) { + return vgpuInstance.GetEccMode() +} + +func (vgpuInstance nvmlVgpuInstance) GetEccMode() (EnableState, Return) { + var eccMode EnableState + ret := nvmlVgpuInstanceGetEccMode(vgpuInstance, &eccMode) + return eccMode, ret +} + +// nvml.VgpuInstanceGetEncoderCapacity() +func (l *library) VgpuInstanceGetEncoderCapacity(vgpuInstance VgpuInstance) (int, Return) { + return vgpuInstance.GetEncoderCapacity() +} + +func (vgpuInstance nvmlVgpuInstance) GetEncoderCapacity() (int, Return) { + var encoderCapacity uint32 + ret := nvmlVgpuInstanceGetEncoderCapacity(vgpuInstance, &encoderCapacity) + return int(encoderCapacity), ret +} + +// nvml.VgpuInstanceSetEncoderCapacity() +func (l *library) VgpuInstanceSetEncoderCapacity(vgpuInstance VgpuInstance, encoderCapacity int) Return { + return vgpuInstance.SetEncoderCapacity(encoderCapacity) +} + +func (vgpuInstance nvmlVgpuInstance) SetEncoderCapacity(encoderCapacity int) Return { + return nvmlVgpuInstanceSetEncoderCapacity(vgpuInstance, uint32(encoderCapacity)) +} + +// nvml.VgpuInstanceGetEncoderStats() +func (l *library) VgpuInstanceGetEncoderStats(vgpuInstance VgpuInstance) (int, uint32, uint32, Return) { + return vgpuInstance.GetEncoderStats() +} + +func (vgpuInstance nvmlVgpuInstance) GetEncoderStats() (int, uint32, uint32, Return) { + var sessionCount, averageFps, averageLatency uint32 + ret := nvmlVgpuInstanceGetEncoderStats(vgpuInstance, &sessionCount, &averageFps, &averageLatency) + return int(sessionCount), averageFps, averageLatency, ret +} + +// nvml.VgpuInstanceGetEncoderSessions() +func (l *library) VgpuInstanceGetEncoderSessions(vgpuInstance VgpuInstance) (int, EncoderSessionInfo, Return) { + return vgpuInstance.GetEncoderSessions() +} + +func (vgpuInstance nvmlVgpuInstance) GetEncoderSessions() (int, EncoderSessionInfo, Return) { + var sessionCount uint32 + var sessionInfo EncoderSessionInfo + ret := nvmlVgpuInstanceGetEncoderSessions(vgpuInstance, &sessionCount, &sessionInfo) + return int(sessionCount), sessionInfo, ret +} + +// nvml.VgpuInstanceGetFBCStats() +func (l *library) VgpuInstanceGetFBCStats(vgpuInstance VgpuInstance) (FBCStats, Return) { + return vgpuInstance.GetFBCStats() +} + +func (vgpuInstance nvmlVgpuInstance) GetFBCStats() (FBCStats, Return) { + var fbcStats FBCStats + ret := nvmlVgpuInstanceGetFBCStats(vgpuInstance, &fbcStats) + return fbcStats, ret +} + +// nvml.VgpuInstanceGetFBCSessions() +func (l *library) VgpuInstanceGetFBCSessions(vgpuInstance VgpuInstance) (int, FBCSessionInfo, Return) { + return vgpuInstance.GetFBCSessions() +} + +func (vgpuInstance nvmlVgpuInstance) GetFBCSessions() (int, FBCSessionInfo, Return) { + var sessionCount uint32 + var sessionInfo FBCSessionInfo + ret := nvmlVgpuInstanceGetFBCSessions(vgpuInstance, &sessionCount, &sessionInfo) + return int(sessionCount), sessionInfo, ret +} + +// nvml.VgpuInstanceGetGpuInstanceId() +func (l *library) VgpuInstanceGetGpuInstanceId(vgpuInstance VgpuInstance) (int, Return) { + return vgpuInstance.GetGpuInstanceId() +} + +func (vgpuInstance nvmlVgpuInstance) GetGpuInstanceId() (int, Return) { + var gpuInstanceId uint32 + ret := nvmlVgpuInstanceGetGpuInstanceId(vgpuInstance, &gpuInstanceId) + return int(gpuInstanceId), ret +} + +// nvml.VgpuInstanceGetGpuPciId() +func (l *library) VgpuInstanceGetGpuPciId(vgpuInstance VgpuInstance) (string, Return) { + return vgpuInstance.GetGpuPciId() +} + +func (vgpuInstance nvmlVgpuInstance) GetGpuPciId() (string, Return) { + var length uint32 = 1 // Will be reduced upon returning + for { + vgpuPciId := make([]byte, length) + ret := nvmlVgpuInstanceGetGpuPciId(vgpuInstance, &vgpuPciId[0], &length) + if ret == SUCCESS { + return string(vgpuPciId[:clen(vgpuPciId)]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return "", ret + } + length *= 2 + } +} + +// nvml.VgpuInstanceGetMetadata() +func (l *library) VgpuInstanceGetMetadata(vgpuInstance VgpuInstance) (VgpuMetadata, Return) { + return vgpuInstance.GetMetadata() +} + +func (vgpuInstance nvmlVgpuInstance) GetMetadata() (VgpuMetadata, Return) { + var vgpuMetadata VgpuMetadata + opaqueDataSize := unsafe.Sizeof(vgpuMetadata.nvmlVgpuMetadata.OpaqueData) + vgpuMetadataSize := unsafe.Sizeof(vgpuMetadata.nvmlVgpuMetadata) - opaqueDataSize + for { + bufferSize := uint32(vgpuMetadataSize + opaqueDataSize) + buffer := make([]byte, bufferSize) + nvmlVgpuMetadataPtr := (*nvmlVgpuMetadata)(unsafe.Pointer(&buffer[0])) + ret := nvmlVgpuInstanceGetMetadata(vgpuInstance, nvmlVgpuMetadataPtr, &bufferSize) + if ret == SUCCESS { + vgpuMetadata.nvmlVgpuMetadata = *nvmlVgpuMetadataPtr + vgpuMetadata.OpaqueData = buffer[vgpuMetadataSize:bufferSize] + return vgpuMetadata, ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return vgpuMetadata, ret + } + opaqueDataSize = 2 * opaqueDataSize + } +} + +// nvml.VgpuInstanceGetAccountingMode() +func (l *library) VgpuInstanceGetAccountingMode(vgpuInstance VgpuInstance) (EnableState, Return) { + return vgpuInstance.GetAccountingMode() +} + +func (vgpuInstance nvmlVgpuInstance) GetAccountingMode() (EnableState, Return) { + var mode EnableState + ret := nvmlVgpuInstanceGetAccountingMode(vgpuInstance, &mode) + return mode, ret +} + +// nvml.VgpuInstanceGetAccountingPids() +func (l *library) VgpuInstanceGetAccountingPids(vgpuInstance VgpuInstance) ([]int, Return) { + return vgpuInstance.GetAccountingPids() +} + +func (vgpuInstance nvmlVgpuInstance) GetAccountingPids() ([]int, Return) { + var count uint32 = 1 // Will be reduced upon returning + for { + pids := make([]uint32, count) + ret := nvmlVgpuInstanceGetAccountingPids(vgpuInstance, &count, &pids[0]) + if ret == SUCCESS { + return uint32SliceToIntSlice(pids[:count]), ret + } + if ret != ERROR_INSUFFICIENT_SIZE { + return nil, ret + } + count *= 2 + } +} + +// nvml.VgpuInstanceGetAccountingStats() +func (l *library) VgpuInstanceGetAccountingStats(vgpuInstance VgpuInstance, pid int) (AccountingStats, Return) { + return vgpuInstance.GetAccountingStats(pid) +} + +func (vgpuInstance nvmlVgpuInstance) GetAccountingStats(pid int) (AccountingStats, Return) { + var stats AccountingStats + ret := nvmlVgpuInstanceGetAccountingStats(vgpuInstance, uint32(pid), &stats) + return stats, ret +} + +// nvml.GetVgpuCompatibility() +func (l *library) GetVgpuCompatibility(vgpuMetadata *VgpuMetadata, pgpuMetadata *VgpuPgpuMetadata) (VgpuPgpuCompatibility, Return) { + var compatibilityInfo VgpuPgpuCompatibility + ret := nvmlGetVgpuCompatibility(&vgpuMetadata.nvmlVgpuMetadata, &pgpuMetadata.nvmlVgpuPgpuMetadata, &compatibilityInfo) + return compatibilityInfo, ret +} + +// nvml.GetVgpuVersion() +func (l *library) GetVgpuVersion() (VgpuVersion, VgpuVersion, Return) { + var supported, current VgpuVersion + ret := nvmlGetVgpuVersion(&supported, ¤t) + return supported, current, ret +} + +// nvml.SetVgpuVersion() +func (l *library) SetVgpuVersion(vgpuVersion *VgpuVersion) Return { + return nvmlSetVgpuVersion(vgpuVersion) +} + +// nvml.VgpuInstanceClearAccountingPids() +func (l *library) VgpuInstanceClearAccountingPids(vgpuInstance VgpuInstance) Return { + return vgpuInstance.ClearAccountingPids() +} + +func (vgpuInstance nvmlVgpuInstance) ClearAccountingPids() Return { + return nvmlVgpuInstanceClearAccountingPids(vgpuInstance) +} + +// nvml.VgpuInstanceGetMdevUUID() +func (l *library) VgpuInstanceGetMdevUUID(vgpuInstance VgpuInstance) (string, Return) { + return vgpuInstance.GetMdevUUID() +} + +func (vgpuInstance nvmlVgpuInstance) GetMdevUUID() (string, Return) { + mdevUUID := make([]byte, DEVICE_UUID_BUFFER_SIZE) + ret := nvmlVgpuInstanceGetMdevUUID(vgpuInstance, &mdevUUID[0], DEVICE_UUID_BUFFER_SIZE) + return string(mdevUUID[:clen(mdevUUID)]), ret +} + +// nvml.VgpuTypeGetCapabilities() +func (l *library) VgpuTypeGetCapabilities(vgpuTypeId VgpuTypeId, capability VgpuCapability) (bool, Return) { + return vgpuTypeId.GetCapabilities(capability) +} + +func (vgpuTypeId nvmlVgpuTypeId) GetCapabilities(capability VgpuCapability) (bool, Return) { + var capResult uint32 + ret := nvmlVgpuTypeGetCapabilities(vgpuTypeId, capability, &capResult) + return (capResult != 0), ret +} + +// nvml.GetVgpuDriverCapabilities() +func (l *library) GetVgpuDriverCapabilities(capability VgpuDriverCapability) (bool, Return) { + var capResult uint32 + ret := nvmlGetVgpuDriverCapabilities(capability, &capResult) + return (capResult != 0), ret +} + +// nvml.VgpuTypeGetBAR1Info() +func (l *library) VgpuTypeGetBAR1Info(vgpuTypeId VgpuTypeId) (VgpuTypeBar1Info, Return) { + return vgpuTypeId.GetBAR1Info() +} + +func (vgpuTypeId nvmlVgpuTypeId) GetBAR1Info() (VgpuTypeBar1Info, Return) { + var bar1Info VgpuTypeBar1Info + bar1Info.Version = STRUCT_VERSION(bar1Info, 1) + ret := nvmlVgpuTypeGetBAR1Info(vgpuTypeId, &bar1Info) + return bar1Info, ret +} + +// nvml.VgpuInstanceGetRuntimeStateSize() +func (l *library) VgpuInstanceGetRuntimeStateSize(vgpuInstance VgpuInstance) (VgpuRuntimeState, Return) { + return vgpuInstance.GetRuntimeStateSize() +} + +func (vgpuInstance nvmlVgpuInstance) GetRuntimeStateSize() (VgpuRuntimeState, Return) { + var pState VgpuRuntimeState + pState.Version = STRUCT_VERSION(pState, 1) + ret := nvmlVgpuInstanceGetRuntimeStateSize(vgpuInstance, &pState) + return pState, ret +} + +// nvml.VgpuTypeGetMaxInstancesPerGpuInstance() +func (l *library) VgpuTypeGetMaxInstancesPerGpuInstance(maxInstance *VgpuTypeMaxInstance) Return { + return nvmlVgpuTypeGetMaxInstancesPerGpuInstance(maxInstance) +} diff --git a/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/zz_generated.api.go b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/zz_generated.api.go new file mode 100644 index 00000000..f4930ffd --- /dev/null +++ b/vendor/github.com/NVIDIA/go-nvml/pkg/nvml/zz_generated.api.go @@ -0,0 +1,1154 @@ +/** +# Copyright 2024 NVIDIA CORPORATION +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +**/ + +// Generated Code; DO NOT EDIT. + +package nvml + +// The variables below represent package level methods from the library type. +var ( + ComputeInstanceDestroy = libnvml.ComputeInstanceDestroy + ComputeInstanceGetInfo = libnvml.ComputeInstanceGetInfo + DeviceClearAccountingPids = libnvml.DeviceClearAccountingPids + DeviceClearCpuAffinity = libnvml.DeviceClearCpuAffinity + DeviceClearEccErrorCounts = libnvml.DeviceClearEccErrorCounts + DeviceClearFieldValues = libnvml.DeviceClearFieldValues + DeviceCreateGpuInstance = libnvml.DeviceCreateGpuInstance + DeviceCreateGpuInstanceWithPlacement = libnvml.DeviceCreateGpuInstanceWithPlacement + DeviceDiscoverGpus = libnvml.DeviceDiscoverGpus + DeviceFreezeNvLinkUtilizationCounter = libnvml.DeviceFreezeNvLinkUtilizationCounter + DeviceGetAPIRestriction = libnvml.DeviceGetAPIRestriction + DeviceGetAccountingBufferSize = libnvml.DeviceGetAccountingBufferSize + DeviceGetAccountingMode = libnvml.DeviceGetAccountingMode + DeviceGetAccountingPids = libnvml.DeviceGetAccountingPids + DeviceGetAccountingStats = libnvml.DeviceGetAccountingStats + DeviceGetActiveVgpus = libnvml.DeviceGetActiveVgpus + DeviceGetAdaptiveClockInfoStatus = libnvml.DeviceGetAdaptiveClockInfoStatus + DeviceGetAddressingMode = libnvml.DeviceGetAddressingMode + DeviceGetApplicationsClock = libnvml.DeviceGetApplicationsClock + DeviceGetArchitecture = libnvml.DeviceGetArchitecture + DeviceGetAttributes = libnvml.DeviceGetAttributes + DeviceGetAutoBoostedClocksEnabled = libnvml.DeviceGetAutoBoostedClocksEnabled + DeviceGetBAR1MemoryInfo = libnvml.DeviceGetBAR1MemoryInfo + DeviceGetBoardId = libnvml.DeviceGetBoardId + DeviceGetBoardPartNumber = libnvml.DeviceGetBoardPartNumber + DeviceGetBrand = libnvml.DeviceGetBrand + DeviceGetBridgeChipInfo = libnvml.DeviceGetBridgeChipInfo + DeviceGetBusType = libnvml.DeviceGetBusType + DeviceGetC2cModeInfoV = libnvml.DeviceGetC2cModeInfoV + DeviceGetCapabilities = libnvml.DeviceGetCapabilities + DeviceGetClkMonStatus = libnvml.DeviceGetClkMonStatus + DeviceGetClock = libnvml.DeviceGetClock + DeviceGetClockInfo = libnvml.DeviceGetClockInfo + DeviceGetClockOffsets = libnvml.DeviceGetClockOffsets + DeviceGetComputeInstanceId = libnvml.DeviceGetComputeInstanceId + DeviceGetComputeMode = libnvml.DeviceGetComputeMode + DeviceGetComputeRunningProcesses = libnvml.DeviceGetComputeRunningProcesses + DeviceGetConfComputeGpuAttestationReport = libnvml.DeviceGetConfComputeGpuAttestationReport + DeviceGetConfComputeGpuCertificate = libnvml.DeviceGetConfComputeGpuCertificate + DeviceGetConfComputeMemSizeInfo = libnvml.DeviceGetConfComputeMemSizeInfo + DeviceGetConfComputeProtectedMemoryUsage = libnvml.DeviceGetConfComputeProtectedMemoryUsage + DeviceGetCoolerInfo = libnvml.DeviceGetCoolerInfo + DeviceGetCount = libnvml.DeviceGetCount + DeviceGetCpuAffinity = libnvml.DeviceGetCpuAffinity + DeviceGetCpuAffinityWithinScope = libnvml.DeviceGetCpuAffinityWithinScope + DeviceGetCreatableVgpus = libnvml.DeviceGetCreatableVgpus + DeviceGetCudaComputeCapability = libnvml.DeviceGetCudaComputeCapability + DeviceGetCurrPcieLinkGeneration = libnvml.DeviceGetCurrPcieLinkGeneration + DeviceGetCurrPcieLinkWidth = libnvml.DeviceGetCurrPcieLinkWidth + DeviceGetCurrentClockFreqs = libnvml.DeviceGetCurrentClockFreqs + DeviceGetCurrentClocksEventReasons = libnvml.DeviceGetCurrentClocksEventReasons + DeviceGetCurrentClocksThrottleReasons = libnvml.DeviceGetCurrentClocksThrottleReasons + DeviceGetDecoderUtilization = libnvml.DeviceGetDecoderUtilization + DeviceGetDefaultApplicationsClock = libnvml.DeviceGetDefaultApplicationsClock + DeviceGetDefaultEccMode = libnvml.DeviceGetDefaultEccMode + DeviceGetDetailedEccErrors = libnvml.DeviceGetDetailedEccErrors + DeviceGetDeviceHandleFromMigDeviceHandle = libnvml.DeviceGetDeviceHandleFromMigDeviceHandle + DeviceGetDisplayActive = libnvml.DeviceGetDisplayActive + DeviceGetDisplayMode = libnvml.DeviceGetDisplayMode + DeviceGetDramEncryptionMode = libnvml.DeviceGetDramEncryptionMode + DeviceGetDriverModel = libnvml.DeviceGetDriverModel + DeviceGetDriverModel_v2 = libnvml.DeviceGetDriverModel_v2 + DeviceGetDynamicPstatesInfo = libnvml.DeviceGetDynamicPstatesInfo + DeviceGetEccMode = libnvml.DeviceGetEccMode + DeviceGetEncoderCapacity = libnvml.DeviceGetEncoderCapacity + DeviceGetEncoderSessions = libnvml.DeviceGetEncoderSessions + DeviceGetEncoderStats = libnvml.DeviceGetEncoderStats + DeviceGetEncoderUtilization = libnvml.DeviceGetEncoderUtilization + DeviceGetEnforcedPowerLimit = libnvml.DeviceGetEnforcedPowerLimit + DeviceGetFBCSessions = libnvml.DeviceGetFBCSessions + DeviceGetFBCStats = libnvml.DeviceGetFBCStats + DeviceGetFanControlPolicy_v2 = libnvml.DeviceGetFanControlPolicy_v2 + DeviceGetFanSpeed = libnvml.DeviceGetFanSpeed + DeviceGetFanSpeedRPM = libnvml.DeviceGetFanSpeedRPM + DeviceGetFanSpeed_v2 = libnvml.DeviceGetFanSpeed_v2 + DeviceGetFieldValues = libnvml.DeviceGetFieldValues + DeviceGetGpcClkMinMaxVfOffset = libnvml.DeviceGetGpcClkMinMaxVfOffset + DeviceGetGpcClkVfOffset = libnvml.DeviceGetGpcClkVfOffset + DeviceGetGpuFabricInfo = libnvml.DeviceGetGpuFabricInfo + DeviceGetGpuFabricInfoV = libnvml.DeviceGetGpuFabricInfoV + DeviceGetGpuInstanceById = libnvml.DeviceGetGpuInstanceById + DeviceGetGpuInstanceId = libnvml.DeviceGetGpuInstanceId + DeviceGetGpuInstancePossiblePlacements = libnvml.DeviceGetGpuInstancePossiblePlacements + DeviceGetGpuInstanceProfileInfo = libnvml.DeviceGetGpuInstanceProfileInfo + DeviceGetGpuInstanceProfileInfoByIdV = libnvml.DeviceGetGpuInstanceProfileInfoByIdV + DeviceGetGpuInstanceProfileInfoV = libnvml.DeviceGetGpuInstanceProfileInfoV + DeviceGetGpuInstanceRemainingCapacity = libnvml.DeviceGetGpuInstanceRemainingCapacity + DeviceGetGpuInstances = libnvml.DeviceGetGpuInstances + DeviceGetGpuMaxPcieLinkGeneration = libnvml.DeviceGetGpuMaxPcieLinkGeneration + DeviceGetGpuOperationMode = libnvml.DeviceGetGpuOperationMode + DeviceGetGraphicsRunningProcesses = libnvml.DeviceGetGraphicsRunningProcesses + DeviceGetGridLicensableFeatures = libnvml.DeviceGetGridLicensableFeatures + DeviceGetGspFirmwareMode = libnvml.DeviceGetGspFirmwareMode + DeviceGetGspFirmwareVersion = libnvml.DeviceGetGspFirmwareVersion + DeviceGetHandleByIndex = libnvml.DeviceGetHandleByIndex + DeviceGetHandleByPciBusId = libnvml.DeviceGetHandleByPciBusId + DeviceGetHandleBySerial = libnvml.DeviceGetHandleBySerial + DeviceGetHandleByUUID = libnvml.DeviceGetHandleByUUID + DeviceGetHandleByUUIDV = libnvml.DeviceGetHandleByUUIDV + DeviceGetHostVgpuMode = libnvml.DeviceGetHostVgpuMode + DeviceGetIndex = libnvml.DeviceGetIndex + DeviceGetInforomConfigurationChecksum = libnvml.DeviceGetInforomConfigurationChecksum + DeviceGetInforomImageVersion = libnvml.DeviceGetInforomImageVersion + DeviceGetInforomVersion = libnvml.DeviceGetInforomVersion + DeviceGetIrqNum = libnvml.DeviceGetIrqNum + DeviceGetJpgUtilization = libnvml.DeviceGetJpgUtilization + DeviceGetLastBBXFlushTime = libnvml.DeviceGetLastBBXFlushTime + DeviceGetMPSComputeRunningProcesses = libnvml.DeviceGetMPSComputeRunningProcesses + DeviceGetMarginTemperature = libnvml.DeviceGetMarginTemperature + DeviceGetMaxClockInfo = libnvml.DeviceGetMaxClockInfo + DeviceGetMaxCustomerBoostClock = libnvml.DeviceGetMaxCustomerBoostClock + DeviceGetMaxMigDeviceCount = libnvml.DeviceGetMaxMigDeviceCount + DeviceGetMaxPcieLinkGeneration = libnvml.DeviceGetMaxPcieLinkGeneration + DeviceGetMaxPcieLinkWidth = libnvml.DeviceGetMaxPcieLinkWidth + DeviceGetMemClkMinMaxVfOffset = libnvml.DeviceGetMemClkMinMaxVfOffset + DeviceGetMemClkVfOffset = libnvml.DeviceGetMemClkVfOffset + DeviceGetMemoryAffinity = libnvml.DeviceGetMemoryAffinity + DeviceGetMemoryBusWidth = libnvml.DeviceGetMemoryBusWidth + DeviceGetMemoryErrorCounter = libnvml.DeviceGetMemoryErrorCounter + DeviceGetMemoryInfo = libnvml.DeviceGetMemoryInfo + DeviceGetMemoryInfo_v2 = libnvml.DeviceGetMemoryInfo_v2 + DeviceGetMigDeviceHandleByIndex = libnvml.DeviceGetMigDeviceHandleByIndex + DeviceGetMigMode = libnvml.DeviceGetMigMode + DeviceGetMinMaxClockOfPState = libnvml.DeviceGetMinMaxClockOfPState + DeviceGetMinMaxFanSpeed = libnvml.DeviceGetMinMaxFanSpeed + DeviceGetMinorNumber = libnvml.DeviceGetMinorNumber + DeviceGetModuleId = libnvml.DeviceGetModuleId + DeviceGetMultiGpuBoard = libnvml.DeviceGetMultiGpuBoard + DeviceGetName = libnvml.DeviceGetName + DeviceGetNumFans = libnvml.DeviceGetNumFans + DeviceGetNumGpuCores = libnvml.DeviceGetNumGpuCores + DeviceGetNumaNodeId = libnvml.DeviceGetNumaNodeId + DeviceGetNvLinkCapability = libnvml.DeviceGetNvLinkCapability + DeviceGetNvLinkErrorCounter = libnvml.DeviceGetNvLinkErrorCounter + DeviceGetNvLinkInfo = libnvml.DeviceGetNvLinkInfo + DeviceGetNvLinkRemoteDeviceType = libnvml.DeviceGetNvLinkRemoteDeviceType + DeviceGetNvLinkRemotePciInfo = libnvml.DeviceGetNvLinkRemotePciInfo + DeviceGetNvLinkState = libnvml.DeviceGetNvLinkState + DeviceGetNvLinkUtilizationControl = libnvml.DeviceGetNvLinkUtilizationControl + DeviceGetNvLinkUtilizationCounter = libnvml.DeviceGetNvLinkUtilizationCounter + DeviceGetNvLinkVersion = libnvml.DeviceGetNvLinkVersion + DeviceGetNvlinkBwMode = libnvml.DeviceGetNvlinkBwMode + DeviceGetNvlinkSupportedBwModes = libnvml.DeviceGetNvlinkSupportedBwModes + DeviceGetOfaUtilization = libnvml.DeviceGetOfaUtilization + DeviceGetP2PStatus = libnvml.DeviceGetP2PStatus + DeviceGetPciInfo = libnvml.DeviceGetPciInfo + DeviceGetPciInfoExt = libnvml.DeviceGetPciInfoExt + DeviceGetPcieLinkMaxSpeed = libnvml.DeviceGetPcieLinkMaxSpeed + DeviceGetPcieReplayCounter = libnvml.DeviceGetPcieReplayCounter + DeviceGetPcieSpeed = libnvml.DeviceGetPcieSpeed + DeviceGetPcieThroughput = libnvml.DeviceGetPcieThroughput + DeviceGetPdi = libnvml.DeviceGetPdi + DeviceGetPerformanceModes = libnvml.DeviceGetPerformanceModes + DeviceGetPerformanceState = libnvml.DeviceGetPerformanceState + DeviceGetPersistenceMode = libnvml.DeviceGetPersistenceMode + DeviceGetPgpuMetadataString = libnvml.DeviceGetPgpuMetadataString + DeviceGetPlatformInfo = libnvml.DeviceGetPlatformInfo + DeviceGetPowerManagementDefaultLimit = libnvml.DeviceGetPowerManagementDefaultLimit + DeviceGetPowerManagementLimit = libnvml.DeviceGetPowerManagementLimit + DeviceGetPowerManagementLimitConstraints = libnvml.DeviceGetPowerManagementLimitConstraints + DeviceGetPowerManagementMode = libnvml.DeviceGetPowerManagementMode + DeviceGetPowerMizerMode_v1 = libnvml.DeviceGetPowerMizerMode_v1 + DeviceGetPowerSource = libnvml.DeviceGetPowerSource + DeviceGetPowerState = libnvml.DeviceGetPowerState + DeviceGetPowerUsage = libnvml.DeviceGetPowerUsage + DeviceGetProcessUtilization = libnvml.DeviceGetProcessUtilization + DeviceGetProcessesUtilizationInfo = libnvml.DeviceGetProcessesUtilizationInfo + DeviceGetRemappedRows = libnvml.DeviceGetRemappedRows + DeviceGetRepairStatus = libnvml.DeviceGetRepairStatus + DeviceGetRetiredPages = libnvml.DeviceGetRetiredPages + DeviceGetRetiredPagesPendingStatus = libnvml.DeviceGetRetiredPagesPendingStatus + DeviceGetRetiredPages_v2 = libnvml.DeviceGetRetiredPages_v2 + DeviceGetRowRemapperHistogram = libnvml.DeviceGetRowRemapperHistogram + DeviceGetRunningProcessDetailList = libnvml.DeviceGetRunningProcessDetailList + DeviceGetSamples = libnvml.DeviceGetSamples + DeviceGetSerial = libnvml.DeviceGetSerial + DeviceGetSramEccErrorStatus = libnvml.DeviceGetSramEccErrorStatus + DeviceGetSramUniqueUncorrectedEccErrorCounts = libnvml.DeviceGetSramUniqueUncorrectedEccErrorCounts + DeviceGetSupportedClocksEventReasons = libnvml.DeviceGetSupportedClocksEventReasons + DeviceGetSupportedClocksThrottleReasons = libnvml.DeviceGetSupportedClocksThrottleReasons + DeviceGetSupportedEventTypes = libnvml.DeviceGetSupportedEventTypes + DeviceGetSupportedGraphicsClocks = libnvml.DeviceGetSupportedGraphicsClocks + DeviceGetSupportedMemoryClocks = libnvml.DeviceGetSupportedMemoryClocks + DeviceGetSupportedPerformanceStates = libnvml.DeviceGetSupportedPerformanceStates + DeviceGetSupportedVgpus = libnvml.DeviceGetSupportedVgpus + DeviceGetTargetFanSpeed = libnvml.DeviceGetTargetFanSpeed + DeviceGetTemperature = libnvml.DeviceGetTemperature + DeviceGetTemperatureThreshold = libnvml.DeviceGetTemperatureThreshold + DeviceGetTemperatureV = libnvml.DeviceGetTemperatureV + DeviceGetThermalSettings = libnvml.DeviceGetThermalSettings + DeviceGetTopologyCommonAncestor = libnvml.DeviceGetTopologyCommonAncestor + DeviceGetTopologyNearestGpus = libnvml.DeviceGetTopologyNearestGpus + DeviceGetTotalEccErrors = libnvml.DeviceGetTotalEccErrors + DeviceGetTotalEnergyConsumption = libnvml.DeviceGetTotalEnergyConsumption + DeviceGetUUID = libnvml.DeviceGetUUID + DeviceGetUtilizationRates = libnvml.DeviceGetUtilizationRates + DeviceGetVbiosVersion = libnvml.DeviceGetVbiosVersion + DeviceGetVgpuCapabilities = libnvml.DeviceGetVgpuCapabilities + DeviceGetVgpuHeterogeneousMode = libnvml.DeviceGetVgpuHeterogeneousMode + DeviceGetVgpuInstancesUtilizationInfo = libnvml.DeviceGetVgpuInstancesUtilizationInfo + DeviceGetVgpuMetadata = libnvml.DeviceGetVgpuMetadata + DeviceGetVgpuProcessUtilization = libnvml.DeviceGetVgpuProcessUtilization + DeviceGetVgpuProcessesUtilizationInfo = libnvml.DeviceGetVgpuProcessesUtilizationInfo + DeviceGetVgpuSchedulerCapabilities = libnvml.DeviceGetVgpuSchedulerCapabilities + DeviceGetVgpuSchedulerLog = libnvml.DeviceGetVgpuSchedulerLog + DeviceGetVgpuSchedulerState = libnvml.DeviceGetVgpuSchedulerState + DeviceGetVgpuTypeCreatablePlacements = libnvml.DeviceGetVgpuTypeCreatablePlacements + DeviceGetVgpuTypeSupportedPlacements = libnvml.DeviceGetVgpuTypeSupportedPlacements + DeviceGetVgpuUtilization = libnvml.DeviceGetVgpuUtilization + DeviceGetViolationStatus = libnvml.DeviceGetViolationStatus + DeviceGetVirtualizationMode = libnvml.DeviceGetVirtualizationMode + DeviceIsMigDeviceHandle = libnvml.DeviceIsMigDeviceHandle + DeviceModifyDrainState = libnvml.DeviceModifyDrainState + DeviceOnSameBoard = libnvml.DeviceOnSameBoard + DevicePowerSmoothingActivatePresetProfile = libnvml.DevicePowerSmoothingActivatePresetProfile + DevicePowerSmoothingSetState = libnvml.DevicePowerSmoothingSetState + DevicePowerSmoothingUpdatePresetProfileParam = libnvml.DevicePowerSmoothingUpdatePresetProfileParam + DeviceQueryDrainState = libnvml.DeviceQueryDrainState + DeviceReadWritePRM_v1 = libnvml.DeviceReadWritePRM_v1 + DeviceRegisterEvents = libnvml.DeviceRegisterEvents + DeviceRemoveGpu = libnvml.DeviceRemoveGpu + DeviceRemoveGpu_v2 = libnvml.DeviceRemoveGpu_v2 + DeviceResetApplicationsClocks = libnvml.DeviceResetApplicationsClocks + DeviceResetGpuLockedClocks = libnvml.DeviceResetGpuLockedClocks + DeviceResetMemoryLockedClocks = libnvml.DeviceResetMemoryLockedClocks + DeviceResetNvLinkErrorCounters = libnvml.DeviceResetNvLinkErrorCounters + DeviceResetNvLinkUtilizationCounter = libnvml.DeviceResetNvLinkUtilizationCounter + DeviceSetAPIRestriction = libnvml.DeviceSetAPIRestriction + DeviceSetAccountingMode = libnvml.DeviceSetAccountingMode + DeviceSetApplicationsClocks = libnvml.DeviceSetApplicationsClocks + DeviceSetAutoBoostedClocksEnabled = libnvml.DeviceSetAutoBoostedClocksEnabled + DeviceSetClockOffsets = libnvml.DeviceSetClockOffsets + DeviceSetComputeMode = libnvml.DeviceSetComputeMode + DeviceSetConfComputeUnprotectedMemSize = libnvml.DeviceSetConfComputeUnprotectedMemSize + DeviceSetCpuAffinity = libnvml.DeviceSetCpuAffinity + DeviceSetDefaultAutoBoostedClocksEnabled = libnvml.DeviceSetDefaultAutoBoostedClocksEnabled + DeviceSetDefaultFanSpeed_v2 = libnvml.DeviceSetDefaultFanSpeed_v2 + DeviceSetDramEncryptionMode = libnvml.DeviceSetDramEncryptionMode + DeviceSetDriverModel = libnvml.DeviceSetDriverModel + DeviceSetEccMode = libnvml.DeviceSetEccMode + DeviceSetFanControlPolicy = libnvml.DeviceSetFanControlPolicy + DeviceSetFanSpeed_v2 = libnvml.DeviceSetFanSpeed_v2 + DeviceSetGpcClkVfOffset = libnvml.DeviceSetGpcClkVfOffset + DeviceSetGpuLockedClocks = libnvml.DeviceSetGpuLockedClocks + DeviceSetGpuOperationMode = libnvml.DeviceSetGpuOperationMode + DeviceSetMemClkVfOffset = libnvml.DeviceSetMemClkVfOffset + DeviceSetMemoryLockedClocks = libnvml.DeviceSetMemoryLockedClocks + DeviceSetMigMode = libnvml.DeviceSetMigMode + DeviceSetNvLinkDeviceLowPowerThreshold = libnvml.DeviceSetNvLinkDeviceLowPowerThreshold + DeviceSetNvLinkUtilizationControl = libnvml.DeviceSetNvLinkUtilizationControl + DeviceSetNvlinkBwMode = libnvml.DeviceSetNvlinkBwMode + DeviceSetPersistenceMode = libnvml.DeviceSetPersistenceMode + DeviceSetPowerManagementLimit = libnvml.DeviceSetPowerManagementLimit + DeviceSetPowerManagementLimit_v2 = libnvml.DeviceSetPowerManagementLimit_v2 + DeviceSetTemperatureThreshold = libnvml.DeviceSetTemperatureThreshold + DeviceSetVgpuCapabilities = libnvml.DeviceSetVgpuCapabilities + DeviceSetVgpuHeterogeneousMode = libnvml.DeviceSetVgpuHeterogeneousMode + DeviceSetVgpuSchedulerState = libnvml.DeviceSetVgpuSchedulerState + DeviceSetVirtualizationMode = libnvml.DeviceSetVirtualizationMode + DeviceValidateInforom = libnvml.DeviceValidateInforom + DeviceWorkloadPowerProfileClearRequestedProfiles = libnvml.DeviceWorkloadPowerProfileClearRequestedProfiles + DeviceWorkloadPowerProfileGetCurrentProfiles = libnvml.DeviceWorkloadPowerProfileGetCurrentProfiles + DeviceWorkloadPowerProfileGetProfilesInfo = libnvml.DeviceWorkloadPowerProfileGetProfilesInfo + DeviceWorkloadPowerProfileSetRequestedProfiles = libnvml.DeviceWorkloadPowerProfileSetRequestedProfiles + ErrorString = libnvml.ErrorString + EventSetCreate = libnvml.EventSetCreate + EventSetFree = libnvml.EventSetFree + EventSetWait = libnvml.EventSetWait + Extensions = libnvml.Extensions + GetExcludedDeviceCount = libnvml.GetExcludedDeviceCount + GetExcludedDeviceInfoByIndex = libnvml.GetExcludedDeviceInfoByIndex + GetVgpuCompatibility = libnvml.GetVgpuCompatibility + GetVgpuDriverCapabilities = libnvml.GetVgpuDriverCapabilities + GetVgpuVersion = libnvml.GetVgpuVersion + GpmMetricsGet = libnvml.GpmMetricsGet + GpmMetricsGetV = libnvml.GpmMetricsGetV + GpmMigSampleGet = libnvml.GpmMigSampleGet + GpmQueryDeviceSupport = libnvml.GpmQueryDeviceSupport + GpmQueryDeviceSupportV = libnvml.GpmQueryDeviceSupportV + GpmQueryIfStreamingEnabled = libnvml.GpmQueryIfStreamingEnabled + GpmSampleAlloc = libnvml.GpmSampleAlloc + GpmSampleFree = libnvml.GpmSampleFree + GpmSampleGet = libnvml.GpmSampleGet + GpmSetStreamingEnabled = libnvml.GpmSetStreamingEnabled + GpuInstanceCreateComputeInstance = libnvml.GpuInstanceCreateComputeInstance + GpuInstanceCreateComputeInstanceWithPlacement = libnvml.GpuInstanceCreateComputeInstanceWithPlacement + GpuInstanceDestroy = libnvml.GpuInstanceDestroy + GpuInstanceGetActiveVgpus = libnvml.GpuInstanceGetActiveVgpus + GpuInstanceGetComputeInstanceById = libnvml.GpuInstanceGetComputeInstanceById + GpuInstanceGetComputeInstancePossiblePlacements = libnvml.GpuInstanceGetComputeInstancePossiblePlacements + GpuInstanceGetComputeInstanceProfileInfo = libnvml.GpuInstanceGetComputeInstanceProfileInfo + GpuInstanceGetComputeInstanceProfileInfoV = libnvml.GpuInstanceGetComputeInstanceProfileInfoV + GpuInstanceGetComputeInstanceRemainingCapacity = libnvml.GpuInstanceGetComputeInstanceRemainingCapacity + GpuInstanceGetComputeInstances = libnvml.GpuInstanceGetComputeInstances + GpuInstanceGetCreatableVgpus = libnvml.GpuInstanceGetCreatableVgpus + GpuInstanceGetInfo = libnvml.GpuInstanceGetInfo + GpuInstanceGetVgpuHeterogeneousMode = libnvml.GpuInstanceGetVgpuHeterogeneousMode + GpuInstanceGetVgpuSchedulerLog = libnvml.GpuInstanceGetVgpuSchedulerLog + GpuInstanceGetVgpuSchedulerState = libnvml.GpuInstanceGetVgpuSchedulerState + GpuInstanceGetVgpuTypeCreatablePlacements = libnvml.GpuInstanceGetVgpuTypeCreatablePlacements + GpuInstanceSetVgpuHeterogeneousMode = libnvml.GpuInstanceSetVgpuHeterogeneousMode + GpuInstanceSetVgpuSchedulerState = libnvml.GpuInstanceSetVgpuSchedulerState + Init = libnvml.Init + InitWithFlags = libnvml.InitWithFlags + SetVgpuVersion = libnvml.SetVgpuVersion + Shutdown = libnvml.Shutdown + SystemEventSetCreate = libnvml.SystemEventSetCreate + SystemEventSetFree = libnvml.SystemEventSetFree + SystemEventSetWait = libnvml.SystemEventSetWait + SystemGetConfComputeCapabilities = libnvml.SystemGetConfComputeCapabilities + SystemGetConfComputeGpusReadyState = libnvml.SystemGetConfComputeGpusReadyState + SystemGetConfComputeKeyRotationThresholdInfo = libnvml.SystemGetConfComputeKeyRotationThresholdInfo + SystemGetConfComputeSettings = libnvml.SystemGetConfComputeSettings + SystemGetConfComputeState = libnvml.SystemGetConfComputeState + SystemGetCudaDriverVersion = libnvml.SystemGetCudaDriverVersion + SystemGetCudaDriverVersion_v2 = libnvml.SystemGetCudaDriverVersion_v2 + SystemGetDriverBranch = libnvml.SystemGetDriverBranch + SystemGetDriverVersion = libnvml.SystemGetDriverVersion + SystemGetHicVersion = libnvml.SystemGetHicVersion + SystemGetNVMLVersion = libnvml.SystemGetNVMLVersion + SystemGetNvlinkBwMode = libnvml.SystemGetNvlinkBwMode + SystemGetProcessName = libnvml.SystemGetProcessName + SystemGetTopologyGpuSet = libnvml.SystemGetTopologyGpuSet + SystemRegisterEvents = libnvml.SystemRegisterEvents + SystemSetConfComputeGpusReadyState = libnvml.SystemSetConfComputeGpusReadyState + SystemSetConfComputeKeyRotationThresholdInfo = libnvml.SystemSetConfComputeKeyRotationThresholdInfo + SystemSetNvlinkBwMode = libnvml.SystemSetNvlinkBwMode + UnitGetCount = libnvml.UnitGetCount + UnitGetDevices = libnvml.UnitGetDevices + UnitGetFanSpeedInfo = libnvml.UnitGetFanSpeedInfo + UnitGetHandleByIndex = libnvml.UnitGetHandleByIndex + UnitGetLedState = libnvml.UnitGetLedState + UnitGetPsuInfo = libnvml.UnitGetPsuInfo + UnitGetTemperature = libnvml.UnitGetTemperature + UnitGetUnitInfo = libnvml.UnitGetUnitInfo + UnitSetLedState = libnvml.UnitSetLedState + VgpuInstanceClearAccountingPids = libnvml.VgpuInstanceClearAccountingPids + VgpuInstanceGetAccountingMode = libnvml.VgpuInstanceGetAccountingMode + VgpuInstanceGetAccountingPids = libnvml.VgpuInstanceGetAccountingPids + VgpuInstanceGetAccountingStats = libnvml.VgpuInstanceGetAccountingStats + VgpuInstanceGetEccMode = libnvml.VgpuInstanceGetEccMode + VgpuInstanceGetEncoderCapacity = libnvml.VgpuInstanceGetEncoderCapacity + VgpuInstanceGetEncoderSessions = libnvml.VgpuInstanceGetEncoderSessions + VgpuInstanceGetEncoderStats = libnvml.VgpuInstanceGetEncoderStats + VgpuInstanceGetFBCSessions = libnvml.VgpuInstanceGetFBCSessions + VgpuInstanceGetFBCStats = libnvml.VgpuInstanceGetFBCStats + VgpuInstanceGetFbUsage = libnvml.VgpuInstanceGetFbUsage + VgpuInstanceGetFrameRateLimit = libnvml.VgpuInstanceGetFrameRateLimit + VgpuInstanceGetGpuInstanceId = libnvml.VgpuInstanceGetGpuInstanceId + VgpuInstanceGetGpuPciId = libnvml.VgpuInstanceGetGpuPciId + VgpuInstanceGetLicenseInfo = libnvml.VgpuInstanceGetLicenseInfo + VgpuInstanceGetLicenseStatus = libnvml.VgpuInstanceGetLicenseStatus + VgpuInstanceGetMdevUUID = libnvml.VgpuInstanceGetMdevUUID + VgpuInstanceGetMetadata = libnvml.VgpuInstanceGetMetadata + VgpuInstanceGetRuntimeStateSize = libnvml.VgpuInstanceGetRuntimeStateSize + VgpuInstanceGetType = libnvml.VgpuInstanceGetType + VgpuInstanceGetUUID = libnvml.VgpuInstanceGetUUID + VgpuInstanceGetVmDriverVersion = libnvml.VgpuInstanceGetVmDriverVersion + VgpuInstanceGetVmID = libnvml.VgpuInstanceGetVmID + VgpuInstanceSetEncoderCapacity = libnvml.VgpuInstanceSetEncoderCapacity + VgpuTypeGetBAR1Info = libnvml.VgpuTypeGetBAR1Info + VgpuTypeGetCapabilities = libnvml.VgpuTypeGetCapabilities + VgpuTypeGetClass = libnvml.VgpuTypeGetClass + VgpuTypeGetDeviceID = libnvml.VgpuTypeGetDeviceID + VgpuTypeGetFrameRateLimit = libnvml.VgpuTypeGetFrameRateLimit + VgpuTypeGetFramebufferSize = libnvml.VgpuTypeGetFramebufferSize + VgpuTypeGetGpuInstanceProfileId = libnvml.VgpuTypeGetGpuInstanceProfileId + VgpuTypeGetLicense = libnvml.VgpuTypeGetLicense + VgpuTypeGetMaxInstances = libnvml.VgpuTypeGetMaxInstances + VgpuTypeGetMaxInstancesPerGpuInstance = libnvml.VgpuTypeGetMaxInstancesPerGpuInstance + VgpuTypeGetMaxInstancesPerVm = libnvml.VgpuTypeGetMaxInstancesPerVm + VgpuTypeGetName = libnvml.VgpuTypeGetName + VgpuTypeGetNumDisplayHeads = libnvml.VgpuTypeGetNumDisplayHeads + VgpuTypeGetResolution = libnvml.VgpuTypeGetResolution +) + +// Interface represents the interface for the library type. +// +//go:generate moq -out mock/interface.go -pkg mock . Interface:Interface +type Interface interface { + ComputeInstanceDestroy(ComputeInstance) Return + ComputeInstanceGetInfo(ComputeInstance) (ComputeInstanceInfo, Return) + DeviceClearAccountingPids(Device) Return + DeviceClearCpuAffinity(Device) Return + DeviceClearEccErrorCounts(Device, EccCounterType) Return + DeviceClearFieldValues(Device, []FieldValue) Return + DeviceCreateGpuInstance(Device, *GpuInstanceProfileInfo) (GpuInstance, Return) + DeviceCreateGpuInstanceWithPlacement(Device, *GpuInstanceProfileInfo, *GpuInstancePlacement) (GpuInstance, Return) + DeviceDiscoverGpus() (PciInfo, Return) + DeviceFreezeNvLinkUtilizationCounter(Device, int, int, EnableState) Return + DeviceGetAPIRestriction(Device, RestrictedAPI) (EnableState, Return) + DeviceGetAccountingBufferSize(Device) (int, Return) + DeviceGetAccountingMode(Device) (EnableState, Return) + DeviceGetAccountingPids(Device) ([]int, Return) + DeviceGetAccountingStats(Device, uint32) (AccountingStats, Return) + DeviceGetActiveVgpus(Device) ([]VgpuInstance, Return) + DeviceGetAdaptiveClockInfoStatus(Device) (uint32, Return) + DeviceGetAddressingMode(Device) (DeviceAddressingMode, Return) + DeviceGetApplicationsClock(Device, ClockType) (uint32, Return) + DeviceGetArchitecture(Device) (DeviceArchitecture, Return) + DeviceGetAttributes(Device) (DeviceAttributes, Return) + DeviceGetAutoBoostedClocksEnabled(Device) (EnableState, EnableState, Return) + DeviceGetBAR1MemoryInfo(Device) (BAR1Memory, Return) + DeviceGetBoardId(Device) (uint32, Return) + DeviceGetBoardPartNumber(Device) (string, Return) + DeviceGetBrand(Device) (BrandType, Return) + DeviceGetBridgeChipInfo(Device) (BridgeChipHierarchy, Return) + DeviceGetBusType(Device) (BusType, Return) + DeviceGetC2cModeInfoV(Device) C2cModeInfoHandler + DeviceGetCapabilities(Device) (DeviceCapabilities, Return) + DeviceGetClkMonStatus(Device) (ClkMonStatus, Return) + DeviceGetClock(Device, ClockType, ClockId) (uint32, Return) + DeviceGetClockInfo(Device, ClockType) (uint32, Return) + DeviceGetClockOffsets(Device) (ClockOffset, Return) + DeviceGetComputeInstanceId(Device) (int, Return) + DeviceGetComputeMode(Device) (ComputeMode, Return) + DeviceGetComputeRunningProcesses(Device) ([]ProcessInfo, Return) + DeviceGetConfComputeGpuAttestationReport(Device, *ConfComputeGpuAttestationReport) Return + DeviceGetConfComputeGpuCertificate(Device) (ConfComputeGpuCertificate, Return) + DeviceGetConfComputeMemSizeInfo(Device) (ConfComputeMemSizeInfo, Return) + DeviceGetConfComputeProtectedMemoryUsage(Device) (Memory, Return) + DeviceGetCoolerInfo(Device) (CoolerInfo, Return) + DeviceGetCount() (int, Return) + DeviceGetCpuAffinity(Device, int) ([]uint, Return) + DeviceGetCpuAffinityWithinScope(Device, int, AffinityScope) ([]uint, Return) + DeviceGetCreatableVgpus(Device) ([]VgpuTypeId, Return) + DeviceGetCudaComputeCapability(Device) (int, int, Return) + DeviceGetCurrPcieLinkGeneration(Device) (int, Return) + DeviceGetCurrPcieLinkWidth(Device) (int, Return) + DeviceGetCurrentClockFreqs(Device) (DeviceCurrentClockFreqs, Return) + DeviceGetCurrentClocksEventReasons(Device) (uint64, Return) + DeviceGetCurrentClocksThrottleReasons(Device) (uint64, Return) + DeviceGetDecoderUtilization(Device) (uint32, uint32, Return) + DeviceGetDefaultApplicationsClock(Device, ClockType) (uint32, Return) + DeviceGetDefaultEccMode(Device) (EnableState, Return) + DeviceGetDetailedEccErrors(Device, MemoryErrorType, EccCounterType) (EccErrorCounts, Return) + DeviceGetDeviceHandleFromMigDeviceHandle(Device) (Device, Return) + DeviceGetDisplayActive(Device) (EnableState, Return) + DeviceGetDisplayMode(Device) (EnableState, Return) + DeviceGetDramEncryptionMode(Device) (DramEncryptionInfo, DramEncryptionInfo, Return) + DeviceGetDriverModel(Device) (DriverModel, DriverModel, Return) + DeviceGetDriverModel_v2(Device) (DriverModel, DriverModel, Return) + DeviceGetDynamicPstatesInfo(Device) (GpuDynamicPstatesInfo, Return) + DeviceGetEccMode(Device) (EnableState, EnableState, Return) + DeviceGetEncoderCapacity(Device, EncoderType) (int, Return) + DeviceGetEncoderSessions(Device) ([]EncoderSessionInfo, Return) + DeviceGetEncoderStats(Device) (int, uint32, uint32, Return) + DeviceGetEncoderUtilization(Device) (uint32, uint32, Return) + DeviceGetEnforcedPowerLimit(Device) (uint32, Return) + DeviceGetFBCSessions(Device) ([]FBCSessionInfo, Return) + DeviceGetFBCStats(Device) (FBCStats, Return) + DeviceGetFanControlPolicy_v2(Device, int) (FanControlPolicy, Return) + DeviceGetFanSpeed(Device) (uint32, Return) + DeviceGetFanSpeedRPM(Device) (FanSpeedInfo, Return) + DeviceGetFanSpeed_v2(Device, int) (uint32, Return) + DeviceGetFieldValues(Device, []FieldValue) Return + DeviceGetGpcClkMinMaxVfOffset(Device) (int, int, Return) + DeviceGetGpcClkVfOffset(Device) (int, Return) + DeviceGetGpuFabricInfo(Device) (GpuFabricInfo, Return) + DeviceGetGpuFabricInfoV(Device) GpuFabricInfoHandler + DeviceGetGpuInstanceById(Device, int) (GpuInstance, Return) + DeviceGetGpuInstanceId(Device) (int, Return) + DeviceGetGpuInstancePossiblePlacements(Device, *GpuInstanceProfileInfo) ([]GpuInstancePlacement, Return) + DeviceGetGpuInstanceProfileInfo(Device, int) (GpuInstanceProfileInfo, Return) + DeviceGetGpuInstanceProfileInfoByIdV(Device, int) GpuInstanceProfileInfoByIdHandler + DeviceGetGpuInstanceProfileInfoV(Device, int) GpuInstanceProfileInfoHandler + DeviceGetGpuInstanceRemainingCapacity(Device, *GpuInstanceProfileInfo) (int, Return) + DeviceGetGpuInstances(Device, *GpuInstanceProfileInfo) ([]GpuInstance, Return) + DeviceGetGpuMaxPcieLinkGeneration(Device) (int, Return) + DeviceGetGpuOperationMode(Device) (GpuOperationMode, GpuOperationMode, Return) + DeviceGetGraphicsRunningProcesses(Device) ([]ProcessInfo, Return) + DeviceGetGridLicensableFeatures(Device) (GridLicensableFeatures, Return) + DeviceGetGspFirmwareMode(Device) (bool, bool, Return) + DeviceGetGspFirmwareVersion(Device) (string, Return) + DeviceGetHandleByIndex(int) (Device, Return) + DeviceGetHandleByPciBusId(string) (Device, Return) + DeviceGetHandleBySerial(string) (Device, Return) + DeviceGetHandleByUUID(string) (Device, Return) + DeviceGetHandleByUUIDV(*UUID) (Device, Return) + DeviceGetHostVgpuMode(Device) (HostVgpuMode, Return) + DeviceGetIndex(Device) (int, Return) + DeviceGetInforomConfigurationChecksum(Device) (uint32, Return) + DeviceGetInforomImageVersion(Device) (string, Return) + DeviceGetInforomVersion(Device, InforomObject) (string, Return) + DeviceGetIrqNum(Device) (int, Return) + DeviceGetJpgUtilization(Device) (uint32, uint32, Return) + DeviceGetLastBBXFlushTime(Device) (uint64, uint, Return) + DeviceGetMPSComputeRunningProcesses(Device) ([]ProcessInfo, Return) + DeviceGetMarginTemperature(Device) (MarginTemperature, Return) + DeviceGetMaxClockInfo(Device, ClockType) (uint32, Return) + DeviceGetMaxCustomerBoostClock(Device, ClockType) (uint32, Return) + DeviceGetMaxMigDeviceCount(Device) (int, Return) + DeviceGetMaxPcieLinkGeneration(Device) (int, Return) + DeviceGetMaxPcieLinkWidth(Device) (int, Return) + DeviceGetMemClkMinMaxVfOffset(Device) (int, int, Return) + DeviceGetMemClkVfOffset(Device) (int, Return) + DeviceGetMemoryAffinity(Device, int, AffinityScope) ([]uint, Return) + DeviceGetMemoryBusWidth(Device) (uint32, Return) + DeviceGetMemoryErrorCounter(Device, MemoryErrorType, EccCounterType, MemoryLocation) (uint64, Return) + DeviceGetMemoryInfo(Device) (Memory, Return) + DeviceGetMemoryInfo_v2(Device) (Memory_v2, Return) + DeviceGetMigDeviceHandleByIndex(Device, int) (Device, Return) + DeviceGetMigMode(Device) (int, int, Return) + DeviceGetMinMaxClockOfPState(Device, ClockType, Pstates) (uint32, uint32, Return) + DeviceGetMinMaxFanSpeed(Device) (int, int, Return) + DeviceGetMinorNumber(Device) (int, Return) + DeviceGetModuleId(Device) (int, Return) + DeviceGetMultiGpuBoard(Device) (int, Return) + DeviceGetName(Device) (string, Return) + DeviceGetNumFans(Device) (int, Return) + DeviceGetNumGpuCores(Device) (int, Return) + DeviceGetNumaNodeId(Device) (int, Return) + DeviceGetNvLinkCapability(Device, int, NvLinkCapability) (uint32, Return) + DeviceGetNvLinkErrorCounter(Device, int, NvLinkErrorCounter) (uint64, Return) + DeviceGetNvLinkInfo(Device) NvLinkInfoHandler + DeviceGetNvLinkRemoteDeviceType(Device, int) (IntNvLinkDeviceType, Return) + DeviceGetNvLinkRemotePciInfo(Device, int) (PciInfo, Return) + DeviceGetNvLinkState(Device, int) (EnableState, Return) + DeviceGetNvLinkUtilizationControl(Device, int, int) (NvLinkUtilizationControl, Return) + DeviceGetNvLinkUtilizationCounter(Device, int, int) (uint64, uint64, Return) + DeviceGetNvLinkVersion(Device, int) (uint32, Return) + DeviceGetNvlinkBwMode(Device) (NvlinkGetBwMode, Return) + DeviceGetNvlinkSupportedBwModes(Device) (NvlinkSupportedBwModes, Return) + DeviceGetOfaUtilization(Device) (uint32, uint32, Return) + DeviceGetP2PStatus(Device, Device, GpuP2PCapsIndex) (GpuP2PStatus, Return) + DeviceGetPciInfo(Device) (PciInfo, Return) + DeviceGetPciInfoExt(Device) (PciInfoExt, Return) + DeviceGetPcieLinkMaxSpeed(Device) (uint32, Return) + DeviceGetPcieReplayCounter(Device) (int, Return) + DeviceGetPcieSpeed(Device) (int, Return) + DeviceGetPcieThroughput(Device, PcieUtilCounter) (uint32, Return) + DeviceGetPdi(Device) (Pdi, Return) + DeviceGetPerformanceModes(Device) (DevicePerfModes, Return) + DeviceGetPerformanceState(Device) (Pstates, Return) + DeviceGetPersistenceMode(Device) (EnableState, Return) + DeviceGetPgpuMetadataString(Device) (string, Return) + DeviceGetPlatformInfo(Device) (PlatformInfo, Return) + DeviceGetPowerManagementDefaultLimit(Device) (uint32, Return) + DeviceGetPowerManagementLimit(Device) (uint32, Return) + DeviceGetPowerManagementLimitConstraints(Device) (uint32, uint32, Return) + DeviceGetPowerManagementMode(Device) (EnableState, Return) + DeviceGetPowerMizerMode_v1(Device) (DevicePowerMizerModes_v1, Return) + DeviceGetPowerSource(Device) (PowerSource, Return) + DeviceGetPowerState(Device) (Pstates, Return) + DeviceGetPowerUsage(Device) (uint32, Return) + DeviceGetProcessUtilization(Device, uint64) ([]ProcessUtilizationSample, Return) + DeviceGetProcessesUtilizationInfo(Device) (ProcessesUtilizationInfo, Return) + DeviceGetRemappedRows(Device) (int, int, bool, bool, Return) + DeviceGetRepairStatus(Device) (RepairStatus, Return) + DeviceGetRetiredPages(Device, PageRetirementCause) ([]uint64, Return) + DeviceGetRetiredPagesPendingStatus(Device) (EnableState, Return) + DeviceGetRetiredPages_v2(Device, PageRetirementCause) ([]uint64, []uint64, Return) + DeviceGetRowRemapperHistogram(Device) (RowRemapperHistogramValues, Return) + DeviceGetRunningProcessDetailList(Device) (ProcessDetailList, Return) + DeviceGetSamples(Device, SamplingType, uint64) (ValueType, []Sample, Return) + DeviceGetSerial(Device) (string, Return) + DeviceGetSramEccErrorStatus(Device) (EccSramErrorStatus, Return) + DeviceGetSramUniqueUncorrectedEccErrorCounts(Device, *EccSramUniqueUncorrectedErrorCounts) Return + DeviceGetSupportedClocksEventReasons(Device) (uint64, Return) + DeviceGetSupportedClocksThrottleReasons(Device) (uint64, Return) + DeviceGetSupportedEventTypes(Device) (uint64, Return) + DeviceGetSupportedGraphicsClocks(Device, int) (int, uint32, Return) + DeviceGetSupportedMemoryClocks(Device) (int, uint32, Return) + DeviceGetSupportedPerformanceStates(Device) ([]Pstates, Return) + DeviceGetSupportedVgpus(Device) ([]VgpuTypeId, Return) + DeviceGetTargetFanSpeed(Device, int) (int, Return) + DeviceGetTemperature(Device, TemperatureSensors) (uint32, Return) + DeviceGetTemperatureThreshold(Device, TemperatureThresholds) (uint32, Return) + DeviceGetTemperatureV(Device) TemperatureHandler + DeviceGetThermalSettings(Device, uint32) (GpuThermalSettings, Return) + DeviceGetTopologyCommonAncestor(Device, Device) (GpuTopologyLevel, Return) + DeviceGetTopologyNearestGpus(Device, GpuTopologyLevel) ([]Device, Return) + DeviceGetTotalEccErrors(Device, MemoryErrorType, EccCounterType) (uint64, Return) + DeviceGetTotalEnergyConsumption(Device) (uint64, Return) + DeviceGetUUID(Device) (string, Return) + DeviceGetUtilizationRates(Device) (Utilization, Return) + DeviceGetVbiosVersion(Device) (string, Return) + DeviceGetVgpuCapabilities(Device, DeviceVgpuCapability) (bool, Return) + DeviceGetVgpuHeterogeneousMode(Device) (VgpuHeterogeneousMode, Return) + DeviceGetVgpuInstancesUtilizationInfo(Device) (VgpuInstancesUtilizationInfo, Return) + DeviceGetVgpuMetadata(Device) (VgpuPgpuMetadata, Return) + DeviceGetVgpuProcessUtilization(Device, uint64) ([]VgpuProcessUtilizationSample, Return) + DeviceGetVgpuProcessesUtilizationInfo(Device) (VgpuProcessesUtilizationInfo, Return) + DeviceGetVgpuSchedulerCapabilities(Device) (VgpuSchedulerCapabilities, Return) + DeviceGetVgpuSchedulerLog(Device) (VgpuSchedulerLog, Return) + DeviceGetVgpuSchedulerState(Device) (VgpuSchedulerGetState, Return) + DeviceGetVgpuTypeCreatablePlacements(Device, VgpuTypeId) (VgpuPlacementList, Return) + DeviceGetVgpuTypeSupportedPlacements(Device, VgpuTypeId) (VgpuPlacementList, Return) + DeviceGetVgpuUtilization(Device, uint64) (ValueType, []VgpuInstanceUtilizationSample, Return) + DeviceGetViolationStatus(Device, PerfPolicyType) (ViolationTime, Return) + DeviceGetVirtualizationMode(Device) (GpuVirtualizationMode, Return) + DeviceIsMigDeviceHandle(Device) (bool, Return) + DeviceModifyDrainState(*PciInfo, EnableState) Return + DeviceOnSameBoard(Device, Device) (int, Return) + DevicePowerSmoothingActivatePresetProfile(Device, *PowerSmoothingProfile) Return + DevicePowerSmoothingSetState(Device, *PowerSmoothingState) Return + DevicePowerSmoothingUpdatePresetProfileParam(Device, *PowerSmoothingProfile) Return + DeviceQueryDrainState(*PciInfo) (EnableState, Return) + DeviceReadWritePRM_v1(Device, *PRMTLV_v1) Return + DeviceRegisterEvents(Device, uint64, EventSet) Return + DeviceRemoveGpu(*PciInfo) Return + DeviceRemoveGpu_v2(*PciInfo, DetachGpuState, PcieLinkState) Return + DeviceResetApplicationsClocks(Device) Return + DeviceResetGpuLockedClocks(Device) Return + DeviceResetMemoryLockedClocks(Device) Return + DeviceResetNvLinkErrorCounters(Device, int) Return + DeviceResetNvLinkUtilizationCounter(Device, int, int) Return + DeviceSetAPIRestriction(Device, RestrictedAPI, EnableState) Return + DeviceSetAccountingMode(Device, EnableState) Return + DeviceSetApplicationsClocks(Device, uint32, uint32) Return + DeviceSetAutoBoostedClocksEnabled(Device, EnableState) Return + DeviceSetClockOffsets(Device, ClockOffset) Return + DeviceSetComputeMode(Device, ComputeMode) Return + DeviceSetConfComputeUnprotectedMemSize(Device, uint64) Return + DeviceSetCpuAffinity(Device) Return + DeviceSetDefaultAutoBoostedClocksEnabled(Device, EnableState, uint32) Return + DeviceSetDefaultFanSpeed_v2(Device, int) Return + DeviceSetDramEncryptionMode(Device, *DramEncryptionInfo) Return + DeviceSetDriverModel(Device, DriverModel, uint32) Return + DeviceSetEccMode(Device, EnableState) Return + DeviceSetFanControlPolicy(Device, int, FanControlPolicy) Return + DeviceSetFanSpeed_v2(Device, int, int) Return + DeviceSetGpcClkVfOffset(Device, int) Return + DeviceSetGpuLockedClocks(Device, uint32, uint32) Return + DeviceSetGpuOperationMode(Device, GpuOperationMode) Return + DeviceSetMemClkVfOffset(Device, int) Return + DeviceSetMemoryLockedClocks(Device, uint32, uint32) Return + DeviceSetMigMode(Device, int) (Return, Return) + DeviceSetNvLinkDeviceLowPowerThreshold(Device, *NvLinkPowerThres) Return + DeviceSetNvLinkUtilizationControl(Device, int, int, *NvLinkUtilizationControl, bool) Return + DeviceSetNvlinkBwMode(Device, *NvlinkSetBwMode) Return + DeviceSetPersistenceMode(Device, EnableState) Return + DeviceSetPowerManagementLimit(Device, uint32) Return + DeviceSetPowerManagementLimit_v2(Device, *PowerValue_v2) Return + DeviceSetTemperatureThreshold(Device, TemperatureThresholds, int) Return + DeviceSetVgpuCapabilities(Device, DeviceVgpuCapability, EnableState) Return + DeviceSetVgpuHeterogeneousMode(Device, VgpuHeterogeneousMode) Return + DeviceSetVgpuSchedulerState(Device, *VgpuSchedulerSetState) Return + DeviceSetVirtualizationMode(Device, GpuVirtualizationMode) Return + DeviceValidateInforom(Device) Return + DeviceWorkloadPowerProfileClearRequestedProfiles(Device, *WorkloadPowerProfileRequestedProfiles) Return + DeviceWorkloadPowerProfileGetCurrentProfiles(Device) (WorkloadPowerProfileCurrentProfiles, Return) + DeviceWorkloadPowerProfileGetProfilesInfo(Device) (WorkloadPowerProfileProfilesInfo, Return) + DeviceWorkloadPowerProfileSetRequestedProfiles(Device, *WorkloadPowerProfileRequestedProfiles) Return + ErrorString(Return) string + EventSetCreate() (EventSet, Return) + EventSetFree(EventSet) Return + EventSetWait(EventSet, uint32) (EventData, Return) + Extensions() ExtendedInterface + GetExcludedDeviceCount() (int, Return) + GetExcludedDeviceInfoByIndex(int) (ExcludedDeviceInfo, Return) + GetVgpuCompatibility(*VgpuMetadata, *VgpuPgpuMetadata) (VgpuPgpuCompatibility, Return) + GetVgpuDriverCapabilities(VgpuDriverCapability) (bool, Return) + GetVgpuVersion() (VgpuVersion, VgpuVersion, Return) + GpmMetricsGet(*GpmMetricsGetType) Return + GpmMetricsGetV(*GpmMetricsGetType) GpmMetricsGetVType + GpmMigSampleGet(Device, int, GpmSample) Return + GpmQueryDeviceSupport(Device) (GpmSupport, Return) + GpmQueryDeviceSupportV(Device) GpmSupportV + GpmQueryIfStreamingEnabled(Device) (uint32, Return) + GpmSampleAlloc() (GpmSample, Return) + GpmSampleFree(GpmSample) Return + GpmSampleGet(Device, GpmSample) Return + GpmSetStreamingEnabled(Device, uint32) Return + GpuInstanceCreateComputeInstance(GpuInstance, *ComputeInstanceProfileInfo) (ComputeInstance, Return) + GpuInstanceCreateComputeInstanceWithPlacement(GpuInstance, *ComputeInstanceProfileInfo, *ComputeInstancePlacement) (ComputeInstance, Return) + GpuInstanceDestroy(GpuInstance) Return + GpuInstanceGetActiveVgpus(GpuInstance) (ActiveVgpuInstanceInfo, Return) + GpuInstanceGetComputeInstanceById(GpuInstance, int) (ComputeInstance, Return) + GpuInstanceGetComputeInstancePossiblePlacements(GpuInstance, *ComputeInstanceProfileInfo) ([]ComputeInstancePlacement, Return) + GpuInstanceGetComputeInstanceProfileInfo(GpuInstance, int, int) (ComputeInstanceProfileInfo, Return) + GpuInstanceGetComputeInstanceProfileInfoV(GpuInstance, int, int) ComputeInstanceProfileInfoHandler + GpuInstanceGetComputeInstanceRemainingCapacity(GpuInstance, *ComputeInstanceProfileInfo) (int, Return) + GpuInstanceGetComputeInstances(GpuInstance, *ComputeInstanceProfileInfo) ([]ComputeInstance, Return) + GpuInstanceGetCreatableVgpus(GpuInstance) (VgpuTypeIdInfo, Return) + GpuInstanceGetInfo(GpuInstance) (GpuInstanceInfo, Return) + GpuInstanceGetVgpuHeterogeneousMode(GpuInstance) (VgpuHeterogeneousMode, Return) + GpuInstanceGetVgpuSchedulerLog(GpuInstance) (VgpuSchedulerLogInfo, Return) + GpuInstanceGetVgpuSchedulerState(GpuInstance) (VgpuSchedulerStateInfo, Return) + GpuInstanceGetVgpuTypeCreatablePlacements(GpuInstance) (VgpuCreatablePlacementInfo, Return) + GpuInstanceSetVgpuHeterogeneousMode(GpuInstance, *VgpuHeterogeneousMode) Return + GpuInstanceSetVgpuSchedulerState(GpuInstance, *VgpuSchedulerState) Return + Init() Return + InitWithFlags(uint32) Return + SetVgpuVersion(*VgpuVersion) Return + Shutdown() Return + SystemEventSetCreate(*SystemEventSetCreateRequest) Return + SystemEventSetFree(*SystemEventSetFreeRequest) Return + SystemEventSetWait(*SystemEventSetWaitRequest) Return + SystemGetConfComputeCapabilities() (ConfComputeSystemCaps, Return) + SystemGetConfComputeGpusReadyState() (uint32, Return) + SystemGetConfComputeKeyRotationThresholdInfo() (ConfComputeGetKeyRotationThresholdInfo, Return) + SystemGetConfComputeSettings() (SystemConfComputeSettings, Return) + SystemGetConfComputeState() (ConfComputeSystemState, Return) + SystemGetCudaDriverVersion() (int, Return) + SystemGetCudaDriverVersion_v2() (int, Return) + SystemGetDriverBranch() (SystemDriverBranchInfo, Return) + SystemGetDriverVersion() (string, Return) + SystemGetHicVersion() ([]HwbcEntry, Return) + SystemGetNVMLVersion() (string, Return) + SystemGetNvlinkBwMode() (uint32, Return) + SystemGetProcessName(int) (string, Return) + SystemGetTopologyGpuSet(int) ([]Device, Return) + SystemRegisterEvents(*SystemRegisterEventRequest) Return + SystemSetConfComputeGpusReadyState(uint32) Return + SystemSetConfComputeKeyRotationThresholdInfo(ConfComputeSetKeyRotationThresholdInfo) Return + SystemSetNvlinkBwMode(uint32) Return + UnitGetCount() (int, Return) + UnitGetDevices(Unit) ([]Device, Return) + UnitGetFanSpeedInfo(Unit) (UnitFanSpeeds, Return) + UnitGetHandleByIndex(int) (Unit, Return) + UnitGetLedState(Unit) (LedState, Return) + UnitGetPsuInfo(Unit) (PSUInfo, Return) + UnitGetTemperature(Unit, int) (uint32, Return) + UnitGetUnitInfo(Unit) (UnitInfo, Return) + UnitSetLedState(Unit, LedColor) Return + VgpuInstanceClearAccountingPids(VgpuInstance) Return + VgpuInstanceGetAccountingMode(VgpuInstance) (EnableState, Return) + VgpuInstanceGetAccountingPids(VgpuInstance) ([]int, Return) + VgpuInstanceGetAccountingStats(VgpuInstance, int) (AccountingStats, Return) + VgpuInstanceGetEccMode(VgpuInstance) (EnableState, Return) + VgpuInstanceGetEncoderCapacity(VgpuInstance) (int, Return) + VgpuInstanceGetEncoderSessions(VgpuInstance) (int, EncoderSessionInfo, Return) + VgpuInstanceGetEncoderStats(VgpuInstance) (int, uint32, uint32, Return) + VgpuInstanceGetFBCSessions(VgpuInstance) (int, FBCSessionInfo, Return) + VgpuInstanceGetFBCStats(VgpuInstance) (FBCStats, Return) + VgpuInstanceGetFbUsage(VgpuInstance) (uint64, Return) + VgpuInstanceGetFrameRateLimit(VgpuInstance) (uint32, Return) + VgpuInstanceGetGpuInstanceId(VgpuInstance) (int, Return) + VgpuInstanceGetGpuPciId(VgpuInstance) (string, Return) + VgpuInstanceGetLicenseInfo(VgpuInstance) (VgpuLicenseInfo, Return) + VgpuInstanceGetLicenseStatus(VgpuInstance) (int, Return) + VgpuInstanceGetMdevUUID(VgpuInstance) (string, Return) + VgpuInstanceGetMetadata(VgpuInstance) (VgpuMetadata, Return) + VgpuInstanceGetRuntimeStateSize(VgpuInstance) (VgpuRuntimeState, Return) + VgpuInstanceGetType(VgpuInstance) (VgpuTypeId, Return) + VgpuInstanceGetUUID(VgpuInstance) (string, Return) + VgpuInstanceGetVmDriverVersion(VgpuInstance) (string, Return) + VgpuInstanceGetVmID(VgpuInstance) (string, VgpuVmIdType, Return) + VgpuInstanceSetEncoderCapacity(VgpuInstance, int) Return + VgpuTypeGetBAR1Info(VgpuTypeId) (VgpuTypeBar1Info, Return) + VgpuTypeGetCapabilities(VgpuTypeId, VgpuCapability) (bool, Return) + VgpuTypeGetClass(VgpuTypeId) (string, Return) + VgpuTypeGetDeviceID(VgpuTypeId) (uint64, uint64, Return) + VgpuTypeGetFrameRateLimit(VgpuTypeId) (uint32, Return) + VgpuTypeGetFramebufferSize(VgpuTypeId) (uint64, Return) + VgpuTypeGetGpuInstanceProfileId(VgpuTypeId) (uint32, Return) + VgpuTypeGetLicense(VgpuTypeId) (string, Return) + VgpuTypeGetMaxInstances(Device, VgpuTypeId) (int, Return) + VgpuTypeGetMaxInstancesPerGpuInstance(*VgpuTypeMaxInstance) Return + VgpuTypeGetMaxInstancesPerVm(VgpuTypeId) (int, Return) + VgpuTypeGetName(VgpuTypeId) (string, Return) + VgpuTypeGetNumDisplayHeads(VgpuTypeId) (int, Return) + VgpuTypeGetResolution(VgpuTypeId, int) (uint32, uint32, Return) +} + +// Device represents the interface for the nvmlDevice type. +// +//go:generate moq -out mock/device.go -pkg mock . Device:Device +type Device interface { + ClearAccountingPids() Return + ClearCpuAffinity() Return + ClearEccErrorCounts(EccCounterType) Return + ClearFieldValues([]FieldValue) Return + CreateGpuInstance(*GpuInstanceProfileInfo) (GpuInstance, Return) + CreateGpuInstanceWithPlacement(*GpuInstanceProfileInfo, *GpuInstancePlacement) (GpuInstance, Return) + FreezeNvLinkUtilizationCounter(int, int, EnableState) Return + GetAPIRestriction(RestrictedAPI) (EnableState, Return) + GetAccountingBufferSize() (int, Return) + GetAccountingMode() (EnableState, Return) + GetAccountingPids() ([]int, Return) + GetAccountingStats(uint32) (AccountingStats, Return) + GetActiveVgpus() ([]VgpuInstance, Return) + GetAdaptiveClockInfoStatus() (uint32, Return) + GetAddressingMode() (DeviceAddressingMode, Return) + GetApplicationsClock(ClockType) (uint32, Return) + GetArchitecture() (DeviceArchitecture, Return) + GetAttributes() (DeviceAttributes, Return) + GetAutoBoostedClocksEnabled() (EnableState, EnableState, Return) + GetBAR1MemoryInfo() (BAR1Memory, Return) + GetBoardId() (uint32, Return) + GetBoardPartNumber() (string, Return) + GetBrand() (BrandType, Return) + GetBridgeChipInfo() (BridgeChipHierarchy, Return) + GetBusType() (BusType, Return) + GetC2cModeInfoV() C2cModeInfoHandler + GetCapabilities() (DeviceCapabilities, Return) + GetClkMonStatus() (ClkMonStatus, Return) + GetClock(ClockType, ClockId) (uint32, Return) + GetClockInfo(ClockType) (uint32, Return) + GetClockOffsets() (ClockOffset, Return) + GetComputeInstanceId() (int, Return) + GetComputeMode() (ComputeMode, Return) + GetComputeRunningProcesses() ([]ProcessInfo, Return) + GetConfComputeGpuAttestationReport(*ConfComputeGpuAttestationReport) Return + GetConfComputeGpuCertificate() (ConfComputeGpuCertificate, Return) + GetConfComputeMemSizeInfo() (ConfComputeMemSizeInfo, Return) + GetConfComputeProtectedMemoryUsage() (Memory, Return) + GetCoolerInfo() (CoolerInfo, Return) + GetCpuAffinity(int) ([]uint, Return) + GetCpuAffinityWithinScope(int, AffinityScope) ([]uint, Return) + GetCreatableVgpus() ([]VgpuTypeId, Return) + GetCudaComputeCapability() (int, int, Return) + GetCurrPcieLinkGeneration() (int, Return) + GetCurrPcieLinkWidth() (int, Return) + GetCurrentClockFreqs() (DeviceCurrentClockFreqs, Return) + GetCurrentClocksEventReasons() (uint64, Return) + GetCurrentClocksThrottleReasons() (uint64, Return) + GetDecoderUtilization() (uint32, uint32, Return) + GetDefaultApplicationsClock(ClockType) (uint32, Return) + GetDefaultEccMode() (EnableState, Return) + GetDetailedEccErrors(MemoryErrorType, EccCounterType) (EccErrorCounts, Return) + GetDeviceHandleFromMigDeviceHandle() (Device, Return) + GetDisplayActive() (EnableState, Return) + GetDisplayMode() (EnableState, Return) + GetDramEncryptionMode() (DramEncryptionInfo, DramEncryptionInfo, Return) + GetDriverModel() (DriverModel, DriverModel, Return) + GetDriverModel_v2() (DriverModel, DriverModel, Return) + GetDynamicPstatesInfo() (GpuDynamicPstatesInfo, Return) + GetEccMode() (EnableState, EnableState, Return) + GetEncoderCapacity(EncoderType) (int, Return) + GetEncoderSessions() ([]EncoderSessionInfo, Return) + GetEncoderStats() (int, uint32, uint32, Return) + GetEncoderUtilization() (uint32, uint32, Return) + GetEnforcedPowerLimit() (uint32, Return) + GetFBCSessions() ([]FBCSessionInfo, Return) + GetFBCStats() (FBCStats, Return) + GetFanControlPolicy_v2(int) (FanControlPolicy, Return) + GetFanSpeed() (uint32, Return) + GetFanSpeedRPM() (FanSpeedInfo, Return) + GetFanSpeed_v2(int) (uint32, Return) + GetFieldValues([]FieldValue) Return + GetGpcClkMinMaxVfOffset() (int, int, Return) + GetGpcClkVfOffset() (int, Return) + GetGpuFabricInfo() (GpuFabricInfo, Return) + GetGpuFabricInfoV() GpuFabricInfoHandler + GetGpuInstanceById(int) (GpuInstance, Return) + GetGpuInstanceId() (int, Return) + GetGpuInstancePossiblePlacements(*GpuInstanceProfileInfo) ([]GpuInstancePlacement, Return) + GetGpuInstanceProfileInfo(int) (GpuInstanceProfileInfo, Return) + GetGpuInstanceProfileInfoByIdV(int) GpuInstanceProfileInfoByIdHandler + GetGpuInstanceProfileInfoV(int) GpuInstanceProfileInfoHandler + GetGpuInstanceRemainingCapacity(*GpuInstanceProfileInfo) (int, Return) + GetGpuInstances(*GpuInstanceProfileInfo) ([]GpuInstance, Return) + GetGpuMaxPcieLinkGeneration() (int, Return) + GetGpuOperationMode() (GpuOperationMode, GpuOperationMode, Return) + GetGraphicsRunningProcesses() ([]ProcessInfo, Return) + GetGridLicensableFeatures() (GridLicensableFeatures, Return) + GetGspFirmwareMode() (bool, bool, Return) + GetGspFirmwareVersion() (string, Return) + GetHostVgpuMode() (HostVgpuMode, Return) + GetIndex() (int, Return) + GetInforomConfigurationChecksum() (uint32, Return) + GetInforomImageVersion() (string, Return) + GetInforomVersion(InforomObject) (string, Return) + GetIrqNum() (int, Return) + GetJpgUtilization() (uint32, uint32, Return) + GetLastBBXFlushTime() (uint64, uint, Return) + GetMPSComputeRunningProcesses() ([]ProcessInfo, Return) + GetMarginTemperature() (MarginTemperature, Return) + GetMaxClockInfo(ClockType) (uint32, Return) + GetMaxCustomerBoostClock(ClockType) (uint32, Return) + GetMaxMigDeviceCount() (int, Return) + GetMaxPcieLinkGeneration() (int, Return) + GetMaxPcieLinkWidth() (int, Return) + GetMemClkMinMaxVfOffset() (int, int, Return) + GetMemClkVfOffset() (int, Return) + GetMemoryAffinity(int, AffinityScope) ([]uint, Return) + GetMemoryBusWidth() (uint32, Return) + GetMemoryErrorCounter(MemoryErrorType, EccCounterType, MemoryLocation) (uint64, Return) + GetMemoryInfo() (Memory, Return) + GetMemoryInfo_v2() (Memory_v2, Return) + GetMigDeviceHandleByIndex(int) (Device, Return) + GetMigMode() (int, int, Return) + GetMinMaxClockOfPState(ClockType, Pstates) (uint32, uint32, Return) + GetMinMaxFanSpeed() (int, int, Return) + GetMinorNumber() (int, Return) + GetModuleId() (int, Return) + GetMultiGpuBoard() (int, Return) + GetName() (string, Return) + GetNumFans() (int, Return) + GetNumGpuCores() (int, Return) + GetNumaNodeId() (int, Return) + GetNvLinkCapability(int, NvLinkCapability) (uint32, Return) + GetNvLinkErrorCounter(int, NvLinkErrorCounter) (uint64, Return) + GetNvLinkInfo() NvLinkInfoHandler + GetNvLinkRemoteDeviceType(int) (IntNvLinkDeviceType, Return) + GetNvLinkRemotePciInfo(int) (PciInfo, Return) + GetNvLinkState(int) (EnableState, Return) + GetNvLinkUtilizationControl(int, int) (NvLinkUtilizationControl, Return) + GetNvLinkUtilizationCounter(int, int) (uint64, uint64, Return) + GetNvLinkVersion(int) (uint32, Return) + GetNvlinkBwMode() (NvlinkGetBwMode, Return) + GetNvlinkSupportedBwModes() (NvlinkSupportedBwModes, Return) + GetOfaUtilization() (uint32, uint32, Return) + GetP2PStatus(Device, GpuP2PCapsIndex) (GpuP2PStatus, Return) + GetPciInfo() (PciInfo, Return) + GetPciInfoExt() (PciInfoExt, Return) + GetPcieLinkMaxSpeed() (uint32, Return) + GetPcieReplayCounter() (int, Return) + GetPcieSpeed() (int, Return) + GetPcieThroughput(PcieUtilCounter) (uint32, Return) + GetPdi() (Pdi, Return) + GetPerformanceModes() (DevicePerfModes, Return) + GetPerformanceState() (Pstates, Return) + GetPersistenceMode() (EnableState, Return) + GetPgpuMetadataString() (string, Return) + GetPlatformInfo() (PlatformInfo, Return) + GetPowerManagementDefaultLimit() (uint32, Return) + GetPowerManagementLimit() (uint32, Return) + GetPowerManagementLimitConstraints() (uint32, uint32, Return) + GetPowerManagementMode() (EnableState, Return) + GetPowerMizerMode_v1() (DevicePowerMizerModes_v1, Return) + GetPowerSource() (PowerSource, Return) + GetPowerState() (Pstates, Return) + GetPowerUsage() (uint32, Return) + GetProcessUtilization(uint64) ([]ProcessUtilizationSample, Return) + GetProcessesUtilizationInfo() (ProcessesUtilizationInfo, Return) + GetRemappedRows() (int, int, bool, bool, Return) + GetRepairStatus() (RepairStatus, Return) + GetRetiredPages(PageRetirementCause) ([]uint64, Return) + GetRetiredPagesPendingStatus() (EnableState, Return) + GetRetiredPages_v2(PageRetirementCause) ([]uint64, []uint64, Return) + GetRowRemapperHistogram() (RowRemapperHistogramValues, Return) + GetRunningProcessDetailList() (ProcessDetailList, Return) + GetSamples(SamplingType, uint64) (ValueType, []Sample, Return) + GetSerial() (string, Return) + GetSramEccErrorStatus() (EccSramErrorStatus, Return) + GetSramUniqueUncorrectedEccErrorCounts(*EccSramUniqueUncorrectedErrorCounts) Return + GetSupportedClocksEventReasons() (uint64, Return) + GetSupportedClocksThrottleReasons() (uint64, Return) + GetSupportedEventTypes() (uint64, Return) + GetSupportedGraphicsClocks(int) (int, uint32, Return) + GetSupportedMemoryClocks() (int, uint32, Return) + GetSupportedPerformanceStates() ([]Pstates, Return) + GetSupportedVgpus() ([]VgpuTypeId, Return) + GetTargetFanSpeed(int) (int, Return) + GetTemperature(TemperatureSensors) (uint32, Return) + GetTemperatureThreshold(TemperatureThresholds) (uint32, Return) + GetTemperatureV() TemperatureHandler + GetThermalSettings(uint32) (GpuThermalSettings, Return) + GetTopologyCommonAncestor(Device) (GpuTopologyLevel, Return) + GetTopologyNearestGpus(GpuTopologyLevel) ([]Device, Return) + GetTotalEccErrors(MemoryErrorType, EccCounterType) (uint64, Return) + GetTotalEnergyConsumption() (uint64, Return) + GetUUID() (string, Return) + GetUtilizationRates() (Utilization, Return) + GetVbiosVersion() (string, Return) + GetVgpuCapabilities(DeviceVgpuCapability) (bool, Return) + GetVgpuHeterogeneousMode() (VgpuHeterogeneousMode, Return) + GetVgpuInstancesUtilizationInfo() (VgpuInstancesUtilizationInfo, Return) + GetVgpuMetadata() (VgpuPgpuMetadata, Return) + GetVgpuProcessUtilization(uint64) ([]VgpuProcessUtilizationSample, Return) + GetVgpuProcessesUtilizationInfo() (VgpuProcessesUtilizationInfo, Return) + GetVgpuSchedulerCapabilities() (VgpuSchedulerCapabilities, Return) + GetVgpuSchedulerLog() (VgpuSchedulerLog, Return) + GetVgpuSchedulerState() (VgpuSchedulerGetState, Return) + GetVgpuTypeCreatablePlacements(VgpuTypeId) (VgpuPlacementList, Return) + GetVgpuTypeSupportedPlacements(VgpuTypeId) (VgpuPlacementList, Return) + GetVgpuUtilization(uint64) (ValueType, []VgpuInstanceUtilizationSample, Return) + GetViolationStatus(PerfPolicyType) (ViolationTime, Return) + GetVirtualizationMode() (GpuVirtualizationMode, Return) + GpmMigSampleGet(int, GpmSample) Return + GpmQueryDeviceSupport() (GpmSupport, Return) + GpmQueryDeviceSupportV() GpmSupportV + GpmQueryIfStreamingEnabled() (uint32, Return) + GpmSampleGet(GpmSample) Return + GpmSetStreamingEnabled(uint32) Return + IsMigDeviceHandle() (bool, Return) + OnSameBoard(Device) (int, Return) + PowerSmoothingActivatePresetProfile(*PowerSmoothingProfile) Return + PowerSmoothingSetState(*PowerSmoothingState) Return + PowerSmoothingUpdatePresetProfileParam(*PowerSmoothingProfile) Return + ReadWritePRM_v1(*PRMTLV_v1) Return + RegisterEvents(uint64, EventSet) Return + ResetApplicationsClocks() Return + ResetGpuLockedClocks() Return + ResetMemoryLockedClocks() Return + ResetNvLinkErrorCounters(int) Return + ResetNvLinkUtilizationCounter(int, int) Return + SetAPIRestriction(RestrictedAPI, EnableState) Return + SetAccountingMode(EnableState) Return + SetApplicationsClocks(uint32, uint32) Return + SetAutoBoostedClocksEnabled(EnableState) Return + SetClockOffsets(ClockOffset) Return + SetComputeMode(ComputeMode) Return + SetConfComputeUnprotectedMemSize(uint64) Return + SetCpuAffinity() Return + SetDefaultAutoBoostedClocksEnabled(EnableState, uint32) Return + SetDefaultFanSpeed_v2(int) Return + SetDramEncryptionMode(*DramEncryptionInfo) Return + SetDriverModel(DriverModel, uint32) Return + SetEccMode(EnableState) Return + SetFanControlPolicy(int, FanControlPolicy) Return + SetFanSpeed_v2(int, int) Return + SetGpcClkVfOffset(int) Return + SetGpuLockedClocks(uint32, uint32) Return + SetGpuOperationMode(GpuOperationMode) Return + SetMemClkVfOffset(int) Return + SetMemoryLockedClocks(uint32, uint32) Return + SetMigMode(int) (Return, Return) + SetNvLinkDeviceLowPowerThreshold(*NvLinkPowerThres) Return + SetNvLinkUtilizationControl(int, int, *NvLinkUtilizationControl, bool) Return + SetNvlinkBwMode(*NvlinkSetBwMode) Return + SetPersistenceMode(EnableState) Return + SetPowerManagementLimit(uint32) Return + SetPowerManagementLimit_v2(*PowerValue_v2) Return + SetTemperatureThreshold(TemperatureThresholds, int) Return + SetVgpuCapabilities(DeviceVgpuCapability, EnableState) Return + SetVgpuHeterogeneousMode(VgpuHeterogeneousMode) Return + SetVgpuSchedulerState(*VgpuSchedulerSetState) Return + SetVirtualizationMode(GpuVirtualizationMode) Return + ValidateInforom() Return + VgpuTypeGetMaxInstances(VgpuTypeId) (int, Return) + WorkloadPowerProfileClearRequestedProfiles(*WorkloadPowerProfileRequestedProfiles) Return + WorkloadPowerProfileGetCurrentProfiles() (WorkloadPowerProfileCurrentProfiles, Return) + WorkloadPowerProfileGetProfilesInfo() (WorkloadPowerProfileProfilesInfo, Return) + WorkloadPowerProfileSetRequestedProfiles(*WorkloadPowerProfileRequestedProfiles) Return +} + +// GpuInstance represents the interface for the nvmlGpuInstance type. +// +//go:generate moq -out mock/gpuinstance.go -pkg mock . GpuInstance:GpuInstance +type GpuInstance interface { + CreateComputeInstance(*ComputeInstanceProfileInfo) (ComputeInstance, Return) + CreateComputeInstanceWithPlacement(*ComputeInstanceProfileInfo, *ComputeInstancePlacement) (ComputeInstance, Return) + Destroy() Return + GetActiveVgpus() (ActiveVgpuInstanceInfo, Return) + GetComputeInstanceById(int) (ComputeInstance, Return) + GetComputeInstancePossiblePlacements(*ComputeInstanceProfileInfo) ([]ComputeInstancePlacement, Return) + GetComputeInstanceProfileInfo(int, int) (ComputeInstanceProfileInfo, Return) + GetComputeInstanceProfileInfoV(int, int) ComputeInstanceProfileInfoHandler + GetComputeInstanceRemainingCapacity(*ComputeInstanceProfileInfo) (int, Return) + GetComputeInstances(*ComputeInstanceProfileInfo) ([]ComputeInstance, Return) + GetCreatableVgpus() (VgpuTypeIdInfo, Return) + GetInfo() (GpuInstanceInfo, Return) + GetVgpuHeterogeneousMode() (VgpuHeterogeneousMode, Return) + GetVgpuSchedulerLog() (VgpuSchedulerLogInfo, Return) + GetVgpuSchedulerState() (VgpuSchedulerStateInfo, Return) + GetVgpuTypeCreatablePlacements() (VgpuCreatablePlacementInfo, Return) + SetVgpuHeterogeneousMode(*VgpuHeterogeneousMode) Return + SetVgpuSchedulerState(*VgpuSchedulerState) Return +} + +// ComputeInstance represents the interface for the nvmlComputeInstance type. +// +//go:generate moq -out mock/computeinstance.go -pkg mock . ComputeInstance:ComputeInstance +type ComputeInstance interface { + Destroy() Return + GetInfo() (ComputeInstanceInfo, Return) +} + +// EventSet represents the interface for the nvmlEventSet type. +// +//go:generate moq -out mock/eventset.go -pkg mock . EventSet:EventSet +type EventSet interface { + Free() Return + Wait(uint32) (EventData, Return) +} + +// GpmSample represents the interface for the nvmlGpmSample type. +// +//go:generate moq -out mock/gpmsample.go -pkg mock . GpmSample:GpmSample +type GpmSample interface { + Free() Return + Get(Device) Return + MigGet(Device, int) Return +} + +// Unit represents the interface for the nvmlUnit type. +// +//go:generate moq -out mock/unit.go -pkg mock . Unit:Unit +type Unit interface { + GetDevices() ([]Device, Return) + GetFanSpeedInfo() (UnitFanSpeeds, Return) + GetLedState() (LedState, Return) + GetPsuInfo() (PSUInfo, Return) + GetTemperature(int) (uint32, Return) + GetUnitInfo() (UnitInfo, Return) + SetLedState(LedColor) Return +} + +// VgpuInstance represents the interface for the nvmlVgpuInstance type. +// +//go:generate moq -out mock/vgpuinstance.go -pkg mock . VgpuInstance:VgpuInstance +type VgpuInstance interface { + ClearAccountingPids() Return + GetAccountingMode() (EnableState, Return) + GetAccountingPids() ([]int, Return) + GetAccountingStats(int) (AccountingStats, Return) + GetEccMode() (EnableState, Return) + GetEncoderCapacity() (int, Return) + GetEncoderSessions() (int, EncoderSessionInfo, Return) + GetEncoderStats() (int, uint32, uint32, Return) + GetFBCSessions() (int, FBCSessionInfo, Return) + GetFBCStats() (FBCStats, Return) + GetFbUsage() (uint64, Return) + GetFrameRateLimit() (uint32, Return) + GetGpuInstanceId() (int, Return) + GetGpuPciId() (string, Return) + GetLicenseInfo() (VgpuLicenseInfo, Return) + GetLicenseStatus() (int, Return) + GetMdevUUID() (string, Return) + GetMetadata() (VgpuMetadata, Return) + GetRuntimeStateSize() (VgpuRuntimeState, Return) + GetType() (VgpuTypeId, Return) + GetUUID() (string, Return) + GetVmDriverVersion() (string, Return) + GetVmID() (string, VgpuVmIdType, Return) + SetEncoderCapacity(int) Return +} + +// VgpuTypeId represents the interface for the nvmlVgpuTypeId type. +// +//go:generate moq -out mock/vgputypeid.go -pkg mock . VgpuTypeId:VgpuTypeId +type VgpuTypeId interface { + GetBAR1Info() (VgpuTypeBar1Info, Return) + GetCapabilities(VgpuCapability) (bool, Return) + GetClass() (string, Return) + GetCreatablePlacements(Device) (VgpuPlacementList, Return) + GetDeviceID() (uint64, uint64, Return) + GetFrameRateLimit() (uint32, Return) + GetFramebufferSize() (uint64, Return) + GetGpuInstanceProfileId() (uint32, Return) + GetLicense() (string, Return) + GetMaxInstances(Device) (int, Return) + GetMaxInstancesPerVm() (int, Return) + GetName() (string, Return) + GetNumDisplayHeads() (int, Return) + GetResolution(int) (uint32, uint32, Return) + GetSupportedPlacements(Device) (VgpuPlacementList, Return) +} diff --git a/vendor/modules.txt b/vendor/modules.txt index 0b053a3a..cc09dfcc 100644 --- a/vendor/modules.txt +++ b/vendor/modules.txt @@ -1,6 +1,10 @@ # github.com/Masterminds/semver/v3 v3.4.0 ## explicit; go 1.21 github.com/Masterminds/semver/v3 +# github.com/NVIDIA/go-nvml v0.13.0-1 +## explicit; go 1.20 +github.com/NVIDIA/go-nvml/pkg/dl +github.com/NVIDIA/go-nvml/pkg/nvml # github.com/NVIDIA/gpu-monitoring-tools v0.0.0-20211102125545-5a2c58442e48 ## explicit; go 1.14 github.com/NVIDIA/gpu-monitoring-tools/bindings/go/nvml From da063c7c98f8f05f54baea7ed65640e41ad653c3 Mon Sep 17 00:00:00 2001 From: Aleksei Sviridkin Date: Sat, 11 Jul 2026 01:50:33 +0300 Subject: [PATCH 4/5] fix(device_plugin): resolve reduced-catalog vGPU types via NVML, not other cards When a configured VF's card catalog is reduced and cannot resolve its type id, resolution fell back to a host-wide catalog merged across all cards. Numeric vGPU type ids are only unique within one physical card, so if the target card contributed no entry for the id and another card mapped the same id to a different profile, that entry looked unambiguous and the VF could be advertised under the wrong extended resource. Drop the host-wide catalog entirely. Resolve strictly card-local sysfs, then the parent Physical Function's own NVML supported-type list, then skip the VF with a log. NVML is authoritative per card and its supported list does not shrink as capacity is allocated, so it covers the fully consumed card without ever borrowing a sibling card's mapping. Assisted-By: Claude Signed-off-by: Aleksei Sviridkin --- pkg/device_plugin/vfio_vgpu.go | 54 +++++++++++------------------ pkg/device_plugin/vfio_vgpu_test.go | 46 ++++++++++++++---------- 2 files changed, 48 insertions(+), 52 deletions(-) diff --git a/pkg/device_plugin/vfio_vgpu.go b/pkg/device_plugin/vfio_vgpu.go index c5a3b109..a14ed92b 100644 --- a/pkg/device_plugin/vfio_vgpu.go +++ b/pkg/device_plugin/vfio_vgpu.go @@ -97,11 +97,14 @@ var isPhysicalFunction = isPhysicalFunctionFunc // PCI address, found via each VF's "physfn" symlink) rather than merged // globally across the whole host, because vGPU type ids are only unique // within one physical card - the same numeric id can mean a different -// profile on a different card, even of the same GPU model. The corner case -// of every function of a card being simultaneously configured with a -// reduced catalog that omits a given type id is not handled here; if this -// is observed in practice, a cached per-card catalog (refreshed once at -// startup) or an NVML-based lookup would be the natural fallback. +// profile on a different card, even of the same GPU model. When every +// function of a card is simultaneously configured and its catalog is +// reduced to the header, the card's own sysfs can no longer resolve the +// type; the plugin then falls back to a per-card NVML lookup (whose +// supported-type list does not shrink as capacity is allocated) and, if +// NVML is unavailable too, skips the VF rather than borrowing another +// card's mapping, since no invariant guarantees that two cards assign the +// same numeric id to the same profile. // // Discovered VFs are added to the shared iommuMap and bdfToIommuMap used by // GenericDevicePlugin, so Allocate and the health check for these devices @@ -204,28 +207,11 @@ func createVfioVGpuMap() { log.Printf("Error discovering vendor-specific VFIO vGPU VFs: %v", walkErr) } - // Host-wide catalog fallback: a fully consumed card reduces the - // creatable_vgpu_types list of EVERY one of its functions (down to the - // header), so the card's own catalog cannot resolve the types of its - // already-configured VFs. Merge all per-card catalogs into a host-wide - // one, tracking numeric ids that different cards resolve to different - // names — those stay ambiguous and are never guessed. - globalTypeNames := make(map[string]string) - ambiguousTypeIDs := make(map[string]bool) - for _, catalog := range vgpuTypeNamesByPF { - for typeID, name := range catalog { - if existing, ok := globalTypeNames[typeID]; ok { - if existing != name { - ambiguousTypeIDs[typeID] = true - } - continue - } - globalTypeNames[typeID] = name - } - } - - // NVML is the last-resort resolution source, scoped per card like the - // sysfs catalogs; results are cached per PF for the duration of the scan. + // NVML is the fallback resolution source, scoped per card like the sysfs + // catalogs and consulted only when the card's own catalog cannot resolve + // a configured type (e.g. a fully consumed card whose creatable list is + // reduced to the header on every function). Results are cached per PF for + // the duration of the scan. nvmlNamesByPF := make(map[string]map[string]string) nvmlSupportedNames := func(pfAddress string) map[string]string { if names, done := nvmlNamesByPF[pfAddress]; done { @@ -243,14 +229,16 @@ func createVfioVGpuMap() { for _, vf := range vfs { vGpuName, ok := vgpuTypeNamesByPF[vf.pfAddress][vf.typeID] if !ok { - if name, found := globalTypeNames[vf.typeID]; found && !ambiguousTypeIDs[vf.typeID] { - log.Printf("Resolved vGPU type %s on VF %s via the host-wide catalog (the card's own catalog is reduced)", vf.typeID, vf.device.addr) - vGpuName = name - } else if name, found := nvmlSupportedNames(vf.pfAddress)[vf.typeID]; found { - log.Printf("Resolved vGPU type %s on VF %s via NVML (no sysfs catalog lists it)", vf.typeID, vf.device.addr) + // The card's own sysfs catalog does not resolve this type. Fall + // back to the parent PF's NVML supported-type list, never to + // another card's catalog: numeric vGPU type ids are only unique + // within one physical card, so borrowing a sibling card's mapping + // could advertise the VF under the wrong profile. + if name, found := nvmlSupportedNames(vf.pfAddress)[vf.typeID]; found { + log.Printf("Resolved vGPU type %s on VF %s via NVML (the card's own sysfs catalog does not list it)", vf.typeID, vf.device.addr) vGpuName = name } else { - log.Printf("Error: could not resolve the name of vGPU type %s configured on VF %s (parent PF %s), skipping device", vf.typeID, vf.device.addr, vf.pfAddress) + log.Printf("Error: could not resolve the name of vGPU type %s configured on VF %s (parent PF %s) from the card's own sysfs catalog or NVML, skipping device", vf.typeID, vf.device.addr, vf.pfAddress) continue } } diff --git a/pkg/device_plugin/vfio_vgpu_test.go b/pkg/device_plugin/vfio_vgpu_test.go index 3dc4a802..b6c617bf 100644 --- a/pkg/device_plugin/vfio_vgpu_test.go +++ b/pkg/device_plugin/vfio_vgpu_test.go @@ -271,10 +271,9 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress4)) }) - It("Resolves a type via the host-wide catalog when the card's own catalog is reduced", func() { - // Fully consumed card: every function's creatable list is reduced - // to the header, so the card's own catalog cannot resolve the - // configured type. + It("Resolves a reduced-catalog type via the parent PF's NVML, never another card's catalog", func() { + // Target card is fully consumed: its creatable list is reduced to + // the header, so the card's own catalog cannot resolve the type. createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ vendor: "0x10de", driver: "nvidia", @@ -284,7 +283,10 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { currentType: "1428", creatableTypes: "ID : vGPU Name\n", }) - // A free function on ANOTHER card still lists the full catalog. + // A free function on ANOTHER card maps the SAME numeric id to a + // DIFFERENT profile. The target card contributed no value, so this + // entry would look "unambiguous" to a host-wide merge, but numeric + // ids are only card-local: it must never be borrowed. createFakeDevice(vfioVfAddress5, fakeDeviceOptions{ vendor: "0x10de", driver: "nvidia", @@ -292,14 +294,24 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { iommuGroup: "81", numaNode: "0", currentType: "0", - creatableTypes: "1428 : " + vfioVgpuTypeName1 + "\n", + creatableTypes: "1428 : NVIDIA OTHER-NAME\n", }) + // NVML is the authoritative per-PF source and reports the correct + // profile for the target card's id 1428. + resolveVgpuTypeNamesViaNVML = func(pfAddress string) (map[string]string, error) { + if pfAddress == vfioPfAddress { + return map[string]string{"1428": vfioVgpuResourceName1}, nil + } + return map[string]string{}, nil + } createVfioVGpuMap() Expect(vfioVGpuMap).To(HaveLen(1)) Expect(vfioVGpuMap[vfioVgpuResourceName1]).To(HaveLen(1)) Expect(vfioVGpuMap[vfioVgpuResourceName1][0].addr).To(Equal(vfioVfAddress1)) + // The other card's mapping for id 1428 was never borrowed. + Expect(vfioVGpuMap).ToNot(HaveKey("NVIDIA_OTHER-NAME")) }) It("Resolves a type via NVML when no sysfs catalog lists it", func() { @@ -347,8 +359,9 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { Expect(bdfToIommuMap).ToNot(HaveKey(vfioVfAddress1)) }) - It("Skips a VF whose type maps to conflicting names across cards", func() { - // Fully consumed card with a reduced catalog. + It("Skips a reduced-catalog VF when NVML is unavailable, never borrowing another card's catalog", func() { + // Target card is fully consumed: its creatable list is reduced to + // the header. createFakeDevice(vfioVfAddress1, fakeDeviceOptions{ vendor: "0x10de", driver: "nvidia", @@ -358,8 +371,9 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { currentType: "1428", creatableTypes: "ID : vGPU Name\n", }) - // Two other cards resolve the same numeric id to DIFFERENT names — - // the host-wide fallback must refuse to guess. + // Another card maps the same numeric id to a DIFFERENT profile. + // With NVML unavailable, the target VF must be skipped rather than + // advertised under this other card's (card-local) mapping. createFakeDevice(vfioVfAddress5, fakeDeviceOptions{ vendor: "0x10de", driver: "nvidia", @@ -367,17 +381,11 @@ var _ = Describe("Vendor-specific VFIO vGPU", func() { iommuGroup: "81", numaNode: "0", currentType: "0", - creatableTypes: "1428 : " + vfioVgpuTypeName1 + "\n", - }) - createFakeDevice("0000:c1:00.4", fakeDeviceOptions{ - vendor: "0x10de", - driver: "nvidia", - physfn: "0000:c1:00.0", - iommuGroup: "91", - numaNode: "0", - currentType: "0", creatableTypes: "1428 : NVIDIA OTHER-NAME\n", }) + resolveVgpuTypeNamesViaNVML = func(pfAddress string) (map[string]string, error) { + return nil, errors.New("nvml unavailable") + } createVfioVGpuMap() From 158ffef9afcd60128e7adbce65125f3f13ba76a7 Mon Sep 17 00:00:00 2001 From: Aleksei Sviridkin Date: Sat, 11 Jul 2026 01:50:44 +0300 Subject: [PATCH 5/5] feat(deployment): add NVML-fallback manifest and document its requirements The NVML name-resolution fallback needs libnvidia-ml.so.1 and the NVIDIA device nodes. The default DaemonSet mounts only /dev/vfio and runs unprivileged, so on a fully consumed node a restart could resolve no profile names and advertise no VFs. Add a dedicated manifest that provides them via a single-file bind of libnvidia-ml.so.1 plus the NVIDIA device nodes under privileged: true, leaving the default manifest minimal and unprivileged. Document in the README both the minimally privileged runtimeClassName: nvidia path and the hostPath variant, why the library must be bound as a single file rather than the whole host lib directory, and which device nodes the plugin's NVML queries actually require. Also document that vGPU profiles must be configured before the plugin starts, since it scans the Virtual Functions once at startup. Assisted-By: Claude Signed-off-by: Aleksei Sviridkin --- README.md | 18 ++++ ...vidia-kubevirt-gpu-device-plugin-nvml.yaml | 99 +++++++++++++++++++ .../nvidia-kubevirt-gpu-device-plugin.yaml | 6 ++ 3 files changed, 123 insertions(+) create mode 100644 manifests/nvidia-kubevirt-gpu-device-plugin-nvml.yaml diff --git a/README.md b/README.md index f5f49cbf..1fefd154 100644 --- a/README.md +++ b/README.md @@ -211,6 +211,24 @@ kubectl apply -f nvidia-kubevirt-gpu-device-plugin.yaml Example YAML files for creating VMs with GPU/vGPU are in the `examples` folder +#### vGPU profile timing (vendor-specific VFIO) + +The plugin scans the vendor-specific VFIO vGPU Virtual Functions once at startup and does not re-scan while running. Every profile a node should advertise must therefore be configured on its Virtual Functions (step 3 of "Preparing a GPU to be used in vGPU mode (Ada Lovelace / Hopper and newer)") **before** the device plugin pod starts. A Virtual Function whose profile is created or changed after startup is not picked up until the pod restarts. After a node reboot, order the plugin after whatever recreates the Virtual Functions (for example an init container that waits until the count of configured Virtual Functions is non-zero and stable) so it discovers the full set. Dynamic rediscovery of profiles created or changed while the plugin is running is tracked as a separate follow-up (PR link to be added). + +#### NVML fallback (fully consumed nodes) + +The profile name of a configured Virtual Function is normally read from its card's `creatable_vgpu_types` catalog. On a fully consumed card that catalog is reduced to its header on every function, so the plugin resolves the configured type ids through NVML instead (`GetSupportedVgpus`, whose list does not shrink as capacity is allocated). This path is reached only on fully consumed nodes; the default manifest does not enable it. + +NVML needs two things inside the container: + +- **`libnvidia-ml.so.1`** from the host driver, loadable by the dynamic linker. Bind the **single file**, not the host library directory: putting the host lib directory on `LD_LIBRARY_PATH` drags the host glibc into the container and breaks the dynamic linker. This single-file requirement was confirmed on hardware. +- **The NVIDIA device nodes** the queries touch: `/dev/nvidiactl` (the NVML control device) and the per-GPU `/dev/nvidiaN` for each physical card. The management-only queries this plugin makes (`DeviceGetHandleByPciBusId`, `GetSupportedVgpus`, `GetName`) do not open a CUDA context or MIG capabilities, so `/dev/nvidia-uvm`, `/dev/nvidia-uvm-tools` and `/dev/nvidia-caps` are not needed. `/dev/nvidiactl` plus the per-GPU nodes were confirmed on hardware; the narrower "uvm/caps not required" scope is inferred from the set of NVML calls above, not separately tested. + +There are two supported ways to provide these: + +1. **NVIDIA Container Toolkit (`runtimeClassName: nvidia`) — minimally privileged, recommended.** When the toolkit is installed on the node, set `runtimeClassName: nvidia` and the container env `NVIDIA_VISIBLE_DEVICES=all` and `NVIDIA_DRIVER_CAPABILITIES=utility` (the `utility` capability is the one that provides NVML). The toolkit's runtime hook then injects `libnvidia-ml.so.1` and the device nodes and adds the matching device-cgroup rules, so the container stays non-privileged. This is the standard toolkit mechanism; the exact `runtimeClassName` form was not part of this feature's hardware validation. +2. **hostPath — privileged.** `manifests/nvidia-kubevirt-gpu-device-plugin-nvml.yaml` single-file-binds `libnvidia-ml.so.1`, sets `LD_LIBRARY_PATH`, mounts only `/dev/nvidiactl` and the per-GPU `/dev/nvidiaN` nodes (not the whole host `/dev`) and runs `privileged: true`. A hostPath device node is not added to the container's device cgroup, so a non-privileged container is denied when it opens the node regardless of the mount, and a plain pod spec has no field to grant a single host device node; `privileged: true` is the only pod-spec-native way to make host device nodes usable without the toolkit. The hardware validation of the fallback used this hostPath + privileged mechanism with the whole host `/dev` mounted; this manifest narrows that to `/dev/nvidiactl` plus the per-GPU nodes, which are the only device nodes the NVML calls above use. Point the `libnvidia-ml.so.1` hostPath at wherever the host driver installed it (find it with `ldconfig -p | grep libnvidia-ml`). + ### Build Build executable binary using make: diff --git a/manifests/nvidia-kubevirt-gpu-device-plugin-nvml.yaml b/manifests/nvidia-kubevirt-gpu-device-plugin-nvml.yaml new file mode 100644 index 00000000..d3fe35cd --- /dev/null +++ b/manifests/nvidia-kubevirt-gpu-device-plugin-nvml.yaml @@ -0,0 +1,99 @@ +# DaemonSet variant that enables the NVML name-resolution fallback. +# +# The default manifest (nvidia-kubevirt-gpu-device-plugin.yaml) mounts only +# /dev/vfio and is enough for every node where at least one function of each +# card still lists its creatable_vgpu_types catalog. On a fully consumed node +# every card's sysfs catalog is reduced to the header, and the plugin then +# resolves configured vGPU type ids through NVML instead. NVML needs the host +# driver library (libnvidia-ml.so.1) and the NVIDIA device nodes, which the +# default manifest does not provide. +# +# This variant provides them via hostPath. Because a device node mounted via +# hostPath is not added to the container's device cgroup, opening it requires +# privileged: true. If the NVIDIA Container Toolkit is installed on the node, +# prefer the runtimeClassName: nvidia approach documented in the README, which +# injects the library and the device nodes (with the matching device-cgroup +# rules) without privileged. See the README section "NVML fallback (fully +# consumed nodes)" for the trade-offs. +--- +apiVersion: apps/v1 +kind: DaemonSet +metadata: + name: nvidia-kubevirt-gpu-dp-daemonset + namespace: kube-system +spec: + selector: + matchLabels: + name: nvidia-kubevirt-gpu-dp-ds + template: + metadata: + labels: + name: nvidia-kubevirt-gpu-dp-ds + spec: + priorityClassName: system-node-critical + tolerations: + # Allow this pod to be rescheduled while the node is in "critical add-ons only" mode. + # This, along with the annotation above marks this pod as a critical add-on. + - key: CriticalAddonsOnly + operator: Exists + containers: + - name: nvidia-kubevirt-gpu-dp-ctr + image: nvcr.io/nvidia/kubevirt-gpu-device-plugin:v1.5.0 + securityContext: + # The NVML fallback opens /dev/nvidiactl and the per-GPU /dev/nvidiaN + # nodes. A hostPath device node is not placed in the container's + # device cgroup, so a non-privileged container is denied access to it + # regardless of the mount, and a plain pod spec has no field to grant + # a single host device node. privileged: true is used here as the + # only pod-spec-native way to make host device nodes accessible + # without the NVIDIA Container Toolkit; for a non-privileged + # alternative use runtimeClassName: nvidia (see the README). + privileged: true + env: + # Make the single-file bind of the host driver library loadable by + # the dynamic linker. Do NOT add the host lib directory itself to the + # search path: it drags the host glibc into the container and breaks + # the linker. + - name: LD_LIBRARY_PATH + value: /host-driver-lib + volumeMounts: + - name: device-plugin + mountPath: /var/lib/kubelet/device-plugins + - name: vfio + mountPath: /dev/vfio + # Single file, not the whole host lib directory - see LD_LIBRARY_PATH. + - name: nvml-lib + mountPath: /host-driver-lib/libnvidia-ml.so.1 + readOnly: true + # NVIDIA device nodes for NVML: the control node plus one node per + # physical GPU. Only these are needed - the plugin's NVML queries do + # not open /dev/nvidia-uvm* or /dev/nvidia-caps, and the whole host + # /dev is deliberately NOT mounted. This lists /dev/nvidia0 for a + # single-GPU node; add a CharDevice volume and mount for each + # additional GPU (/dev/nvidia1, /dev/nvidia2, ...). + - name: nvidiactl + mountPath: /dev/nvidiactl + - name: nvidia0 + mountPath: /dev/nvidia0 + volumes: + - name: device-plugin + hostPath: + path: /var/lib/kubelet/device-plugins + - name: vfio + hostPath: + path: /dev/vfio + - name: nvidiactl + hostPath: + path: /dev/nvidiactl + type: CharDevice + - name: nvidia0 + hostPath: + path: /dev/nvidia0 + type: CharDevice + # Point this at wherever the host driver installed libnvidia-ml.so.1. + # Debian/Ubuntu use /usr/lib/x86_64-linux-gnu; other distributions + # differ (find it with: ldconfig -p | grep libnvidia-ml). + - name: nvml-lib + hostPath: + path: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 + type: File diff --git a/manifests/nvidia-kubevirt-gpu-device-plugin.yaml b/manifests/nvidia-kubevirt-gpu-device-plugin.yaml index e756a12b..ae5e1b9e 100644 --- a/manifests/nvidia-kubevirt-gpu-device-plugin.yaml +++ b/manifests/nvidia-kubevirt-gpu-device-plugin.yaml @@ -1,3 +1,9 @@ +# This manifest mounts only /dev/vfio and runs unprivileged. It covers every +# node where each card still exposes its creatable_vgpu_types catalog on at +# least one function. On a fully consumed node the plugin resolves vGPU type +# names through NVML instead, which needs the host driver library and the +# NVIDIA device nodes - use nvidia-kubevirt-gpu-device-plugin-nvml.yaml (see +# the README section "NVML fallback (fully consumed nodes)") on such nodes. apiVersion: apps/v1 kind: DaemonSet metadata: