diff --git a/triton/ref/gpu.rst b/triton/ref/gpu.rst index 1d3fdda61..91fe66979 100644 --- a/triton/ref/gpu.rst +++ b/triton/ref/gpu.rst @@ -7,6 +7,7 @@ NVIDIA B300(*) | ``b300`` | ``288`` | 10.3 (``103``) | 16 | gpu[64-65] | 8 | 18944 | ``gpu-b300-288g-ellis``, ``gpu-b300-288g-short`` | NVIDIA H200(*) | ``h200`` | ``141`` | 9.0 (``90``) | 112 | gpu[50-63] | 8 | 16896 | ``gpu-h200-141g-ellis``, ``gpu-h200-141g-short`` | NVIDIA H200(**) | ``h200_2g.35gb`` | ``35`` | 9.0 (``90``) | 24 | gpu[49] | 24 | 4224 | ``gpu-h200-35g-ia-ellis``, ``gpu-h200-35g-ia`` | + NVIDIA :doc:`Grace-H200(+) ` | ``h200`` | ``141`` | 9.0 (``90``) | 4 | gpuarm[1-2] | 2 | 16896 | ``gpu-grace-h200-141g`` | NVIDIA H100 | ``h100`` | ``80`` | 9.0 (``90``) | 16 | gpu[45-48] | 4 | 16896 | ``gpu-h100-80g`` | NVIDIA A100 | ``a100`` | ``80`` | 8.0 (``80``) | 56 | gpu[11-17,38-44] | 4 | 7936 | ``gpu-a100-80g`` | NVIDIA V100 | ``v100`` | ``32`` | 7.0 (``70``) | 40 | gpu[28-37] | 4 | 5120 | ``gpu-v100-32g`` | @@ -27,3 +28,5 @@ preempted if a job requiring the resources comes in from the Ellis queue. (**) These GPUs are split from a single GPU with NVIDIA's `Multi-Instance GPU `__-feature. + +(+) These computers have the Nvidia Grace CPU, an ARM based cpu. Normal software compiled for x86 does not run on these nodes. See :doc:`/triton/usage/gracehopper`. diff --git a/triton/ref/hardware.rst b/triton/ref/hardware.rst index 933ac7450..c5d4d1d42 100644 --- a/triton/ref/hardware.rst +++ b/triton/ref/hardware.rst @@ -21,3 +21,4 @@ gpu[49] | 1 | Dell PowerEdge XE9680 | 2025 | emerald avx2 h200 hopper | 2x32 core `Xeon® Platinum 8562Y+ `__ 2.8GHz | 2048GB DDR5-5600 | HDR | 8x `H200 SXM `__ each split to 3x35GB | 20 TB SSD gpu[50-63] | 14 | Dell PowerEdge XE9680 | 2025 | emerald avx2 h200 hopper | 2x32 core `Xeon® Platinum 8562Y+ `__ 2.8GHz | 2048GB DDR5-5600 | HDR | 8x `H200 SXM `__ 141GB | 20 TB SSD gpu[64-65] | 2 | Dell PowerEdge XE9780 | 2026 | granite avx2 b300 blackwell | 2x64 core `Xeon® 6767P `__ 2.4GHz | 3072GB DDR5-6400 | NDR | 8x `B300 SXM6 AC `__ 288GB | 28 TB SSD + gpuarm[1-2] | 2 | Supermicro ARS-221GL-NHIR | 2026 | NVIDIA :doc:`grace h200 hopper ` | 2x72 core `Grace A02 Neoverse-V2`__ 3.4GHz | 1318GB LPDDR5-6400 | NDR | 2x H200 `H200 SXM `__ 144GB HBM3e| 21 TB SSD diff --git a/triton/usage/gracehopper.rst b/triton/usage/gracehopper.rst new file mode 100644 index 000000000..3cffda4a0 --- /dev/null +++ b/triton/usage/gracehopper.rst @@ -0,0 +1,13 @@ +Grace Hopper Super Chips +======================== + + +Hardware in Triton +------------------ + +Triton has currently two compute nodes (gpuarm[1,2]) with the Grace-Hopper superchip. +To gain access, please use the slurm partition: ``--partition gpu-grace-h200-141g``. + +Arm vs x86 +---------- +