Skip to content

Refactor: encapsulate dep pool operations and reorder orchestrator pipeline #320

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ChaoZheng109 merged 1 commit into from
Mar 18, 2026
Merged

Refactor: encapsulate dep pool operations and reorder orchestrator pipeline #320

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion src/a2a3/platform/onboard/host/host_regs.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -116,7 +116,7 @@ static void get_aicore_regs(std::vector<int64_t>& regs, uint64_t device_id) {
if (rt != 0) {
LOG_ERROR("get_aicore_reg_info failed, using placeholder addresses");
// Fallback: generate placeholder addresses
for (int i = 0; i < DAV_2201::PLATFORM_MAX_PHYSICAL_CORES; i++) {
for (uint32_t i = 0; i < DAV_2201::PLATFORM_MAX_PHYSICAL_CORES; i++) {
aic.push_back(0xDEADBEEF00000000ULL + (i * 0x800000)); // 8M stride
aiv.push_back(0xDEADBEEF00000000ULL + (i * 0x800000) + 0x100000);
aiv.push_back(0xDEADBEEF00000000ULL + (i * 0x800000) + 0x200000);
Expand Down
67 changes: 43 additions & 24 deletions src/a2a3/runtime/tensormap_and_ringbuffer/aicpu/aicpu_executor.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1777,34 +1777,53 @@ int32_t AicpuExecutor::run(Runtime* runtime) {
// Print orchestrator profiling data
#if PTO2_ORCH_PROFILING
PTO2OrchProfilingData p = pto2_orchestrator_get_profiling();
uint64_t total = p.sync_cycle + p.alloc_cycle + p.params_cycle +
p.lookup_cycle + p.heap_cycle + p.insert_cycle +
p.fanin_cycle;
uint64_t total = p.sync_cycle + p.alloc_cycle + p.params_cycle + p.lookup_cycle + p.heap_cycle +
p.insert_cycle + p.fanin_cycle;
if (total == 0) total = 1; // avoid div-by-zero
DEV_ALWAYS("Thread %d: === Orchestrator Profiling: %lld tasks, total=%.3fus ===", thread_idx,
(long long)p.submit_count, cycles_to_us(total));
DEV_ALWAYS("Thread %d: sync_tensormap : %.3fus (%.1f%%)", thread_idx, cycles_to_us(p.sync_cycle), p.sync_cycle * 100.0 / total);
DEV_ALWAYS("Thread %d: task_ring_alloc: %.3fus (%.1f%%) work=%.3fus wait=%.3fus atomics=%llu", thread_idx,
cycles_to_us(p.alloc_cycle), p.alloc_cycle * 100.0 / total,
cycles_to_us(p.alloc_cycle - p.alloc_wait_cycle), cycles_to_us(p.alloc_wait_cycle),
DEV_ALWAYS("Thread %d: === Orchestrator Profiling: %lld tasks, total=%.3fus ===",
thread_idx,
(long long)p.submit_count,
cycles_to_us(total));
DEV_ALWAYS("Thread %d: task_ring_alloc: %.3fus (%.1f%%) work=%.3fus wait=%.3fus atomics=%llu",
thread_idx,
cycles_to_us(p.alloc_cycle),
p.alloc_cycle * 100.0 / total,
cycles_to_us(p.alloc_cycle - p.alloc_wait_cycle),
cycles_to_us(p.alloc_wait_cycle),
(unsigned long long)p.alloc_atomic_count);
DEV_ALWAYS("Thread %d: param_copy : %.3fus (%.1f%%) atomics=%llu", thread_idx,
cycles_to_us(p.params_cycle), p.params_cycle * 100.0 / total,
(unsigned long long)p.params_atomic_count);
DEV_ALWAYS("Thread %d: lookup+dep : %.3fus (%.1f%%)", thread_idx, cycles_to_us(p.lookup_cycle), p.lookup_cycle * 100.0 / total);
DEV_ALWAYS("Thread %d: heap_alloc : %.3fus (%.1f%%) work=%.3fus wait=%.3fus atomics=%llu", thread_idx,
cycles_to_us(p.heap_cycle), p.heap_cycle * 100.0 / total,
cycles_to_us(p.heap_cycle - p.heap_wait_cycle), cycles_to_us(p.heap_wait_cycle),
DEV_ALWAYS("Thread %d: heap_alloc : %.3fus (%.1f%%) work=%.3fus wait=%.3fus atomics=%llu",
thread_idx,
cycles_to_us(p.heap_cycle),
p.heap_cycle * 100.0 / total,
cycles_to_us(p.heap_cycle - p.heap_wait_cycle),
cycles_to_us(p.heap_wait_cycle),
(unsigned long long)p.heap_atomic_count);
DEV_ALWAYS("Thread %d: tensormap_ins : %.3fus (%.1f%%)", thread_idx, cycles_to_us(p.insert_cycle), p.insert_cycle * 100.0 / total);
DEV_ALWAYS("Thread %d: fanin+ready : %.3fus (%.1f%%) work=%.3fus wait=%.3fus atomics=%llu", thread_idx,
cycles_to_us(p.fanin_cycle), p.fanin_cycle * 100.0 / total,
cycles_to_us(p.fanin_cycle - p.fanin_wait_cycle), cycles_to_us(p.fanin_wait_cycle),
DEV_ALWAYS("Thread %d: sync_tensormap : %.3fus (%.1f%%)",
thread_idx,
cycles_to_us(p.sync_cycle),
p.sync_cycle * 100.0 / total);
DEV_ALWAYS("Thread %d: lookup+dep : %.3fus (%.1f%%)",
thread_idx,
cycles_to_us(p.lookup_cycle),
p.lookup_cycle * 100.0 / total);
DEV_ALWAYS("Thread %d: tensormap_ins : %.3fus (%.1f%%)",
thread_idx,
cycles_to_us(p.insert_cycle),
p.insert_cycle * 100.0 / total);
DEV_ALWAYS("Thread %d: param_copy : %.3fus (%.1f%%) atomics=%llu",
thread_idx,
cycles_to_us(p.params_cycle),
p.params_cycle * 100.0 / total,
(unsigned long long)p.params_atomic_count);
DEV_ALWAYS("Thread %d: fanin+ready : %.3fus (%.1f%%) work=%.3fus wait=%.3fus atomics=%llu",
thread_idx,
cycles_to_us(p.fanin_cycle),
p.fanin_cycle * 100.0 / total,
cycles_to_us(p.fanin_cycle - p.fanin_wait_cycle),
cycles_to_us(p.fanin_wait_cycle),
(unsigned long long)p.fanin_atomic_count);
DEV_ALWAYS("Thread %d: scope_end : %.3fus atomics=%llu", thread_idx,
cycles_to_us(p.scope_end_cycle),
(unsigned long long)p.scope_end_atomic_count);
DEV_ALWAYS("Thread %d: avg/task : %.3fus", thread_idx,
DEV_ALWAYS("Thread %d: avg/task : %.3fus",
thread_idx,
p.submit_count > 0 ? cycles_to_us(total) / p.submit_count : 0.0);

#if PTO2_TENSORMAP_PROFILING
Expand Down
1 change: 0 additions & 1 deletion src/a2a3/runtime/tensormap_and_ringbuffer/docs/MULTI_RING.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,6 @@ PTO2DepListPool dep_pool;

// After: per-ring array
PTO2RingSet rings[PTO2_MAX_RING_DEPTH];
PTO2DepListEntry* dep_pool_cur_entries[PTO2_MAX_RING_DEPTH];
int32_t dep_pool_last_reclaimed[PTO2_MAX_RING_DEPTH];
```

Expand Down
Loading
Loading