Skip to content

Fix: parallelize A5 dummy task resolution #1679

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

Open
yanghaoran29 wants to merge 1 commit into
base: main
Choose a base branch
from
+6 −4
Open

Fix: parallelize A5 dummy task resolution #1679

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
10 changes: 6 additions & 4 deletions src/a5/runtime/tensormap_and_ringbuffer/runtime/scheduler/scheduler_dispatch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1018,13 +1018,15 @@ int32_t SchedulerContext::resolve_and_dispatch(Runtime *runtime, int32_t thread_
continue;
}

// Phase 3: Drain dummy ready queue (thread 0 only).
// Phase 3: Drain dummy ready queue (S0/S1/S2/S3).
//
// Dependency-only tasks bypass AICore dispatch: they go through the
// scheduler so fanin/fanout edges stay consistent, but completion is
// signalled inline here. Pinned to thread 0 to avoid cross-thread races.
if (thread_idx == 0) {
constexpr int DUMMY_DRAIN_BATCH = 16;
// signalled inline here. The ready queue is MPMC, and the fanout path
// uses per-slot locks/atomics, so multiple scheduler threads can share
// the dependency-only resolve work.
if (thread_idx < 4) {
constexpr int DUMMY_DRAIN_BATCH = 8;
PTO2TaskSlotState *dummy_batch[DUMMY_DRAIN_BATCH];
int dummy_got = sched_->dummy_ready_queue.pop_batch(dummy_batch, DUMMY_DRAIN_BATCH);
#if SIMPLER_DFX
Expand Down
Loading