Skip to content

fix an intermittent accuracy isse #3233

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
dishengbin wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix an intermittent accuracy isse #3233

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
9 changes: 5 additions & 4 deletions examples/python/CuTeDSL/cute/blackwell_geforce/kernel/dense_gemm/dense_gemm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -869,10 +869,7 @@ def kernel(
tRS_sD[(None, None, None, epi_buffer)],
)

cute.arch.fence_proxy(
"async.shared",
space="cta",
)
cute.arch.fence_view_async_shared()
# barrier for sync
self.epilog_sync_barrier.arrive_and_wait()

Expand All @@ -888,6 +885,10 @@ def kernel(
tma_store_pipeline.producer_commit()
tma_store_pipeline.producer_acquire()

# Do not let non-store warps reuse sC until warp 0 has
# committed the TMA store and acquired the next store slot.
self.epilog_sync_barrier.arrive_and_wait()

# Advance to the next work tile
tile_sched.advance_to_next_work()
work_tile = tile_sched.get_current_work()
Expand Down