Skip to content

[JAX] Fix classification of sync-tagged collective start instructions #3425

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
KshitijLakhani merged 1 commit into from
Aug 26, 2026
+7 −4
Merged

[JAX] Fix classification of sync-tagged collective start instructions #3425

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
11 changes: 7 additions & 4 deletions tests/jax/distributed_test_base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -168,13 +168,16 @@ def count_collectives(splitted_hlo):
is_sync_collective = "collective_backend_config" in line and sync_symb in line

if is_async_start or is_sync_collective:
if is_sync_collective:
is_all_reduce = re.search(r"\ball-reduce\s*\(", line)
is_all_gather = re.search(r"\ball-gather\s*\(", line)
else:
# Some direct *-start instructions are tagged with
# `"is_sync":true`. Classify the explicit async form first so
# it is not mistaken for an unsuffixed synchronous collective.
if is_async_start:
called_collective = get_called_collective_type(line)
is_all_reduce = COLL_AR_KEY in txt[0] or called_collective == COLL_AR_KEY
is_all_gather = COLL_AG_KEY in txt[0] or called_collective == COLL_AG_KEY
else:
is_all_reduce = re.search(r"\ball-reduce\s*\(", line)
is_all_gather = re.search(r"\ball-gather\s*\(", line)

if is_all_reduce:
result[COLL_AR_KEY] += count_bytes(txt)
Expand Down
Loading