Repository navigation
Conversation
Signed-off-by: Jack Chang <jianbinc@nvidia.com>
shjwudp
force-pushed
the
fix/mfsdp-finalize-scope-stream-capture
branch
from
November 6, 2026 06:23
156ea5e to
0a02887
Compare
shjwudp
changed the title
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Combined 1F1B bypasses
FsdpModule.pre_backward(), so its first gradient staging allocation can occur on a reduction stream that has not joined CUDA-graph capture.Join the inner reduction stream from the current compute stream when entering
microbatch(). This reuses the existing MCoreno_sync()scope without adding scheduler hooks or per-module waits.Validation: Black 26, isort, Ruff, and diff checks passed. On NSC-LHR (4 GB300 GPUs, NeMo 26.08), both the regression and microbatch-scope tests passed on all four ranks. Removing only the added wait reproduced
cudaErrorStreamCaptureIsolationon all four ranks. Slurm642295:COMPLETED, exit 0. This is functional validation, not CI-image parity.Related: #9000. Regression test: #8683.