Feat: Enable FSDP manual registration #1828
Merged
Conversation
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
youngeunkwon0405
requested review from
a team and
malay-nagda
as code owners
erhoo82
previously approved these changes
Dec 31, 2025
Contributor
Author
|
/ok to test 881fa7e |
erhoo82
previously approved these changes
Dec 31, 2025
Contributor
Author
|
/ok to test 3ee00c0 |
Contributor
Author
|
/ok to test f7114bb |
Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> draft Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> test Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> env Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> test Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> print Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> print Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> a Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> add nccl_ub Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> remove prints Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> a Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> fix Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> minor fixes Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> fix type Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> fix Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> type error Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> fix Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> revert temp commet Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> add sample recipe Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> enhance readability Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> fix lint error Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com> white space Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com>
youngeunkwon0405
force-pushed
the
youngeunk/fsdp-manual-reg
branch
from
f7114bb to
9beb822
Compare
Contributor
Author
|
/ok to test 9beb822 |
ko3n1g
approved these changes
Jan 6, 2026
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.
What does this PR do ?
Enabling FSDP manual registration for fast NCCL symmetric registration
Related Megatron-LM PRs
Changelog
GitHub Actions CI
See the CI sectionin the Contributing doc for how to trigger the CI. A Nvidia developer will need to approve and trigger the CI for external contributors.
Before your PR is "Ready for review"
Pre checks:
If you haven't finished some of the above items you can still open "Draft" PR.
Additional Information