Skip to content

Fix: nccl-ub in ddp path#3181

Merged
yaox12 merged 3 commits into
NVIDIA:devfrom
youngeunkwon0405:fix_nccl_ub_in_ddp
Feb 1, 2026
Merged

Fix: nccl-ub in ddp path#3181
yaox12 merged 3 commits into
NVIDIA:devfrom
youngeunkwon0405:fix_nccl_ub_in_ddp

Conversation

@youngeunkwon0405

@youngeunkwon0405 youngeunkwon0405 commented Jan 30, 2026 •
edited
Loading

Copy link
Copy Markdown
Member

What does this PR do ?

PyTorch memory registration doesn't invoke the lazy initialization of the communication group anymore.
We need to manually trigger lazy initialization of the communication group for the buffer registration.

Contribution process

flowchart LR
    A[Pre-checks] --> B[PR Tests]
    subgraph Code Review/Approval
        C1[Expert Review] --> C2[Final Review]
    end
    B --> C1
    C2 --> D[Merge]
Loading

Pre-checks

  • I want this PR in a versioned release and have added the appropriate Milestone (e.g., Core 0.8)
  • I have added relevant unit tests
  • I have added relevant functional tests
  • I have added proper typing to my code Typing guidelines
  • I have added relevant documentation
  • I have run the autoformatter.sh on my PR

Code review

The following process is enforced via the CODEOWNERS file for changes into megatron/core. For changes outside of megatron/core, it is up to the PR author whether or not to tag the Final Reviewer team.

For MRs into `main` branch

Feel free to message or comment the @mcore-oncall to help accelerate your merge into main. The less complex your PR is, the faster it will be approved and merged!

(Step 1): Add PR label Expert Review

(Step 2): Collect the expert reviewers reviews

  1. Attach the Expert Review label when your PR is ready for review.
  2. GitHub auto-assigns expert reviewers based on your changes. They will get notified and pick up your PR soon.

⚠️ Only proceed to the next step once all reviewers have approved, merge-conflict are resolved and the CI is passing.
Final Review might get declined if these requirements are not fulfilled.

(Step 3): Final Review

  1. Add Final Review label
  2. GitHub auto-assigns final reviewers based on your changes. They will get notified and pick up your PR soon.

(Optional Step 4): Cherry-pick into release branch

If this PR also needs to be merged into core_r* release branches, after this PR has been merged, select Cherry-pick to open a new PR into the release branch.

For MRs into `dev` branch The proposed review process for `dev` branch is under active discussion.

MRs are mergable after one approval by either eharper@nvidia.com or zijiey@nvidia.com.

Merging your PR

Any member of core-adlr and core-nemo will be able to merge your PR.

@youngeunkwon0405
fix
c7039b9 
Signed-off-by: Youngeun Kwon <youngeunk@nvidia.com>
@youngeunkwon0405 youngeunkwon0405 requested review from a team as code owners January 30, 2026 23:20
@copy-pr-bot

copy-pr-bot Bot commented Jan 30, 2026

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@youngeunkwon0405 youngeunkwon0405 added the dev2main: mbridge dev to main: this PR is needed in main for mbridge label Jan 30, 2026
@youngeunkwon0405 youngeunkwon0405 self-assigned this Jan 30, 2026
@youngeunkwon0405 youngeunkwon0405 added the bug Something isn't working label Jan 30, 2026
@youngeunkwon0405

youngeunkwon0405 commented Jan 30, 2026

Copy link
Copy Markdown
Member Author

/ok to test c7039b9

@ko3n1g ko3n1g added this to the Core 0.16 milestone Jan 30, 2026
erhoo82
erhoo82 approved these changes Jan 30, 2026
View reviewed changes

@erhoo82 erhoo82 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@erhoo82 erhoo82 added complexity: low Expert Review [deprecated] Apply this label to indicate that your PR is ready for expert review. labels Jan 30, 2026
sanandaraj5597
sanandaraj5597 reviewed Jan 30, 2026
View reviewed changes
Comment thread megatron/core/distributed/param_and_grad_buffer.py Outdated
@youngeunkwon0405

youngeunkwon0405 commented Jan 30, 2026

Copy link
Copy Markdown
Member Author

/ok to test 8a62df3

sanandaraj5597
sanandaraj5597 approved these changes Jan 30, 2026
View reviewed changes

@sanandaraj5597 sanandaraj5597 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@yaox12 yaox12 disabled auto-merge February 1, 2026 08:33
@yaox12 yaox12 enabled auto-merge February 1, 2026 08:34
@yaox12

yaox12 commented Feb 1, 2026

Copy link
Copy Markdown
Member

/ok to test 13376d1

@ko3n1g

ko3n1g commented Feb 1, 2026

Copy link
Copy Markdown
Contributor

❌ Cherry-pick to main failed

The cherry-pick encountered conflicts and could not be completed automatically.

Next steps:

  1. Manually create a PR with these changes to main
  2. Resolve any conflicts

@yaox12 yaox12 added this pull request to the merge queue Feb 1, 2026
Merged via the queue into NVIDIA:dev with commit 9375be4 Feb 1, 2026
43 of 46 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@yaox12 yaox12 yaox12 approved these changes

@erhoo82 erhoo82 erhoo82 approved these changes

@sanandaraj5597 sanandaraj5597 sanandaraj5597 approved these changes

Assignees

@youngeunkwon0405 youngeunkwon0405

Labels

bug Something isn't working complexity: low dev2main: mbridge dev to main: this PR is needed in main for mbridge Expert Review [deprecated] Apply this label to indicate that your PR is ready for expert review. mirror-to-main

Projects

None yet

Milestone

Core 0.16

Development

Successfully merging this pull request may close these issues.

5 participants

@youngeunkwon0405 @yaox12 @ko3n1g @erhoo82 @sanandaraj5597