Skip to content

Fix hybrid fp8 test guard - #3492

Merged
negvet merged 1 commit into
NVIDIA:mainfrom
negvet:fix_hybrid_fp8_test_guard
Sep 8, 2026
Merged

Fix hybrid fp8 test guard#3492
negvet merged 1 commit into
NVIDIA:mainfrom
negvet:fix_hybrid_fp8_test_guard

Conversation

@negvet

@negvet negvet commented Sep 7, 2026

Copy link
Copy Markdown
Collaborator

Description

Please include a brief summary of the changes, relevant motivation and context.

Fixes # (issue)

Type of change

  • Documentation change (change only to the documentation, either a fix or a new content)
  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Infra/Build change
  • Code refactoring

Changes

Please list the changes introduced in this PR:

  • Change A
  • Change B

Checklist:

  • I have read and followed the contributing guidelines
  • The functionality is complete
  • I have commented my code, particularly in hard-to-understand areas
  • I have made corresponding changes to the documentation
  • My changes generate no new warnings
  • I have added tests that prove my fix is effective or that my feature works
  • New and existing unit tests pass locally with my changes

Signed-off-by: Evgeny <etsykunov@nvidia.com>
@negvet

negvet commented Sep 7, 2026

Copy link
Copy Markdown
Collaborator Author

/te-ci pytorch

@negvet
negvet requested a review from ptrendx September 7, 2026 13:44
@greptile-apps

greptile-apps Bot commented Sep 7, 2026

Copy link
Copy Markdown
Contributor

Greptile Summary

This PR prevents two hybrid FSDP2 tests from failing on systems without FP8 support by adding matching pytest skip guards.

  • Guards the FusedAdam hybrid FP8-current-scaling test.
  • Guards the all-Identity test because its enabled Transformer Engine autocast context still requires FP8 availability.

Confidence Score: 5/5

The PR appears safe to merge; both guards align with the runtime requirements of their respective tests.

No actionable defects were identified, and both tests would otherwise reach code paths that reject environments where the shared FP8 availability check fails.

Important Files Changed

Filename Overview
tests/pytorch/distributed/fsdp2_tests/run_fsdp2_fused_adam.py Adds an appropriate FP8 availability guard to a hybrid current-scaling test.
tests/pytorch/distributed/fsdp2_tests/run_fsdp2_model.py Adds an FP8 availability guard matching the unconditional requirement of the test’s enabled autocast context.

Reviews (1): Last reviewed commit: "Fix hybrid fp8 test guard" | Re-trigger Greptile

@vthumbe1503 vthumbe1503 left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

LGTM

@negvet
negvet merged commit cd24504 into NVIDIA:main Sep 8, 2026
16 of 28 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants