Skip to content

FA 2.8.3 support#443

Open
ipanfilo wants to merge 2 commits intodevfrom
ipanfilo/fa_283
Open

FA 2.8.3 support#443
ipanfilo wants to merge 2 commits intodevfrom
ipanfilo/fa_283

Conversation

@ipanfilo
Copy link
Collaborator

@ipanfilo ipanfilo commented Feb 5, 2026

Description

Update max supported FA version to 2.8.3

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:

  • Bump FA version (cherry-pick from upstream)
  • Update tests limits for FA

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

ipanfilo and others added 2 commits February 4, 2026 23:18
Signed-off-by: 将来 <jianglai.zhw@alibaba-inc.com>
Co-authored-by: 将来 <jianglai.zhw@alibaba-inc.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
(cherry picked from commit a7a69ca)
torch.half: (5e-3, 5e-3),
torch.bfloat16: (3.5e-2, 3.5e-2),
}
# With FA on ROCm it may not fit default tolerance
Copy link
Collaborator

Choose a reason for hiding this comment

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

Is this relax of tolerance due to upgrading to 2.8.3?

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

Yes, even though I did see some numerical errors on ROCm 7.2 image with FA 2.8.0 too, 2.8.3 showed higher delta. It should be dues to new CK I suppose, didn't try Triton backend

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