Skip to content

fix cudnn sdpa invalid seqlen for unused segments #30023

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Jul 11, 2025
Merged

fix cudnn sdpa invalid seqlen for unused segments #30023

merged 1 commit into from
Jul 11, 2025

Conversation

Cjkkkk
Copy link
Contributor

@Cjkkkk Cjkkkk commented Jul 7, 2025

cudnn sdpa on blackwell requires seqlen to be 0 for any unused segments in sequence packing. Using -1 will result in flaky cuda error sometimes.

mattjj
mattjj approved these changes Jul 9, 2025
View reviewed changes
Copy link
Collaborator

@mattjj mattjj left a comment

Choose a reason for hiding this comment

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

Thank you!

@google-ml-butler google-ml-butler bot added kokoro:force-run pull ready Ready for copybara import and testing labels Jul 9, 2025
@mattjj mattjj self-assigned this Jul 9, 2025
@copybara-service copybara-service bot merged commit 8a1dbef into jax-ml:main Jul 11, 2025
21 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@mattjj mattjj mattjj approved these changes

Assignees

@mattjj mattjj

Labels
kokoro:force-run pull ready Ready for copybara import and testing
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@Cjkkkk @mattjj