[BACKEND][EZ] Tighten TMA multicta verifier#9941
Merged
Merged
Conversation
lezcano
changed the title
Jokeren
approved these changes
Apr 7, 2026
Contributor
Author
|
needs #9958. I'll merge it after that one is merged |
As per the PTX docs, TMAs have a very specific behaviour when executed in a 2CTA kernel: >.cta_group::1 : The mbarrier signal is also multicasted to the same offset as mbar in the shared memory of the destination CTA. .cta_group::2 : The mbarrier signal is multicasted either to all the odd numbered CTAs or the even numbered CTAs within the corresponding CTA-Pair. For each destination CTA specified in the ctaMask, the mbarrier signal is sent either to the destination CTA or its peer-CTA based on CTAs %cluster_ctarank parity of shared memory where the mbarrier object mbar resides. As such, we require these CTA layouts in TMA barriers.
Co-authored-by: Codex <noreply@openai.com>
plognjen
pushed a commit
to plognjen/triton
that referenced
this pull request
Apr 14, 2026
As per the PTX docs, TMAs have a very specific behaviour when executed in a 2CTA kernel: >.cta_group::1 : The mbarrier signal is also multicasted to the same offset as mbar in the shared memory of the destination CTA. .cta_group::2 : The mbarrier signal is multicasted either to all the odd numbered CTAs or the even numbered CTAs within the corresponding CTA-Pair. For each destination CTA specified in the ctaMask, the mbarrier signal is sent either to the destination CTA or its peer-CTA based on CTAs %cluster_ctarank parity of shared memory where the mbarrier object mbar resides. As such, we require these CTA layouts in TMA barriers.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
As per the PTX docs, TMAs have a very specific behaviour when executed
in a 2CTA kernel:
As such, we require these CTA layouts in TMA barriers.