Fix: disable 2-CTA backward mode when block_sparse_tensors is used#2433
Merged
drisspg merged 1 commit intoDao-AILab:mainfrom Apr 3, 2026
Merged
Conversation
The SM100 2-CTA backward kernel does not properly handle block_sparse_tensors. When block sparsity is combined with 2-CTA mode, the kernel hits an assertion: 'AssertionError: 2-CTA mode does not support block sparsity' This fix adds block_sparse_tensors to the disable_2cta condition in the backward path, forcing the 1-CTA kernel when block sparsity is active. The 1-CTA backward kernel already supports block_sparse_tensors correctly. Without this fix, any backward pass using block_sparse_tensors on SM100 (B200/GB200) with head_dim >= 128 will crash with the above assertion.
jduprat
pushed a commit
to jduprat/flash-attention
that referenced
this pull request
Apr 3, 2026
* Initial FA-2 aiter Triton Windows build support * minimize diff * bump commit * bump commit * minimize diff * bump commit * bump aiter submodule * bump aiter submodule to merged Dao-AILab#2433 * fix: guard distributed.py fallbacks with hasattr for Windows --------- Co-authored-by: 0xDELUXA <djernovevo@gmail.com>
drisspg
approved these changes
Apr 3, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The SM100 2-CTA backward kernel does not properly handle block_sparse_tensors. When block sparsity is combined with 2-CTA mode, the kernel hits an assertion:
'AssertionError: 2-CTA mode does not support block sparsity'
This fix adds block_sparse_tensors to the disable_2cta condition in the backward path, forcing the 1-CTA kernel when block sparsity is active. The 1-CTA backward kernel already supports block_sparse_tensors correctly.
Without this fix, any backward pass using block_sparse_tensors on SM100 (B200/GB200) with head_dim >= 128 will crash with the above assertion.