CUDA: fix race conditions in FlashAttention kernels #13438
Merged
Conversation
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
github-actions
bot
added
Nvidia GPU
JohannesGaessler
changed the title
CISC
approved these changes
May 10, 2025
ikawrakow
pushed a commit
to ikawrakow/ik_llama.cpp
that referenced
this pull request
May 11, 2025
ikawrakow
added a commit
to ikawrakow/ik_llama.cpp
that referenced
this pull request
May 11, 2025
Reference: ggml-org/llama.cpp#13438 Co-authored-by: Iwan Kawrakow <[email protected]>
|
Just found an issue after this commit, #13461 It does give GGML_ASSERT(cur_p->size > 0) failed. I'm not exactly sure why it happens, but if you revert to the latest commit before this commit, or on latest commit, if reverting this specific commit, it works fine. |
|
Should be fixed by #13469 . |
Nexesenex
added a commit
to Nexesenex/croco.cpp
that referenced
this pull request
May 14, 2025
)" This reverts commit 0208355.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
There are 2 race conditions in the CUDA FlashAttention code as reported by
compute-sanitizer --tool=racecheck ./build/bin/test-backend-ops. This PR adds the necessary calls to__syncthreadsto fix them.