Add num_splits for mha_varlen_fwd FA2, support batch invariant #110
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
Signed-off-by: quanliu <[email protected]>
|
Thks. I have noticed that vllm 0.11.2 released today, but it currently unsupported SM89 GPUs yet, right? |
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Outdated suggestions cannot be applied.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.
Add
num_splitsparameter formha_varlen_fwdFA2 to support batch-invariant processingBatch-invariant is currently unsupported on FA2 for SM80 GPUs, which can lead to inconsistent outputs for the same input.
This PR adds a num_splits parameter. Setting num_splits=1 ensures consistent outputs when batch-invariant is enabled.
vllm-project/vllm#27433 (comment)
Verified on A800 GPU with Qwen-3 32B model.
Testing done with --disable-cascade-attn disabled, under TP=1 and TP=2 settings.
Outputs are consistent.