Flash Attention v3 #36190
Draft
Flash Attention v3 #36190
Conversation
This file contains 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
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
vasqu
reviewed
Feb 14, 2025
There was a problem hiding this comment.
Just a preheader to warn/inform you on some stuff regarding the current status of fa3:
- sm80 is supported (A100 etc) (and up)
- (arm64 is supported now I think, not sure if it was before)
- it doesn't seem like dropout will be supported ( Flash attention 3 does not use Dropout_p? Dao-AILab/flash-attention#1377 )
- (barebones) padding is included in hopper ( https://github.com/Dao-AILab/flash-attention/blob/main/hopper/padding.py )
seqused_(q/k) is now forced in the varlen interface ( https://github.com/Dao-AILab/flash-attention/blob/fa445ff6c215026438cca496a97242b8269aa428/hopper/flash_attn_interface.py#L566-L567 ) but tbh not sure if this was unintended ( opened an issue at [newest main shouldnt require it anymoreFA3] Forced usage ofseqused_(q/k)in varlen Dao-AILab/flash-attention#1495 )- qkv packed exisits for base fa3 forward (but not the others)
- softcapping should be supported now ( e.g. https://github.com/Dao-AILab/flash-attention/blob/fa445ff6c215026438cca496a97242b8269aa428/hopper/flash_attn_interface.py#L576 )
- fp8 backward doesnt look like it will be added soon ( Is there a plan to support flash_attn_varlen_backward with fp8 Dao-AILab/flash-attention#1420 (comment) )
| if torch.version.cuda: | ||
| compute_capability = torch.cuda.get_device_capability() | ||
| major, _ = compute_capability | ||
| if major < 9: |
There was a problem hiding this comment.
A100 support has been recently added Dao-AILab/flash-attention#1481 (comment)
|
cc @bn999 if you're interested about the progress |
|
@vasqu Yup, I'm following. Good stuff. |
|
Thanks for the info @vasqu |
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.
What does this PR do?
Replaces #33522 to avoid conflicts and allow those using it to continue while we get it updated for #35235
Initial commit of this PR adds auxiliary code so we can discuss the core FAv3 integration.
cc @ArthurZucker
_flash_attention_forward/flash_attention_forwardas before or create new functions?FlashAttention2classes, is refactoring all models to use the new style planned? Integrate FAv3 as before or do the refactor in this PR?Also to check:
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.