Faster two pass sdpa - #3023
Merged
Merged
Conversation
awni
force-pushed
the
faster_two_pass_sdpa
branch
2 times, most recently
from
January 21, 2026 06:13
bae68d6 to
d9c9eab
Compare
awni
force-pushed
the
faster_two_pass_sdpa
branch
from
January 22, 2026 19:21
d9c9eab to
6c6de3c
Compare
Member
Author
|
Here's the pre-post on an M3 Ultra with KV heads = 8, head dim = 128, varying sequence length and gqa factor: Other cases I tested have similar results and I think the tuning is mostly generalizable (as I tuned the block sizes for head dim 128 per machine). |
awni
marked this pull request as ready for review
January 22, 2026 20:42
Member
Author
|
Closes ml-explore/mlx-lm#763 |
awni
force-pushed
the
faster_two_pass_sdpa
branch
from
January 22, 2026 21:16
5cb12d8 to
5fdc7bb
Compare
1 task
CC-Yeh
added a commit
to CC-Yeh/mlx
that referenced
this pull request
Jan 29, 2026
CC-Yeh
added a commit
to CC-Yeh/mlx
that referenced
this pull request
Jan 29, 2026
CC-Yeh
added a commit
to CC-Yeh/mlx
that referenced
this pull request
Feb 7, 2026
CC-Yeh
added a commit
to CC-Yeh/mlx
that referenced
this pull request
Apr 16, 2026
jzdziarski
pushed a commit
to jzdziarski/mlx
that referenced
this pull request
Jul 26, 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR does two things:
I tuned the block sizes on m3u, m4m and m5