[CuTe,Fwd,Sm100] refactor mla sm100 forward and add page table#2558
Open
jayhshah wants to merge 2 commits into
Open
[CuTe,Fwd,Sm100] refactor mla sm100 forward and add page table#2558jayhshah wants to merge 2 commits into
jayhshah wants to merge 2 commits into
Conversation
jayhshah
commented
May 13, 2026
| ), | ||
| mCuBlockIdxOffsets=( | ||
| blocksparse_tensors.cu_block_idx_offsets if blocksparse_tensors is not None else None | ||
| blocksparse_tensors.cu_block_idx_offsets |
Collaborator
Author
There was a problem hiding this comment.
This change is just to fix the linter error.
jayhshah
force-pushed
the
jshah/mla-paged-kv-refactor
branch
from
e4e954f to
0705919
Compare
Collaborator
Author
|
Note that 64k test in benchmark script throws IMA on cutlass 4.5.0 in the refactor, but this is due to issue identified here: NVIDIA/cutlass#3208 |
jayhshah
force-pushed
the
jshah/mla-paged-kv-refactor
branch
from
0705919 to
9eaec0a
Compare
Collaborator
Author
|
Benchmark shows ~15% dropoff for paged using cp async loads. |
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Superseding #2468.
We refactor the kernel file to remove duplication around splitting V and add page table support.
We also support not providing Q and K (which semantically mean q_pe and k_pe). This computes attention according to the formula
as in Deepseek v4 core attention.