[CUDA] fp and int4 quants for qmm_sm80#3268
Merged
zcbenz merged 1 commit intoml-explore:mainfrom Mar 19, 2026
Merged
Conversation
angeloskath
approved these changes
Mar 18, 2026
Member
angeloskath
left a comment
angeloskath
left a comment
There was a problem hiding this comment.
Looks great!
How 's the initial perf looking?
Collaborator
Author
|
The memory bandwidth of int4 is about 50% of int8 on A100:
I think it means that the kernel is bottlenecked at loading quantized weights. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Followup to #3255 finishing the missing quants, was blocked by NVIDIA/cutlass#3111.
We can't support int2/3/5/6 in this version unfortunately since the lengths supported by
cp.asyncinstruction are limited.