[KERNELS] simplify mx shuffled weights defaults#9986
Merged
Conversation
plognjen
pushed a commit
to plognjen/triton
that referenced
this pull request
Apr 14, 2026
Simplify use of shuffled blackwell mx value weights - convert directly to BlackwellMX4ValueShuffledLayout; don't require first going through BlackwelllValueLayout - use block sizes from BlackwellMX4ValueShuffledLayout as opt flag constraints. removes complicated code needed to infer the block sizes before making the layout. pick a better default of block_n = 256, block_k = 128 which generally works well and is the inferred one except in cases where N < 256. also makes it simpler to just use, instead of also needing to override disable_mx4_block_swap = True when shuffled weights are used. - add more test coverage same perf from running `torchrun --nproc-per-node=1 python/triton_kernels/bench/bench_mlp.py`
plognjen
pushed a commit
to plognjen/triton
that referenced
this pull request
Apr 14, 2026
…riton-lang#9999) two commits forgotten in triton-lang#9986
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Simplify use of shuffled blackwell mx value weights
same perf from running
torchrun --nproc-per-node=1 python/triton_kernels/bench/bench_mlp.py