[AMD] f16_gemm gluon kernel improve pipeline#10057
Merged
Merged
Conversation
Contributor
Author
|
Converting to draft while I verify that gfx1250 tests pass. |
Contributor
Author
|
Ready for review; I've verified that the kernel tests pass, and that this change alone improves performance significantly. |
guacamoleo
force-pushed
the
dtanner/f16_gemm_pipeline
branch
from
9ab90c1 to
849ac59
Compare
antiagainst
approved these changes
Apr 17, 2026
bingyizh233
pushed a commit
to bingyizh233/triton
that referenced
this pull request
Apr 20, 2026
) Improve f16 gemm gfx1250-gluon performance. Improves gemm_tdm_pipelined_single_warp_per_simd_schedule_kernel by moving tdm.load earlier; from the top of the loop (which hides 3/4th of a loop-iteration's worth of cycles) to right after the wait (which hides a full loop-iteration's worth of cycles). This only fixes the mentioned kernel; other kernels need independent benchmarking and improving.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Improve f16 gemm gfx1250-gluon performance. Improves gemm_tdm_pipelined_single_warp_per_simd_schedule_kernel by moving tdm.load earlier; from the top of the loop (which hides 3/4th of a loop-iteration's worth of cycles) to right after the wait (which hides a full loop-iteration's worth of cycles).
This only fixes the mentioned kernel; other kernels need independent benchmarking and improving.