[gfx1250][FlyDSL] Optimize a8w8 ptpc gemm kernel#3969
Open
aoli26 wants to merge 2 commits into
Open
Conversation
Contributor
🏷️ CI GuideRuns automatically on every PR:
Extended tests (opt-in via labels):
|
aoli26
force-pushed
the
gfx1250/gemm_flydsl_perf
branch
from
158951e to
552f30c
Compare
Contributor
There was a problem hiding this comment.
Pull request overview
This PR updates the gfx1250 FlyDSL A8W8/PTPC GEMM implementation to a new optimized kernel path and refreshes the tuned-shape configuration table used for kernel selection/performance.
Changes:
- Remove the vendored OOB-capable TDM descriptor builder and associated FlyDSL feature-detection shim.
- Rework
gemm_fp8fp4_gfx1250to a wave-specialized TDM load path with updated scale-loading strategies (ascale_load_path) and scheduling tweaks. - Refresh
a8w8_bpreshuffle_tuned_gemm.csventries for gfx1250 tuned shapes/configs.
Reviewed changes
Copilot reviewed 2 out of 3 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| aiter/ops/flydsl/kernels/tdm_oob.py | Deleted the vendored OOB-capable TDM descriptor builder fallback. |
| aiter/ops/flydsl/kernels/gemm_fp8fp4_gfx1250.py | Major kernel path update: new scale load options, updated scheduling, direct TDM descriptor usage with early_timeout/oob_outer_bound, and other tuning-related refactors. |
| aiter/configs/a8w8_bpreshuffle_tuned_gemm.csv | Updated gfx1250 tuned entries (latency/throughput and chosen kernel configs) to match the new optimized kernel behavior. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
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
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.
Motivation
Update gfx1250 FlyDSL a8w8 bpreshuffle optimized kernel and tuned configs.
Technical Details
Update new optimized FlyDSL gfx1250 gemm a8w8 ptpc kernel and refresh
a8w8_bpreshuffle_tuned_gemm.csv.Test Plan
Test Result
All 48 flydsl-tuned shapes pass. Key gains (no regressions): (N=7168, K=16384) M=16: ~30us → ~10us (~3×); (N=2112, K=7168) M=16: ~7.5us → ~4.8us (~36%); remaining shapes improved or unchanged.
Submission Checklist