Skip to content

Add MiniMax M25 A8W8 blockscale GEMM tunings#2979

Closed
akii96 wants to merge 1 commit intomainfrom
gemm-tuning-minimax-m25-gfx950
Closed

Add MiniMax M25 A8W8 blockscale GEMM tunings#2979
akii96 wants to merge 1 commit intomainfrom
gemm-tuning-minimax-m25-gfx950

Conversation

@akii96
Copy link
Copy Markdown
Contributor

@akii96 akii96 commented Apr 30, 2026 •
edited
Loading

Adds MiniMax M25 A8W8 blockscale GEMM tuning entries and keeps the tuning table deduplicated and sorted

Pre-requirments for this to be merged:

@akii96 akii96 requested a review from a team April 30, 2026 13:32
@github-actions
Copy link
Copy Markdown
Contributor

github-actions Bot commented Apr 30, 2026

🏷️ CI Guide

Runs automatically on every PR:

  • ✅ Pre-checks (submodule verification, code formatting)
  • ✅ Aiter op tests (gfx942 + gfx950)
  • ✅ Triton tests on MI35X (only when aiter/ops/triton/** or related paths are changed)

Extended tests (opt-in via labels):

Label Tests
ci:triton-300x Run an additional Triton test job on MI300X in PRs; main branch always runs both MI35X and MI300X
ci:sglang SGLang integration tests
ci:atom ATOM benchmark (DeepSeek-R1 + GPT-OSS)
ci:vllm vLLM benchmark
ci:all All of the above

Add labels via the sidebar or gh pr edit 2979 --add-label <label>

@akii96 akii96 marked this pull request as draft April 30, 2026 13:40
@akii96
Copy link
Copy Markdown
Contributor Author

akii96 commented Apr 30, 2026 •
edited
Loading

I checked now that the upstream aiter main has now already changed a bit and now takes an additional gfx_arch column as input.

This should be a small fix and I can address once I rebase and test it myself on workloads

@sunway513 sunway513 mentioned this pull request May 1, 2026
Merged
2 tasks
@akii96 akii96 force-pushed the gemm-tuning-minimax-m25-gfx950 branch from 95cf9fd to aa8d197 Compare May 4, 2026 06:39
@akii96
Copy link
Copy Markdown
Contributor Author

akii96 commented May 4, 2026

Latest updates

  • Rebased onto current origin/main

  • Re-tuned the full M25 shape set on MI355X (gfx950, cu_num=256, 8 GPUs)

Dependency on PR #2541

@akii96 akii96 marked this pull request as ready for review May 4, 2026 06:58
@akii96 akii96 requested a review from amd-yashagar May 4, 2026 06:59
@amd-yashagar
Copy link
Copy Markdown
Contributor

amd-yashagar commented May 4, 2026

Looks good to me. Thank you @akii96.

@akii96 akii96 force-pushed the gemm-tuning-minimax-m25-gfx950 branch from aa8d197 to 4f555c5 Compare May 4, 2026 11:33
@sunway513
Copy link
Copy Markdown
Collaborator

sunway513 commented May 4, 2026

This PR's content was bulk-merged via #3004 ([Silo] Bulk merge: tuned GEMM and FMoE configs, merged 2026-05-02 03:16 UTC). Please close this PR as superseded.

Tracking issue: ROCm/AI-Frameworks-Dashboard#141

sunway513 added a commit that referenced this pull request May 4, 2026
@sunway513
Cherry-pick: [Silo] Bulk merge tuned configs for v0.1.13.post1 (#3004)
e4b2d6a 
Squash-merged from main commit 52c4554.

Includes 5 atomic Silo PRs:
- #2923 GLM-4.7 FP8 tuned/untuned FMoE configs (new)
- #2938 Kimi-K2.5 FP4 fused MoE tunings (TP2 / 256 CU refresh)
- #2979 MiniMax-M2.5 A8W8 blockscale GEMM tunings
- #2981 DeepSeek-V3.2 MI355X tuned GEMM and FMoE configs
- #2982 MiniMax-M2.5 FMoE tunings

Conflict in aiter/configs/model_configs/kimik2_fp4_tuned_fmoe.csv:
two blocks resolved by taking theirs (Silo). Block 1 upgrades existing
M=256/N=512 rows from base kernel suffixes (w3) to tuner-discovered
variants (w3_xcd4, _bnt2_persist, _sbm32, _sbm64). Block 2 is purely
additive: 30+ new rows for previously-uncovered N=7168/K=1024 shapes
plus a flydsl_fallback section.

Driver: vLLM 0.21 freeze 2026-05-08 — Silo customers need these tunings
on the AITER release wheel, not nightly.

Verification gate before tag:
- Kernel suffix parser smoke (Kimi-K2.5-MXFP4 1-token inference,
  confirm new suffixes JIT-compile without falling back)
- ATOM 5-model accuracy unchanged within +/- 0.005 vs v0.1.13-rc1
- Perf delta on Kimi-K2.5 / MiniMax-M2.5 / DSv3.2 (expect flat or better)

(cherry picked from commit 52c4554)
@azaidy azaidy mentioned this pull request May 4, 2026
Merged
2 tasks
@akii96 akii96 marked this pull request as draft May 5, 2026 11:41
@akii96
Copy link
Copy Markdown
Contributor Author

akii96 commented May 5, 2026

Merged with #3024

@akii96 akii96 closed this May 5, 2026
@akii96 akii96 deleted the gemm-tuning-minimax-m25-gfx950 branch May 5, 2026 19:18
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@amd-yashagar amd-yashagar Awaiting requested review from amd-yashagar

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@akii96 @amd-yashagar @sunway513