[Fyzyukk]

Summary

Add an OPUS backend for blockscale mxfp8 GEMM with B preshuffle on gfx950, and improve performance for small shapes (M <= 1024). Enable the existing gemm_a8w8_blockscale_bpreshuffle interface to select tuned OPUS kernels.

Details

• direct_b: Directly load b into vgpr and extend it to handle workloads with small M.
• stage_b: Extend M-repeat computation to handle workloads with small M.
• Support FP8 inputs, native E8M0 scales (1x128 for A and 128x128 for B), and BF16 output through the existing blockscale GEMM interface, reusing the scale bytes through their logical tensor views.

Speedup Distribution

634 gfx950/CU256 small-M shapes (M <= 1024); the tuned table selects OPUS for 425 shapes.

Speedup range	Shapes
0–5%	108
5–10%	135
10–15%	100
15–20%	55
≥20%	27

[github-actions]

🏷️ 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:gfx1250-ffm-triton	Run the five-shard gfx1250 FFM Triton test suite
ci:triton-300x	Run an additional Triton test job on MI300X in PRs (added automatically when gfx942 configs change); main branch always runs both MI35X and MI300X
ci:triton-355	Run the full Triton test suite on MI35X, not only the tests the change affects
ci:windows	Windows JIT op smoke test on the gfx1201 ROCm runner
ci:multi-gpu	Aiter multi-GPU tests on the 8-GPU runner
ci:sglang	SGLang integration tests: DeepSeek-R1-MXFP4 accuracy, Qwen 3.5 accuracy
ci:atom	ATOM benchmark: DeepSeek-R1-0528, GPT-OSS-120B
ci:atom_full	ATOM accuracy suite for PR and main models from ATOM models_accuracy.json
ci:vllm	vLLM benchmark: DeepSeek-V4-Pro, MiniMax-M3-MXFP4
ci:all	All standard extended tests (excludes ci:atom_full)

Only add ci:atom_full for FlyDSL or Triton upgrades.
Add labels via the sidebar or gh pr edit 7387 --add-label <label>

One backend per PR:
A PR changes one kernel backend: [Triton/Gluon] (Triton and Gluon count as one), [HIP], [ASM], [CK], [OPUS] or [FlyDSL]. If the title ends up with two backend tags, split the PR -- as stacked pull requests when one part cannot merge without the other.

PR title tags & labels:
Component tags ([Triton/Gluon], [HIP], [CK], [ASM], ...) are added to the PR title and as PR labels automatically from the changed files and re-synced on every push — change-type tags like [fix]/[Perf], op tags like [MLA], and human labels (ci:*) are left untouched. Add the no-auto-title label to stop the title rewrites; labels stay in sync either way.

[aiter-bot]

WAITING ON AUTHOR

PR Self-Review

Revision d4bed4246d99 · mode shadow

Waiting for the PR author. The readiness report will appear after the self-review answers are complete for this revision.

Questions to answer in the PR description

1. What behavior does this PR change, and why is the existing behavior insufficient?
2. What did you validate, and what remains unverified? Include commands, hardware/configuration and results.
3. Which architecture, dtype and shape boundaries select the changed kernel path, and what invariant prevents unsupported inputs from reaching it?
4. For every changed op_tests/ file, what behavior does it cover and why is it not redundant? Write None if no op test changed.

Edit the PR Self-Review section in the PR description. Include the current revision SHA; this comment refreshes automatically.

Missing or invalid answers

• FAIL — missing ### Reviewed revision in the PR description
• FAIL — missing ### Behavior and intent in the PR description
• FAIL — missing ### Validation and unverified scope in the PR description
• FAIL — missing ### PR-specific question in the PR description
• FAIL — missing ### Changed op tests in the PR description
• FAIL — Reviewed revision must contain at least the first 7 characters of the current head SHA
• FAIL — Behavior and intent has no substantive answer
• FAIL — Validation and unverified scope has no substantive answer
• FAIL — PR-specific question has no substantive answer
• FAIL — changed op test not explained in Changed op tests: op_tests/test_opus_a8w8_interface.py
• FAIL — changed op test not explained in Changed op tests: op_tests/test_opus_co_integration.py
• FAIL — Changed op tests must explain why each changed op test is necessary and non-redundant

_{Shadow mode reports what would block but keeps the status successful. Set AITER_PR_SELF_REVIEW_MODE=enforce only after reviewing real false positives.}