[zufayu]

Motivation

Reverts #6376 (7cb93dda2281ae361d033b3d70acc1f0a20135f2) because it introduced a gfx950 vLLM regression for amd/MiniMax-M3-MXFP4.

The same TP4 job succeeded before #6376 and fails afterward in CK-Tile MoE stage 2 with:

moe_dispatch: unsupported (activation=2, has_bias=0, split_k=1) combination

The generated dispatcher set does not contain the required non-split/no-bias SwiGLU specialization. Later unrelated commits continue to reproduce the same failure, so revert first and re-land Relu2 support after the missing dispatch coverage is fixed and validated.

• Last successful run: https://github.com/ROCm/aiter/actions/runs/38246983845/job/114831705123
• First completed failing run: https://github.com/ROCm/aiter/actions/runs/38258209597/job/114858476100
• Later reproduction: https://github.com/ROCm/aiter/actions/runs/38334870056

Validation

• Reverted paths exactly match the tree immediately before 7cb93dd.
• git diff --check
• Python syntax check for the reverted Python files

Full vLLM MiniMax TP4/TP8 validation is expected from CI.

[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 6463 --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.

[zufayu]

Closing this PR as a duplicate of #6460, which reverts the same commit and is already running ci:vllm.