Skip to content

[hotfix] always use fp4x2 for swiglu separated per_1x32 path#3591

Open
yadaish wants to merge 1 commit into
mainfrom
dev/hot_fix_swiglu_separate
Open

[hotfix] always use fp4x2 for swiglu separated per_1x32 path#3591
yadaish wants to merge 1 commit into
mainfrom
dev/hot_fix_swiglu_separate

Conversation

@yadaish
Copy link
Copy Markdown
Contributor

@yadaish yadaish commented Jun 8, 2026

No description provided.

Drop the M < _SWIGLU_MXFP4_BF16_BOUND threshold that fell back to bf16
for small M in the Swiglu + SEPARATED gate path; use fp4x2 unconditionally.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@yadaish yadaish requested a review from a team June 8, 2026 03:19
@github-actions
Copy link
Copy Markdown
Contributor

github-actions Bot commented Jun 8, 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: 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: GPT-OSS-120B, DeepSeek-R1-0528, Kimi-K2.5
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 3591 --add-label <label>

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant