minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 13 Packages Projects Releases Wiki Activity

Delete CUTLASS FP8 blockwise for SM90 and SM100, move SM120 to JIT and add SwapAB (#30438)

Browse Source 
Co-authored-by: Brayden Zhong <brayden.zhong@radixark.ai>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
Co-authored-by: root <root@sgl-b300-inference.datacrunch.io>
Co-authored-by: Brayden Zhong <brayden@radixark.ai>
This commit is contained in:
Brayden Zhong
2026-07-14 09:31:32 +08:00
committed by GitHub
co-authored by Brayden Zhong Claude Sonnet 5 root Brayden Zhong
parent c124bec99d
commit 7431f35fd8
17 changed files with 765 additions and 1159 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/ci/cuda/warmup_deep_gemm.py
+1 -1
View File
@@ -115,7 +115,7 @@ def compute_deepseek_v2v3_shapes(config, tp):
Shape derivation based on:
- MoE: python/sglang/srt/layers/moe/fused_moe_triton/layer.py
- MLA: python/sglang/srt/models/deepseek_v2.py
- FP8: python/sglang/srt/layers/quantization/fp8_kernel.py
- FP8: python/sglang/kernels/ops/quantization/fp8_kernel.py
"""
shapes = []