chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[sgl-kernel] misc: update deepgemm version for sgl-kernel (#9340)

Browse Source 
Co-authored-by: Yineng Zhang <me@zhyncs.com>
Co-authored-by: fzyzcjy <ch271828n@outlook.com>
This commit is contained in:
PGFLMG
2025-08-28 03:01:30 +08:00
committed by GitHub
parent 07ee0ab750
commit aa3eba8eb4
25 changed files with 210 additions and 383 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/sglang/srt/layers/quantization/fp8_utils.py
View File

@@ -459,7 +459,7 @@ def _requant_weight_ue8m0(
import deep_gemm.utils.layout
sf = sf.index_select(-2, torch.arange(mn, device=sf.device) // 128)
sf = deep_gemm.utils.layout.get_col_major_tma_aligned_packed_tensor(sf)
sf = deep_gemm.utils.layout.get_mn_major_tma_aligned_packed_ue8m0_tensor(sf)
return sf
out_s = _transform_scale(out_s, mn=out_w.shape[-2])