iHore/vLLM
8
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Bugfix][Triton] Make fp8_min/fp8_max constexpr in _quantize_pad_fp8_kernel (#51872)

Browse Source 
Signed-off-by: Max Hu <maxhu@nvidia.com>
Co-authored-by: Max Hu <maxhu@nvidia.com>
This commit is contained in:
Max Hu
2026-08-12 03:09:54 +00:00
committed by GitHub
co-authored by Max Hu
parent 3fa89225bc
commit d462dee37d
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
vllm/kernels/triton/qkv_padded_fp8_quant.py
+2 -2
View File
@@ -34,8 +34,8 @@ def _quantize_pad_fp8_kernel(
n_rows,
n_cols,
n_cols_padded,
fp8_min,
fp8_max,
fp8_min: tl.constexpr,
fp8_max: tl.constexpr,
SKIP_SCALE: tl.constexpr,
BLOCK_M: tl.constexpr,
BLOCK_N: tl.constexpr,