biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[NVIDIA] Support Cutlass w8a8 FP8 for Blackwell Geforce GPUs (sm120) (#17280)

Browse Source 
Signed-off-by: kaln27 <liaojuncheng123@foxmail.com>
Co-authored-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Joonchen Liau
2025-07-02 20:47:19 +08:00
committed by GitHub
parent 0c600b9ab6
commit 9e5552aa13
7 changed files with 238 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
csrc/quantization/cutlass_w8a8/c3x/scaled_mm_kernels.hpp
View File

@@ -36,6 +36,12 @@ void cutlass_scaled_mm_sm100_fp8(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& b_scales,
std::optional<torch::Tensor> const& bias);
void cutlass_scaled_mm_sm120_fp8(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
std::optional<torch::Tensor> const& bias);
void cutlass_scaled_mm_blockwise_sm100_fp8(torch::Tensor& out,
torch::Tensor const& a,
torch::Tensor const& b,