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

[Bugfix][Build/CI] Fix sparse CUTLASS compilation on CUDA [12.0, 12.2) (#11311)

Browse Source 
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
This commit is contained in:
Tyler Michael Smith
2024-12-18 21:43:30 -05:00
committed by GitHub
parent fdea8ec167
commit 5a9da2e6e9
12 changed files with 89 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
csrc/sparse/cutlass/sparse_compressor_entry.cu
View File

@@ -5,7 +5,7 @@
#include "cutlass_extensions/common.hpp"
#if defined ENABLE_SCALED_MM_C3X && ENABLE_SCALED_MM_C3X
#if defined ENABLE_SPARSE_SCALED_MM_C3X && ENABLE_SPARSE_SCALED_MM_C3X
bool cutlass_sparse_compress_sm90(torch::Tensor& a_nzs, torch::Tensor& a_meta,
torch::Tensor const& a);
#endif
@@ -28,7 +28,7 @@ bool cutlass_sparse_compress_entry(torch::Tensor& a_nzs, torch::Tensor& a_meta,
int32_t version_num = get_sm_version_num();
// Guard against compilation issues for sm90 kernels
#if defined ENABLE_SCALED_MM_C3X && ENABLE_SCALED_MM_C3X
#if defined ENABLE_SPARSE_SCALED_MM_C3X && ENABLE_SPARSE_SCALED_MM_C3X
if (version_num >= 90) {
return cutlass_sparse_compress_sm90(a_nzs, a_meta, a);
}