leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 24 Packages Projects Releases Wiki Activity

v3.8.0 update (#2082)

Browse Source 
* 3.8 update

* fix Markus' name

---------

Co-authored-by: yuzhai <yuzhai@nvidia.com>
This commit is contained in:
Yujia Zhai
2025-02-06 21:33:40 -05:00
committed by GitHub
co-authored by yuzhai
parent affd1b693d
commit 833f6990e0
168 changed files with 24945 additions and 3436 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/58_ada_fp8_gemm/ada_fp8_gemm.cu
+15 -5
View File
@@ -768,16 +768,26 @@ int main(int argc, char const** argv) {
return -1;
}
if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 4) ||
(props.major != 8 && props.minor != 9)) {
bool satisfied;
if (props.major < 10) {
// Pre-Blackwell
satisfied = (__CUDACC_VER_MAJOR__ > 12) || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 4);
satisfied &= (props.major > 8) || (props.major == 8 && props.minor == 9);
}
else {
satisfied = (__CUDACC_VER_MAJOR__ > 12) || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8);
}
if (!satisfied) {
//
// This example requires an NVIDIA Ada-architecture GPU.
// This example requires an NVIDIA GPU with compute capability 8.9 or greater.
//
std::cout
<< "CUTLASS's FP8 SM89 example requires a GPU of NVIDIA's Ada architecture "
<< "and CUDA toolkit version 12.4 or later.\n";
<< "CUTLASS's FP8 SM89 example requires an NVIDIA GPU with compute capability 8.9 or greater "
<< "and CUDA toolkit version 12.4 or later"
<< " (12.8 or later needed for SM100+)"
<< std::endl;
return 0;
}