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

v4.0 update. (#2371)

Browse Source
This commit is contained in:
Junkai-Wu
2025-06-06 14:39:20 +08:00
committed by GitHub
parent 2e2af190bd
commit 8bdbfca682
254 changed files with 29751 additions and 1980 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/cutlass/experimental/distributed/device/dist_gemm_universal_wrapper.hpp
View File

@@ -446,7 +446,7 @@ public:
Status
construct_graph(bool launch_with_pdl) {
#if ((__CUDACC_VER_MAJOR__ >= 12) && (__CUDACC_VER_MINOR__ >= 4))
#if (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
Status status = Status::kSuccess;
// Destroy existing graph, if created

2
include/cutlass/experimental/distributed/device/full_barrier.hpp
View File

@@ -47,7 +47,7 @@ void launch_full_barrier(
cudaStream_t stream,
bool launch_with_pdl) {
#if (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 4))
#if (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
// Legacy (kernel) launch with PDL
cudaLaunchAttribute attributes[1];
attributes[0].id = cudaLaunchAttributeProgrammaticStreamSerialization;