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

CUTLASS 3.4.0 (#1286)

Browse Source 
* CUTLASS 3.4.0

* Update CHANGELOG.md

---------

Co-authored-by: Pradeep Ramani <prramani@nvidia.com>
This commit is contained in:
Pradeep Ramani
2023-12-29 15:21:31 -05:00
committed by GitHub
co-authored by Pradeep Ramani
parent b7508e3379
commit 8236f30675
211 changed files with 11409 additions and 2763 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/cutlass/gemm/kernel/gemm_universal.hpp
+14
View File
@@ -61,6 +61,19 @@ template <
>
class GemmUniversal;
////////////////////////////////////////////////////////////////////////////////
// In cases where ProblemShape is not a tuple, this is used to check if the
// underlying problem shape type is aliased within or not.
// Used for dispatching GemmUniversal to 2.x API or 3.x API
template <class ProblemShape, class = void>
struct IsCutlass3ArrayKernel : cute::false_type { };
template <typename ProblemShape>
struct IsCutlass3ArrayKernel<ProblemShape, cute::void_t<typename ProblemShape::UnderlyingProblemShape>>
: cute::true_type { };
////////////////////////////////////////////////////////////////////////////////
} // namespace cutlass::gemm::kernel
@@ -75,4 +88,5 @@ class GemmUniversal;
#include "cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized.hpp"
#include "cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized_pingpong.hpp"
#include "cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized_cooperative.hpp"
#include "cutlass/gemm/kernel/sm90_gemm_array_tma_warpspecialized_cooperative.hpp"
////////////////////////////////////////////////////////////////////////////////