CUTLASS 3.2 (#1024)

* CUTLASS 3.2
This commit is contained in:
ANIKET SHIVAM
2023-08-07 20:50:32 -04:00
committed by GitHub
parent a0d787b746
commit 4575443d44
392 changed files with 47559 additions and 7940 deletions
-4
View File
@@ -64,7 +64,6 @@ __global__
void Kernel(typename Operator::Params params) {
// Dynamic shared memory base pointer
extern __shared__ int SharedStorageBase[];
// Declare pointer to dynamic shared memory.
typename Operator::SharedStorage *shared_storage =
reinterpret_cast<typename Operator::SharedStorage *>(SharedStorageBase);
@@ -81,13 +80,11 @@ __global__
void Kernel2(typename Operator::Params params) {
// Dynamic shared memory base pointer
extern __shared__ int SharedStorageBase[];
// Declare pointer to dynamic shared memory.
typename Operator::SharedStorage *shared_storage =
reinterpret_cast<typename Operator::SharedStorage *>(SharedStorageBase);
Operator::invoke(params, *shared_storage);
}
@@ -108,7 +105,6 @@ void device_kernel(CUTLASS_GRID_CONSTANT typename Operator::Params const params)
{
// Dynamic shared memory base pointer
extern __shared__ char smem[];
Operator op;
op(params, smem);
}