CUTLASS 3.2.1 (#1113)

* Updates for 3.2.1 release.

* Minor fix in gemm op profiler for raster order.

* Add scheduler mapping for raster order in the kernels.
This commit is contained in:
ANIKET SHIVAM
2023-09-26 17:24:26 -04:00
committed by GitHub
parent e0aaa3c3b3
commit 90d3b0fb18
428 changed files with 22252 additions and 21761 deletions
+3 -28
View File
@@ -37,40 +37,15 @@
#include "cutlass/coord.h"
#include "cutlass/gemm_coord.h"
#include "cutlass/layout/matrix.h"
#include "cutlass/gemm/gemm_enumerated_types.h"
#include "cute/layout.hpp"
#include "cutlass/detail/layout.hpp"
/////////////////////////////////////////////////////////////////////////////////////////////////
namespace cutlass {
namespace gemm {
/////////////////////////////////////////////////////////////////////////////////////////////////
/// GEMM operand enumeration: D = A * B + C
enum class Operand {
kA, /// A multiplicand
kB, /// B multiplicand
kC, /// Source accumulator
kD /// Destination accumulator
};
/////////////////////////////////////////////////////////////////////////////////////////////////
enum class GemmUniversalMode {
kGemm,
kGemmSplitKParallel,
kBatched,
kArray,
kInvalid
};
////////////////////////////////////////////////////////////////////////////////
/// Some options for clearing shared memory
enum class SharedMemoryClearOption {
kNone, ///< SMEM is in don't-care state
kZfill, ///< Kernels fill out of bounds accesses with zeros
kClearLastStage ///< Last SMEM stage is explicitly cleared. Mainloop uses 'kNone'
};
////////////////////////////////////////////////////////////////////////////////////////////////////