#include <default_mma_core_sm75.h>

Public Types
using	Shape = Shape_

using	WarpShape = WarpShape_

using	InstructionShape = InstructionShape_

using	ElementA = ElementA_

using	LayoutA = layout::RowMajor

using	ElementB = ElementB_

using	LayoutB = layout::ColumnMajor

using	ElementC = ElementC_

using	LayoutC = LayoutC_

using	OperatorClass = arch::OpClassTensorOp

using	WarpCount = GemmShape< Shape::kM/WarpShape::kM, Shape::kN/WarpShape::kN, Shape::kK/WarpShape::kK >
	Number of warps present. More...

using	Operator = Operator_
	Default Operator. More...

using	SmemLayoutA = layout::RowMajorTensorOpMultiplicandCrosswise< sizeof_bits< ElementA >::value, Shape::kK >

using	SmemLayoutB = layout::ColumnMajorTensorOpMultiplicandCrosswise< sizeof_bits< ElementB >::value, Shape::kK >

using	IteratorThreadMapA = transform::PitchLinearWarpRakedThreadMap< layout::PitchLinearShape< Shape::kK, Shape::kM >, kThreads, layout::PitchLinearShape< kWarpThreadArrangementContiguousA, kWarpThreadArrangementStridedA >, kAccessSizeInBits/sizeof_bits< ElementA >::value >
	ThreadMap of iterator A. More...

using	SmemIteratorA = transform::threadblock::RegularTileIterator< MatrixShape< Shape::kM, Shape::kK >, ElementA, SmemLayoutA, 0, IteratorThreadMapA >
	Shared memory iterator to A operand. More...

using	IteratorThreadMapB = transform::PitchLinearWarpRakedThreadMap< layout::PitchLinearShape< Shape::kK, Shape::kN >, kThreads, layout::PitchLinearShape< kWarpThreadArrangementContiguousB, kWarpThreadArrangementStridedB >, kAccessSizeInBits/sizeof_bits< ElementB >::value >
	ThreadMap of iterator B. More...

using	SmemIteratorB = transform::threadblock::RegularTileIterator< MatrixShape< Shape::kK, Shape::kN >, ElementB, SmemLayoutB, 1, IteratorThreadMapB >
	Shared memory iterator to B operand. More...

using	MmaTensorOp = typename cutlass::gemm::warp::DefaultMmaTensorOp< WarpShape, InstructionShape, ElementA, SmemLayoutA, ElementB, SmemLayoutB, ElementC, LayoutC, Operator, WarpCount::kK >::Type

using	MmaPolicy = MmaPolicy< MmaTensorOp, MatrixShape< 0, 0 >, MatrixShape< 0, 0 >, WarpCount::kK >
	Policy used to define MmaPipelined. More...

Static Public Attributes
static int const	kWarpSize = warp::WarpSize<arch::OpClassTensorOp>::value
	Number of threads per warp. More...

static int const	kThreads = WarpCount::kCount * kWarpSize
	Number of threads total. More...

static int const	kAccessSizeInBits = 128
	Size of a threadblock-scoped access. More...

static int const	kWarpThreadArrangementContiguousA

static int const	kWarpThreadArrangementStridedA

static int const	kWarpThreadArrangementContiguousB

static int const	kWarpThreadArrangementStridedB

Detailed Description

template<typename Shape_, typename WarpShape_, typename InstructionShape_, typename ElementA_, typename ElementB_, typename ElementC_, typename LayoutC_, typename Operator_>
struct cutlass::gemm::threadblock::DefaultMmaCore< Shape_, WarpShape_, InstructionShape_, ElementA_, layout::RowMajor, ElementB_, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ >

Partial specialization:

A: row-major B: column-major Operator: tensor op class

This uses the default warp-level operator given tile sizes