Public Types
using	WarpCount = gemm::GemmShape< ThreadblockShape::kM/WarpShape::kM, ThreadblockShape::kN/WarpShape::kN, kPartitionsK >
	Number of warps. More...

Static Public Attributes
static int const	kWarpSize = 32

static int const	kGroupCount
	Computes number of thread-level matrix multiplies are needed to span a warp. More...

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

static int const	kIterations = MmaSimtPolicy::LaneMmaShape::kM * kGroupCount
	Number of iterations. More...

Member Typedef Documentation

template<typename ThreadblockShape_ , typename WarpShape_ , typename MmaSimtPolicy_ , int PartitionsK, typename Element_ , int ElementsPerAccess>

using cutlass::epilogue::threadblock::DefaultThreadMapSimt< ThreadblockShape_, WarpShape_, MmaSimtPolicy_, PartitionsK, Element_, ElementsPerAccess >::Detail::WarpCount = gemm::GemmShape< ThreadblockShape::kM / WarpShape::kM, ThreadblockShape::kN / WarpShape::kN, kPartitionsK >

Member Data Documentation

template<typename ThreadblockShape_ , typename WarpShape_ , typename MmaSimtPolicy_ , int PartitionsK, typename Element_ , int ElementsPerAccess>

int const cutlass::epilogue::threadblock::DefaultThreadMapSimt< ThreadblockShape_, WarpShape_, MmaSimtPolicy_, PartitionsK, Element_, ElementsPerAccess >::Detail::kGroupCount

static

Initial value:

=

WarpShape::kM / (MmaSimtPolicy::WarpShape::kRow * MmaSimtPolicy::LaneMmaShape::kM)

template<typename ThreadblockShape_ , typename WarpShape_ , typename MmaSimtPolicy_ , int PartitionsK, typename Element_ , int ElementsPerAccess>

int const cutlass::epilogue::threadblock::DefaultThreadMapSimt< ThreadblockShape_, WarpShape_, MmaSimtPolicy_, PartitionsK, Element_, ElementsPerAccess >::Detail::kIterations = MmaSimtPolicy::LaneMmaShape::kM * kGroupCount

static

template<typename ThreadblockShape_ , typename WarpShape_ , typename MmaSimtPolicy_ , int PartitionsK, typename Element_ , int ElementsPerAccess>

int const cutlass::epilogue::threadblock::DefaultThreadMapSimt< ThreadblockShape_, WarpShape_, MmaSimtPolicy_, PartitionsK, Element_, ElementsPerAccess >::Detail::kThreads = WarpCount::kCount * kWarpSize

static

template<typename ThreadblockShape_ , typename WarpShape_ , typename MmaSimtPolicy_ , int PartitionsK, typename Element_ , int ElementsPerAccess>

int const cutlass::epilogue::threadblock::DefaultThreadMapSimt< ThreadblockShape_, WarpShape_, MmaSimtPolicy_, PartitionsK, Element_, ElementsPerAccess >::Detail::kWarpSize = 32

static

The documentation for this struct was generated from the following file: