CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers
Classes | Public Types | Public Member Functions | Static Public Attributes | List of all members
cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ > Struct Template Reference

#include <gemm_splitk_parallel.h>

Classes

struct  Params
 Parameters structure. More...
 
union  SharedStorage
 Shared memory storage structure. More...
 

Public Types

using Mma = Mma_
 
using Epilogue = Epilogue_
 
using OutputOp = typename Epilogue::OutputOp
 
using ThreadblockSwizzle = ThreadblockSwizzle_
 
using WarpCount = typename Mma::WarpCount
 Warp count (concept: GemmShape) More...
 

Public Member Functions

CUTLASS_HOST_DEVICE GemmSplitKParallel ()
 
CUTLASS_DEVICE void operator() (Params const &params, SharedStorage &shared_storage)
 Executes one GEMM. More...
 

Static Public Attributes

static int const kThreadCount = 32 * WarpCount::kCount
 
static int const kAlignmentK = Mma::Operator::Shape::kK
 

Member Typedef Documentation

template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
using cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Epilogue = Epilogue_
template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
using cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Mma = Mma_
template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
using cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::OutputOp = typename Epilogue::OutputOp
template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
using cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::ThreadblockSwizzle = ThreadblockSwizzle_
template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
using cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::WarpCount = typename Mma::WarpCount

Constructor & Destructor Documentation

template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
CUTLASS_HOST_DEVICE cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::GemmSplitKParallel ( )
inline

Member Function Documentation

template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
CUTLASS_DEVICE void cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::operator() ( Params const &  params,
SharedStorage shared_storage 
)
inline

Member Data Documentation

template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
int const cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::kAlignmentK = Mma::Operator::Shape::kK
static
template<typename Mma_ , typename Epilogue_ , typename ThreadblockSwizzle_ >
int const cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::kThreadCount = 32 * WarpCount::kCount
static

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