CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers
|
Parameters structure.
#include <gemm_splitk_parallel.h>
Public Member Functions | |
CUTLASS_HOST_DEVICE | Params () |
CUTLASS_HOST_DEVICE | Params (cutlass::gemm::GemmCoord const &problem_size, cutlass::gemm::GemmCoord const &grid_tiled_shape, typename Mma::IteratorA::TensorRef ref_A, typename Mma::IteratorB::TensorRef ref_B, typename Epilogue::OutputTileIterator::TensorRef ref_D, typename OutputOp::Params output_op, int64_t splitk_slice_stride) |
Public Attributes | |
cutlass::gemm::GemmCoord | problem_size |
cutlass::gemm::GemmCoord | grid_tiled_shape |
Mma::IteratorA::Params | params_A |
Mma::IteratorA::TensorRef | ref_A |
Mma::IteratorB::Params | params_B |
Mma::IteratorB::TensorRef | ref_B |
Epilogue::OutputTileIterator::Params | params_D |
Epilogue::OutputTileIterator::TensorRef | ref_D |
OutputOp::Params | output_op |
int64_t | splitk_slice_stride |
int | gemm_k_size |
|
inline |
|
inline |
int cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::gemm_k_size |
cutlass::gemm::GemmCoord cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::grid_tiled_shape |
OutputOp::Params cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::output_op |
Mma::IteratorA::Params cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::params_A |
Mma::IteratorB::Params cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::params_B |
Epilogue::OutputTileIterator::Params cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::params_D |
cutlass::gemm::GemmCoord cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::problem_size |
Mma::IteratorA::TensorRef cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::ref_A |
Mma::IteratorB::TensorRef cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::ref_B |
Epilogue::OutputTileIterator::TensorRef cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::ref_D |
int64_t cutlass::gemm::kernel::GemmSplitKParallel< Mma_, Epilogue_, ThreadblockSwizzle_ >::Params::splitk_slice_stride |