CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers
|
Epilogue for threadblock scoped GEMMs using Tensor Ops. More...
#include <assert.h>
#include "cutlass/cutlass.h"
#include "cutlass/matrix_shape.h"
#include "cutlass/numeric_types.h"
#include "cutlass/array.h"
#include "cutlass/layout/vector.h"
#include "cutlass/layout/tensor.h"
#include "cutlass/tensor_coord.h"
#include "cutlass/aligned_buffer.h"
#include "cutlass/gemm/gemm.h"
#include "cutlass/transform/pitch_linear_thread_map.h"
Go to the source code of this file.
Classes | |
class | cutlass::epilogue::threadblock::EpilogueBase< Shape_, WarpMmaOperator_, PartitionsK, AccumulatorFragmentIterator_, WarpTileIterator_, Padding_ > |
Base class for epilogues defining warp-level. More... | |
struct | cutlass::epilogue::threadblock::EpilogueBase< Shape_, WarpMmaOperator_, PartitionsK, AccumulatorFragmentIterator_, WarpTileIterator_, Padding_ >::SharedStorage |
Shared storage allocation needed by the epilogue. More... | |
Namespaces | |
cutlass | |
cutlass::epilogue | |
cutlass::epilogue::threadblock | |
The epilogue rearranges the result of a matrix product through shared memory to match canonical tensor layouts in global memory. Epilogues support conversion and reduction operations.