CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers
Public Types | Public Member Functions | List of all members
cutlass::arch::Mma< gemm::GemmShape< 8, 8, 32 >, 32, int4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd > Struct Template Reference

Matrix multiply-add operation: S32 = S4 * U4 + S32.

#include <mma_sm75.h>

Public Types

using Shape = gemm::GemmShape< 8, 8, 32 >
 
using ElementA = int4b_t
 
using LayoutA = layout::RowMajor
 
using FragmentA = Array< int4b_t, 8 >
 
using ElementB = uint4b_t
 
using LayoutB = layout::ColumnMajor
 
using FragmentB = Array< uint4b_t, 8 >
 
using ElementC = int
 
using LayoutC = layout::RowMajor
 
using FragmentC = Array< int, 2 >
 
using Operator = OpMultiplyAdd
 

Public Member Functions

CUTLASS_HOST_DEVICE void operator() (FragmentC &d, FragmentA const &a, FragmentB const &b, FragmentC const &c) const
 Computes multiply-add. More...
 

Member Typedef Documentation

using cutlass::arch::Mma< gemm::GemmShape< 8, 8, 32 >, 32, int4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd >::FragmentC = Array<int, 2>
using cutlass::arch::Mma< gemm::GemmShape< 8, 8, 32 >, 32, int4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd >::Operator = OpMultiplyAdd

Member Function Documentation

CUTLASS_HOST_DEVICE void cutlass::arch::Mma< gemm::GemmShape< 8, 8, 32 >, 32, int4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd >::operator() ( FragmentC d,
FragmentA const &  a,
FragmentB const &  b,
FragmentC const &  c 
) const
inline

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