Cutlass
CUDA Templates for Linear Algebra Subroutines and Solvers
|
Implements tile iterators to partition the thread block tile into 2D subtiles and efficiently load each. Applies permute transformation to construct 'interleaved K-strided' data layout in which 4-element dot products from the same K index are arranged in consecutive locations within shared memory. More...
#include "cutlass/coord.h"
#include "cutlass/gemm/gemm_global_tile.h"
#include "cutlass/matrix_traits.h"
Go to the source code of this file.
Classes | |
struct | cutlass::gemm::IgemmGlobalTileTraits< kOperand_, kLayout_, Scalar_, Tile_, Threads_, kAccessSize_ > |
struct | cutlass::gemm::IgemmGlobalTileTraits< kOperand_, kLayout_, Scalar_, Tile_, Threads_, kAccessSize_ >::ThreadOffset |
Computes the thread offset in (H, W) based on thread ID. More... | |
struct | cutlass::gemm::IgemmGlobalIteratorAb< TileTraits_, Index_ > |
Namespaces | |
cutlass | |
cutlass::gemm | |
Supports efficient loads from shared memory to target the DP4A instruction.