/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/develop/include/ck_tile/ops/gemm/kernel/gemm_tile_partitioner.hpp File Reference

• develop
• include
• ck_tile
• ops
• gemm
• kernel

Classes | Namespaces

gemm_tile_partitioner.hpp File Reference

#include "ck_tile/core.hpp"
#include "ck_tile/ops/common.hpp"

Go to the source code of this file.

Classes
struct	ck_tile::GemmTile2DPartitioner< BlockGemmShapeType >
	Class providing 2D workgroup index mapping into 2D output GEMM C-tile space. More...
struct	ck_tile::GemmTile1DPartitioner< BlockGemmShape_ >
	Class providing 1D WGP index mapping into 2D output C-tile space. More...
struct	ck_tile::HasFnOneArgImpl< typename, typename >
	GemmTile1DPartitioner::GetOutputTileIndex's std::false specialization, checking expression validity in-place for ill-formed. More...
struct	ck_tile::HasFnOneArgImpl< T, std::void_t< decltype(std::declval< T >().GetOutputTileIndex(1))> >
	GemmTile1DPartitioner::GetOutputTileIndex's std::true specialization, checking expression validity in-place for well-formed. More...
struct	ck_tile::OffsettedTile1DPartitioner< TilePartitioner, typename >
	Struct used to calculate offseted tile indexes. More...
struct	ck_tile::GemmSpatiallyLocalTilePartitioner< BlockGemmShapeType, GroupNum, M01 >
	Class mapping 1D block index into 2D output tile space. More...
struct	ck_tile::StreamKTilePartitioner< BlockGemmShapeType, ReductionStrategy, TileSwizzleSubM >
	Stream-K tile partitioner that dynamically balances work across workgroups. More...

Namespaces
	ck_tile

Detailed Description

GemmTilePartitioner allows customized mapping between a workgroup and the C-tile it computes.