CUTLASS: cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment > Class Template Reference

CUTLASS

CUDA Templates for Linear Algebra Subroutines and Solvers

Regular tile iterator specialized for interleaved layout + 2d thread-tiled threadmapping.

#include <regular_tile_iterator_pitch_linear_2dthreadtile.h>

Public Types
using	Shape = Shape_

using	Element = Element_

using	Layout = layout::ColumnMajorInterleaved< 4 >

using	ThreadMap = ThreadMap_

using	Index = typename Layout::Index

using	LongIndex = typename Layout::LongIndex

using	TensorRef = TensorRef< Element, Layout >

using	TensorCoord = typename Layout::TensorCoord

using	Fragment = Array< Element, ThreadMap::Iterations::kCount *ThreadMap::ThreadAccessShape::kCount >

using	PitchLinearThreadMap = PitchLinearStripminedThreadMap< layout::PitchLinearShape< Shape::kRow, Shape::kColumn >, ThreadMap::kThreads, ThreadMap::ThreadAccessShape::kCount >

using	Underlying = RegularTileIterator2dThreadTile< layout::PitchLinearShape< Shape::kRow, Shape::kColumn >, Element, layout::PitchLinear,(kAdvanceRank==0?0:1), ThreadMap >

Public Member Functions
CUTLASS_DEVICE	RegularTileIterator2dThreadTile ()

CUTLASS_DEVICE	RegularTileIterator2dThreadTile (TensorRef const &ref, int thread_idx)

CUTLASS_HOST_DEVICE void	load_with_pointer_offset (Fragment &frag, Index pointer_offset)
	Loads a fragment. More...

CUTLASS_HOST_DEVICE void	load (Fragment &frag, TensorCoord const &tile_offset)
	Loads a fragment. More...

CUTLASS_HOST_DEVICE void	load (Fragment &frag)
	Loads a fragment. More...

CUTLASS_HOST_DEVICE void	store_with_pointer_offset (Fragment const &frag, Index pointer_offset)
	Stores a fragment. More...

CUTLASS_HOST_DEVICE void	store (Fragment const &frag, TensorCoord const &tile_offset)
	Stores a fragment. More...

CUTLASS_HOST_DEVICE void	store (Fragment const &frag)
	Stores a fragment. More...

CUTLASS_HOST_DEVICE RegularTileIterator2dThreadTile &	operator++ ()
	Advances the pointer. More...

CUTLASS_HOST_DEVICE RegularTileIterator2dThreadTile &	operator-- ()
	Advances the pointer. More...

CUTLASS_HOST_DEVICE void	add_pointer_offset (LongIndex pointer_offset)
	Adds a pointer offset in units of Element. More...

CUTLASS_DEVICE void	add_tile_offset (TensorCoord const &coord)
	Adds a tile offset. More...

Static Public Attributes
static int const	kAdvanceRank = AdvanceRank

static int const	kAlignment = Alignment

Member Typedef Documentation

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::Element = Element_

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::Fragment = Array<Element, ThreadMap::Iterations::kCount * ThreadMap::ThreadAccessShape::kCount>

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::Index = typename Layout::Index

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::Layout = layout::ColumnMajorInterleaved<4>

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::LongIndex = typename Layout::LongIndex

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::PitchLinearThreadMap = PitchLinearStripminedThreadMap< layout::PitchLinearShape<Shape::kRow, Shape::kColumn>, ThreadMap::kThreads, ThreadMap::ThreadAccessShape::kCount >

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::Shape = Shape_

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::TensorCoord = typename Layout::TensorCoord

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::TensorRef = TensorRef<Element, Layout>

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::ThreadMap = ThreadMap_

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

using cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::Underlying = RegularTileIterator2dThreadTile< layout::PitchLinearShape<Shape::kRow, Shape::kColumn>, Element, layout::PitchLinear, (kAdvanceRank == 0 ? 0 : 1), ThreadMap >

Constructor & Destructor Documentation

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_DEVICE cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::RegularTileIterator2dThreadTile ( )

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_DEVICE cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::RegularTileIterator2dThreadTile	(	TensorRef const &	ref,
		int	thread_idx
	)

inline

Member Function Documentation

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::add_pointer_offset ( LongIndex pointer_offset )

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::add_tile_offset ( TensorCoord const & coord )

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::load	(	Fragment &	frag,
		TensorCoord const &	tile_offset
	)

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::load ( Fragment & frag )

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::load_with_pointer_offset	(	Fragment &	frag,
		Index	pointer_offset
	)

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE RegularTileIterator2dThreadTile& cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::operator++ ( )

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE RegularTileIterator2dThreadTile& cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::operator-- ( )

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::store	(	Fragment const &	frag,
		TensorCoord const &	tile_offset
	)

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::store ( Fragment const & frag )

inline

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

CUTLASS_HOST_DEVICE void cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::store_with_pointer_offset	(	Fragment const &	frag,
		Index	pointer_offset
	)

inline

Member Data Documentation

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

int const cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::kAdvanceRank = AdvanceRank

static

template<typename Shape_ , typename Element_ , int AdvanceRank, typename ThreadMap_ , int Alignment>

int const cutlass::transform::threadblock::RegularTileIterator2dThreadTile< Shape_, Element_, layout::ColumnMajorInterleaved< 4 >, AdvanceRank, ThreadMap_, Alignment >::kAlignment = Alignment

static

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

regular_tile_iterator_pitch_linear_2dthreadtile.h