|  | CUTLASS
    CUDA Templates for Linear Algebra Subroutines and Solvers | 
#include <regular_tile_access_iterator_tensor_op.h>
| Classes | |
| struct | Detail | 
| Internal details made public to facilitate introspection.  More... | |
| Public Types | |
| using | Shape = Shape_ | 
| using | Element = Element_ | 
| using | Layout = layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise > | 
| using | Index = typename Layout::Index | 
| using | LongIndex = typename Layout::LongIndex | 
| using | TensorRef = TensorRef< Element, Layout > | 
| using | TensorCoord = typename Layout::TensorCoord | 
| using | ThreadMap = ThreadMap_ | 
| using | AccessType = Array< Element, Layout::kElementsPerAccess > | 
| Element type per access.  More... | |
| Public Member Functions | |
| CUTLASS_HOST_DEVICE | RegularTileAccessIterator (TensorRef ref, int thread_id) | 
| Construct a TileIterator with zero threadblock offset.  More... | |
| CUTLASS_HOST_DEVICE void | set_iteration_index (int index) | 
| Overrides the internal iteration index.  More... | |
| CUTLASS_HOST_DEVICE void | add_pointer_offset (LongIndex pointer_offset) | 
| Adds a pointer offset in units of Element.  More... | |
| CUTLASS_HOST_DEVICE AccessType * | get () const | 
| Returns a pointer.  More... | |
| CUTLASS_HOST_DEVICE RegularTileAccessIterator & | operator++ () | 
| Advances to the next tile in memory.  More... | |
| CUTLASS_HOST_DEVICE RegularTileAccessIterator | operator++ (int) | 
| Advances to the next tile in memory.  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 | 
| static int const | kCrosswise = Crosswise | 
Tile iterator specialized for crosswise arrangements for TensorOps
Satisfies: ForwardTileIteratorConcept | ReadableContiguousTileIteratorConcept | WriteableContiguousTileIteratorConcept
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::AccessType = Array<Element, Layout::kElementsPerAccess> | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::Element = Element_ | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::Index = typename Layout::Index | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::Layout = layout::TensorOpMultiplicandCrosswise<sizeof_bits<Element_>::value, Crosswise> | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::LongIndex = typename Layout::LongIndex | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::Shape = Shape_ | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::TensorCoord = typename Layout::TensorCoord | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::TensorRef = TensorRef<Element, Layout> | 
| using cutlass::transform::threadblock::RegularTileAccessIterator< Shape_, Element_, layout::TensorOpMultiplicandCrosswise< sizeof_bits< Element_ >::value, Crosswise >, AdvanceRank, ThreadMap_, Alignment >::ThreadMap = ThreadMap_ | 
| 
 | inline | 
| ref | Pointer to start of tensor | 
| thread_id | ID of each participating thread | 
| 
 | inline | 
| 
 | inline | 
| 
 | inline | 
| 
 | inline | 
| 
 | inline | 
| 
 | inline | 
| 
 | static | 
| 
 | static | 
| 
 | static | 
 1.8.11
 1.8.11