![]() |
CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers
|
#include <tensor_op_multiplicand_sm75.h>
Public Types | |
| using | Index = int32_t |
| Index type used for coordinates. More... | |
| using | LongIndex = int64_t |
| Long index type used for offsets. More... | |
| using | TensorCoord = PitchLinearCoord |
| Logical coordinate. More... | |
| using | Stride = Coord< kStrideRank, Index, LongIndex > |
| Stride vector. More... | |
| using | TileShape = PitchLinearShape< kTileShapeContiguous, kTileShapeStride > |
| using | PartitionShape = PitchLinearShape< 4, 4 > |
| Fundamental partition shape in units of vectors. More... | |
| using | PartitionCount = PitchLinearShape< TileShape::kContiguous/PartitionShape::kContiguous, TileShape::kStrided/PartitionShape::kStrided > |
| using | AccessCount = PitchLinearShape< PartitionShape::kContiguous, PartitionShape::kStrided > |
Public Member Functions | |
| CUTLASS_HOST_DEVICE | TensorOpMultiplicand (Index ldm=0) |
| Ctor. More... | |
| CUTLASS_HOST_DEVICE | TensorOpMultiplicand (Stride stride) |
| Ctor. More... | |
| CUTLASS_HOST_DEVICE LongIndex | operator() (TensorCoord const &coord) const |
| CUTLASS_HOST_DEVICE Stride | stride () const |
| Returns the stride of the layout. More... | |
| CUTLASS_HOST_DEVICE Stride & | stride () |
| Returns the stride of the layout. More... | |
| CUTLASS_HOST_DEVICE LongIndex | capacity (TensorCoord const &extent) const |
Static Public Member Functions | |
| static CUTLASS_HOST_DEVICE TensorOpMultiplicand | packed (TensorCoord const &extent) |
| Helper returns a layout to a tightly packed tensor. More... | |
Static Public Attributes | |
| static int const | kRank = 2 |
| Logical rank of tensor. More... | |
| static int const | kStrideRank = 1 |
| Rank of stride vector. More... | |
| static int const | kAccessSize = 128 |
| This layout is optimized for 128b accesses. More... | |
| static int const | kElementSize = ElementSize |
| static int const | kElementsPerAccess = kAccessSize / kElementSize |
| static int const | kCrosswise = Crosswise |
| static int const | kTileShapeContiguous = 128 / (kAccessSize / 8) |
| static int const | kFactor |
| Number of kblocks to store PartitionShape::kContiguous Elements. More... | |
| static int const | kTileShapeStride |
Template based on element size (in bits) - defined in terms of pitch-linear memory and Crosswise size (in elements).
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::AccessCount = PitchLinearShape<PartitionShape::kContiguous, PartitionShape::kStrided> |
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::Index = int32_t |
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::LongIndex = int64_t |
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::PartitionCount = PitchLinearShape<TileShape::kContiguous / PartitionShape::kContiguous, TileShape::kStrided / PartitionShape::kStrided> |
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::PartitionShape = PitchLinearShape<4, 4> |
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::Stride = Coord<kStrideRank, Index, LongIndex> |
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::TensorCoord = PitchLinearCoord |
| using cutlass::layout::TensorOpMultiplicand< ElementSize, Crosswise >::TileShape = PitchLinearShape<kTileShapeContiguous, kTileShapeStride> |
Fundamental tile shape in units of vectors For TN kblock=32 and 8x8x16 shapes, TileShape = <8, 4>. For the rest, TileShape = <8, 8>
|
inline |
|
inline |
|
inline |
Compute the number of contiguous elements needed to store a tensor with the given size
|
inline |
Returns the offset of a coordinate in linear memory. Assumes coordinate has convention (contiguous, strided)
|
inlinestatic |
|
inline |
|
inline |
|
static |
|
static |
|
static |
|
static |
|
static |
|
static |
|
static |
|
static |
Contiguous dimension of the tile shape matches one shared memory cache line - 128B. For 128bit access size, it equals to 8 accesses.
|
static |
The strided dimension needs to be at least WarpSize(32) / kTileShapeContiguous for a warp to access. To ensure conflict free access, it also needs to be at least (kTileShapeContiguous / kFactor).
1.8.11