CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers
|
Defines layout functions used by TensorRef and derived classes for common 4-D and 5-D tensor formats. More...
#include "assert.h"
#include "cutlass/cutlass.h"
#include "cutlass/fast_math.h"
#include "cutlass/layout/matrix.h"
#include "cutlass/coord.h"
#include "cutlass/tensor_coord.h"
Go to the source code of this file.
Classes | |
class | cutlass::layout::TensorNHWC |
Mapping function for 4-D NHWC tensors. More... | |
class | cutlass::layout::TensorNCHW |
Mapping function for 4-D NCHW tensors. More... | |
class | cutlass::layout::TensorNCxHWx< Interleave > |
Mapping function for 4-D NC/xHWx tensors. More... | |
class | cutlass::layout::TensorCxRSKx< Interleave > |
Mapping function for 4-D CxRSKx tensors. More... | |
Namespaces | |
cutlass | |
cutlass::layout | |
Layout functions map logical coordinates to linear memory. They often require additional data to describe strides between elements.
Layout functions must implement all members in the public interface of IdentityTensorLayout<> defined in cutlass/tensor_ref.h.