SMAUG
Simulating Machine Learning Applications on gem5-Aladdin
|
Pad a given tensor in any number of dimensions with arbitrary size. More...
#include <padding_op.h>
Pad a given tensor in any number of dimensions with arbitrary size.
This has a software-based implementation.
Backend | The Backend that sets Alignment. |
Public Types | |
enum | { kInput, kNumInputs } |
enum | { kOutput, kNumOutputs } |
Public Member Functions | |
PaddingOp (const std::string &name, Workspace *workspace) | |
void | setPaddingSize (const RepeatedField< google::protobuf::int32 > &val) |
Set the paddingSize of the Tensor along each dimension. More... | |
void | setPaddingSize (std::vector< int > const &val) |
const std::vector< int > & | getPaddingSize () const |
void | run () override |
void | createAllTensors () override |
bool | validate () override |
|
inline |
Set the paddingSize of the Tensor along each dimension.
The paddingSize is orgainized as <{dim0_begin, dim0_end, dim1_begin, dim1_end, ... >
Definition at line 34 of file padding_op.h.