MagmaDNN  1.0
c++NeuralNetworkFramework
matmulop.h File Reference
#include <vector>
#include "compute/operation.h"
#include "compute/variable.h"
#include "compute/dot/dotop.h"
#include "compute/transpose/transposeop.h"
#include "tensor/tensor.h"
#include "math/matmul.h"
#include "math/dot.h"
#include "gemm_internal.h"
Include dependency graph for matmulop.h:
This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes

class  magmadnn::op::MatmulOp< T >
 

Functions

template<typename T >
MatmulOp< T > * magmadnn::op::matmul (Operation< T > *a, Operation< T > *b, bool needs_grad)
 
template<typename T >
MatmulOp< T > * magmadnn::op::matmul (T alpha, Operation< T > *a, Operation< T > *b, T beta, Tensor< T > *c, bool copy=true, bool needs_grad=true)
 

Detailed Description

Author
Daniel Nichols
Version
1.0
Date
2019-02-20

Function Documentation

◆ matmul() [1/2]

template<typename T >
MatmulOp<T>* magmadnn::op::matmul ( alpha,
Operation< T > *  a,
Operation< T > *  b,
beta,
Tensor< T > *  c,
bool  copy = true,
bool  needs_grad = true 
)

Computes the full gemm C = alpha*(AB) + beta*(C). Overwrites C and returns it if copy is false. If true, then it returns a copy of C.

Template Parameters
T
Parameters
alpha
a
b
beta
c
copy
Returns
MatmulOp<T>*

◆ matmul() [2/2]

template<typename T >
MatmulOp< T > * magmadnn::op::matmul ( Operation< T > *  a,
Operation< T > *  b,
bool  needs_grad = true 
)

Returns a new operation of type matmul. It computes the matrix product of A and B.

Template Parameters
T
Parameters
a
b
copy
Returns
MatmulOp<T>*
Here is the caller graph for this function: