gemm_internal.h

Go to the documentation of this file.

#pragma once
#include "cblas.h"
#include "tensor/tensor.h"

#if defined(_HAS_CUDA_)
#include "magma.h"
#endif

namespace magmadnn {
namespace internal {

template <typename T>
bool gemm_check(Tensor<T> *A, Tensor<T> *B, Tensor<T> *C, unsigned int &M, unsigned int &N, unsigned int &K);

template <typename T>
void gemm_full(T alpha, Tensor<T>* A, Tensor<T>* B, T beta, Tensor<T>* C);


}   // namespace internal
}   // namespace magmadnn