sum_internal.h

Go to the documentation of this file.

#pragma once

#include <vector>
#include "tensor/tensor.h"

namespace magmadnn {
namespace internal {

template <typename T>
void sum_full(std::vector<Tensor<T> *> &vals, Tensor<T> &out);


#if defined(_HAS_CUDA_)
template <typename T>
void sum_full_device(std::vector<Tensor<T> *> &vals, Tensor<T> &out);
#endif

}   // namespace internal
}   // namespace magmadnn