blob: d5aaf9c3649360ae4fb239894b4d760e451755d3 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
|
#if GOOGLE_CUDA
#define EIGEN_USE_GPU
#include "tensorflow/core/kernels/softmax_op.h"
#include "tensorflow/core/platform/port.h"
#include "tensorflow/core/framework/tensor_types.h"
namespace tensorflow {
typedef Eigen::GpuDevice GPUDevice;
// Partial specialization for a GPUDevice, that uses the Eigen implementation
// from SoftmaxEigenImpl.
namespace functor {
template <typename T>
struct SoftmaxFunctor<GPUDevice, T> {
void operator()(const GPUDevice& d, typename TTypes<T>::ConstMatrix logits,
typename TTypes<T>::Matrix softmax) {
SoftmaxEigenImpl<GPUDevice, T>::Compute(d, logits, softmax);
}
};
} // end namespace functor
// Instantiate the GPU implementation for float.
template struct functor::SoftmaxFunctor<GPUDevice, float>;
} // end namespace tensorflow
#endif // GOOGLE_CUDA
|
