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
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
|
/* Copyright 2015 The TensorFlow Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
==============================================================================*/
#if GOOGLE_CUDA
#define EIGEN_USE_GPU
#include "tensorflow/core/kernels/xent_op.h"
#include "tensorflow/core/framework/tensor_types.h"
#include "tensorflow/core/platform/types.h"
namespace tensorflow {
typedef Eigen::GpuDevice GPUDevice;
// Partial specialization for a GPUDevice, that uses the Eigen implementation
// from XentEigenImpl.
namespace functor {
template <typename T>
struct XentFunctor<GPUDevice, T> {
void operator()(const GPUDevice &d,
const Eigen::DSizes<Eigen::DenseIndex, 2> &shape,
const Eigen::array<Eigen::DenseIndex, 2> &logits_bcast,
const Eigen::array<Eigen::DenseIndex, 2> &labels_bcast,
typename TTypes<T>::ConstMatrix logits,
typename TTypes<T>::ConstMatrix labels,
typename TTypes<T>::Matrix scratch,
typename TTypes<T>::Vec loss,
typename TTypes<T>::Matrix backprop) {
XentEigenImpl<GPUDevice, T>::Compute(d, shape, logits_bcast, labels_bcast,
logits, labels, scratch, loss,
backprop);
}
};
} // end namespace functor
// Instantiate the GPU implementation for half, float and double.
template struct functor::XentFunctor<GPUDevice, Eigen::half>;
template struct functor::XentFunctor<GPUDevice, float>;
template struct functor::XentFunctor<GPUDevice, double>;
} // end namespace tensorflow
#endif // GOOGLE_CUDA
|
