Repository navigation
Expand file tree
/
Copy pathgpu-impl-linear.h
More file actions
61 lines (42 loc) · 1.71 KB
/
Copy pathgpu-impl-linear.h
File metadata and controls
61 lines (42 loc) · 1.71 KB
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
58
59
60
61
// tensor/gpu-impl-linear.h
// Copyright 2019 Johns Hopkins University (author: Daniel Povey)
// See ../../COPYING for clarification regarding multiple authors
//
// 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
//
// THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
// WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
// MERCHANTABLITY OR NON-INFRINGEMENT.
// See the Apache 2 License for the specific language governing permissions and
// limitations under the License.
#ifndef KALDI_TENSOR_GPU_IMPL_LINEAR_H_
#define KALDI_TENSOR_GPU_IMPL_LINEAR_H_ 1
#include "tensor/tensor.h"
// This header actually contains implementations of functions that are required
// by tensor-impl-linear.cc. It should not be included by users of this
// library.
namespace kaldi {
namespace tensor {
template <typename Real>
inline static void AddProductScalar3GPU(
float alpha, float beta,
const TensorImpl &a, const TensorImpl &b, const TensorImpl *c) {
// TODO: make this actually work on GPU, probably by calling the 1-d vector version.
Real *a_data = static_cast<Real*>(a->data),
*b_data = static_cast<Real*>(b->data),
*c_data = static_cast<Real*>(c->data);
if (beta != 0.0) {
*c_data = (beta * *c_data) + alpha * (*a_data + *b_data);
} else { // don't propagate NaN
*c_data = alpha * (*a_data + *b_data);
}
}
}
} // namespace tensor
} // namespace kaldi
#endif // KALDI_TENSOR_GPU_IMPL_LINEAR_H_