1 | /* Copyright 2017 The TensorFlow Authors. All Rights Reserved. |
2 | |
3 | Licensed under the Apache License, Version 2.0 (the "License"); |
4 | you may not use this file except in compliance with the License. |
5 | You may obtain a copy of the License at |
6 | |
7 | http://www.apache.org/licenses/LICENSE-2.0 |
8 | |
9 | Unless required by applicable law or agreed to in writing, software |
10 | distributed under the License is distributed on an "AS IS" BASIS, |
11 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
12 | See the License for the specific language governing permissions and |
13 | limitations under the License. |
14 | ==============================================================================*/ |
15 | #ifndef TENSORFLOW_LITE_TOCO_ALLOCATE_TRANSIENT_ARRAYS_H_ |
16 | #define TENSORFLOW_LITE_TOCO_ALLOCATE_TRANSIENT_ARRAYS_H_ |
17 | |
18 | #include "tensorflow/lite/toco/model.h" |
19 | |
20 | namespace toco { |
21 | |
22 | // We align the allocated sizes to the next multiple of a cache line, |
23 | // to get simple performance characteristics without side effects of |
24 | // accesses to one buffer on accesses to another buffer. |
25 | // That also takes care of data type alignment for any reasonable type |
26 | // (no reasonable data type should have alignment greater than a cache line). |
27 | // Here we make CPU-centric assumptions, in particular, we assume 64-byte cache |
28 | // lines. Getting this wrong by a factor of 2x (if this ever changes) wouldn't |
29 | // be terrible. |
30 | // Embedded architectures may use a different value for alignment. |
31 | constexpr std::size_t kDefaultTransientDataAlignment = 64; |
32 | |
33 | // Rounds up dividend to a value divisible by divisor. |
34 | inline std::size_t RoundUpToNextMultipleOf(std::size_t dividend, |
35 | std::size_t divisor) { |
36 | return ((dividend + divisor - 1) / divisor) * divisor; |
37 | } |
38 | |
39 | void AllocateTransientArrays(Model* model, |
40 | std::size_t transient_data_alignment); |
41 | |
42 | } // namespace toco |
43 | |
44 | #endif // TENSORFLOW_LITE_TOCO_ALLOCATE_TRANSIENT_ARRAYS_H_ |
45 | |