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
62
63
64
65
66
67
68
69
70
71
72
73
74
|
/*
* Copyright (c) 2018 Samsung Electronics Co., Ltd. All Rights Reserved
* Copyright (c) 2016-2018 ARM Limited.
*
* 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.
*/
#include "arm_compute/runtime/NEON/functions/NENormalizationLayerEx.h"
#include "arm_compute/runtime/NEON/NEScheduler.h"
using namespace arm_compute;
NENormalizationLayerEx::NENormalizationLayerEx(std::shared_ptr<IMemoryManager> memory_manager)
: _memory_group(std::move(memory_manager)), _norm_kernel(), _multiply_kernel(),
_border_handler(), _input_squared()
{
}
void NENormalizationLayerEx::configure(const ITensor *input, ITensor *output,
const NormalizationLayerInfo &norm_info)
{
ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
TensorInfo tensor_info(input->info()->tensor_shape(), 1, input->info()->data_type(),
input->info()->quantization_info());
_input_squared.allocator()->init(tensor_info);
// Manage intermediate buffers
_memory_group.manage(&_input_squared);
// Configure kernels
_norm_kernel.configure(input, &_input_squared, output, norm_info);
_multiply_kernel.configure(input, input, &_input_squared, 1.0f, ConvertPolicy::SATURATE,
RoundingPolicy::TO_ZERO);
_border_handler.configure(&_input_squared, _norm_kernel.border_size(), BorderMode::CONSTANT,
PixelValue(0.0f));
// Allocate the tensor once the configure methods have been called
_input_squared.allocator()->allocate();
}
Status NENormalizationLayerEx::validate(const ITensorInfo *input, const ITensorInfo *output,
const NormalizationLayerInfo &norm_info)
{
// Perform validation step
ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
ARM_COMPUTE_RETURN_ON_ERROR(
NENormalizationLayerExKernel::validate(input, input, output, norm_info));
ARM_COMPUTE_RETURN_ON_ERROR(NEPixelWiseMultiplicationKernel::validate(
input, input, output, 1.0f, ConvertPolicy::SATURATE, RoundingPolicy::TO_ZERO));
return Status{};
}
void NENormalizationLayerEx::run()
{
_memory_group.acquire();
NEScheduler::get().schedule(&_multiply_kernel, Window::DimY);
NEScheduler::get().schedule(&_border_handler, Window::DimY);
NEScheduler::get().schedule(&_norm_kernel, Window::DimY);
_memory_group.release();
}
|
