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
75
76
77
78
79
80
81
82
|
/*
* Copyright (c) 2017 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to
* deal in the Software without restriction, including without limitation the
* rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
* sell copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#include "arm_compute/core/CL/kernels/CLTransposeKernel.h"
#include "arm_compute/core/AccessWindowTranspose.h"
#include "arm_compute/core/CL/CLHelpers.h"
#include "arm_compute/core/CL/CLKernelLibrary.h"
#include "arm_compute/core/CL/ICLTensor.h"
#include "arm_compute/core/CL/OpenCL.h"
#include "arm_compute/core/Error.h"
#include "arm_compute/core/Helpers.h"
#include "arm_compute/core/Types.h"
#include <set>
#include <sstream>
#include <string>
using namespace arm_compute;
void CLTransposeKernel::configure(const ICLTensor *input, ICLTensor *output)
{
ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::U8, DataType::S8, DataType::QS8, DataType::U16, DataType::S16, DataType::U32, DataType::S32, DataType::F16, DataType::F32);
ARM_COMPUTE_ERROR_ON(output == nullptr);
TensorShape output_shape{ input->info()->tensor_shape() };
const size_t w_out = input->info()->dimension(1);
const size_t h_out = input->info()->dimension(0);
output_shape.set(0, w_out);
output_shape.set(1, h_out);
// Output tensor auto inizialitation if not yet initialized
auto_init_if_empty(*output->info(), output_shape, 1, input->info()->data_type(), input->info()->fixed_point_position());
ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
ARM_COMPUTE_ERROR_ON_MISMATCHING_DIMENSIONS(output->info()->tensor_shape(), output_shape);
_input = input;
_output = output;
_lws_hint = cl::NDRange(2, 8);
std::set<std::string> build_opts;
std::ostringstream data_type_in_bytes;
data_type_in_bytes << input->info()->element_size();
build_opts.emplace("-DDATA_TYPE_IN_BYTES=" + data_type_in_bytes.str());
_kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("transpose", build_opts));
// Configure kernel window
const unsigned int num_elems_processed_per_iteration = max_cl_vector_width / input->info()->element_size();
Window win = calculate_max_window(*input->info(), Steps(num_elems_processed_per_iteration, num_elems_processed_per_iteration));
AccessWindowRectangle input_access(input->info(), 0, 0, num_elems_processed_per_iteration, num_elems_processed_per_iteration);
AccessWindowTranspose output_access(output->info(), 0, 0, num_elems_processed_per_iteration, num_elems_processed_per_iteration);
update_window_and_padding(win, input_access, output_access);
output_access.set_valid_region(win, input->info()->valid_region());
ICLKernel::configure(win);
}
|
