1 /*
2 * Copyright (c) 2018-2020 Arm Limited.
3 *
4 * SPDX-License-Identifier: MIT
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to
8 * deal in the Software without restriction, including without limitation the
9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10 * sell copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22 * SOFTWARE.
23 */
24 #include "src/core/CL/kernels/CLSpaceToBatchLayerKernel.h"
25
26 #include "arm_compute/core/CL/CLHelpers.h"
27 #include "arm_compute/core/CL/ICLTensor.h"
28 #include "arm_compute/core/utils/misc/ShapeCalculator.h"
29 #include "src/core/CL/CLValidate.h"
30 #include "src/core/helpers/AutoConfiguration.h"
31 #include "src/core/helpers/WindowHelpers.h"
32 #include "support/StringSupport.h"
33
34 using namespace arm_compute::misc::shape_calculator;
35
36 namespace arm_compute
37 {
38 namespace
39 {
validate_arguments(const ITensorInfo * input,const ITensorInfo * block_info,const ITensorInfo * paddings,const ITensorInfo * output)40 Status validate_arguments(const ITensorInfo *input, const ITensorInfo *block_info, const ITensorInfo *paddings, const ITensorInfo *output)
41 {
42 ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, block_info, paddings, output);
43 ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(block_info, 1, DataType::S32);
44 ARM_COMPUTE_RETURN_ERROR_ON(input->data_type() == DataType::UNKNOWN);
45 ARM_COMPUTE_RETURN_ERROR_ON(input->num_dimensions() > 4);
46 ARM_COMPUTE_RETURN_ERROR_ON(block_info->num_dimensions() > 1);
47 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(block_info->tensor_shape(), TensorShape{ 2 });
48 ARM_COMPUTE_RETURN_ERROR_ON(paddings->num_dimensions() > 2);
49 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(paddings->tensor_shape(), TensorShape{ 2, 2 });
50
51 // Validate output if initialized
52 if(output->total_size() != 0)
53 {
54 const DataLayout data_layout = input->data_layout();
55 const int idx_channel = get_data_layout_dimension_index(data_layout, DataLayoutDimension::CHANNEL);
56 ARM_COMPUTE_RETURN_ERROR_ON(input->tensor_shape()[idx_channel] != output->tensor_shape()[idx_channel]);
57 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
58 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_QUANTIZATION_INFO(input, output);
59 }
60
61 return Status{};
62 }
validate_arguments_static(const ITensorInfo * input,const int block_shape_x,const int block_shape_y,const Size2D & padding_left,const Size2D & padding_right,const ITensorInfo * output)63 Status validate_arguments_static(const ITensorInfo *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right,
64 const ITensorInfo *output)
65 {
66 ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
67 ARM_COMPUTE_RETURN_ERROR_ON(input->data_type() == DataType::UNKNOWN);
68 ARM_COMPUTE_RETURN_ERROR_ON(input->num_dimensions() > 4);
69 ARM_COMPUTE_RETURN_ERROR_ON(block_shape_x < 1 || block_shape_y < 1);
70
71 // Validate output if initialized
72 if(output->total_size() != 0)
73 {
74 TensorShape expected_output_shape = misc::shape_calculator::compute_space_to_batch_shape(input, block_shape_x, block_shape_y, padding_left, padding_right);
75 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(output->tensor_shape(), expected_output_shape);
76 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
77 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_QUANTIZATION_INFO(input, output);
78 }
79
80 return Status{};
81 }
82 } // namespace
83
CLSpaceToBatchLayerKernel()84 CLSpaceToBatchLayerKernel::CLSpaceToBatchLayerKernel()
85 : _input(nullptr), _block_shape(nullptr), _paddings(nullptr), _output(nullptr)
86 {
87 }
88
configure(const ICLTensor * input,const ICLTensor * block_shape,const ICLTensor * paddings,ICLTensor * output)89 void CLSpaceToBatchLayerKernel::configure(const ICLTensor *input, const ICLTensor *block_shape, const ICLTensor *paddings, ICLTensor *output)
90 {
91 configure(CLKernelLibrary::get().get_compile_context(), input, block_shape, paddings, output);
92 }
93
configure(const CLCompileContext & compile_context,const ICLTensor * input,const ICLTensor * block_shape,const ICLTensor * paddings,ICLTensor * output)94 void CLSpaceToBatchLayerKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *block_shape, const ICLTensor *paddings, ICLTensor *output)
95 {
96 ARM_COMPUTE_ERROR_ON_NULLPTR(input, block_shape, paddings, output);
97 ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), block_shape->info(), paddings->info(), output->info()));
98
99 _input = input;
100 _block_shape = block_shape;
101 _paddings = paddings;
102 _output = output;
103
104 const DataLayout data_layout = input->info()->data_layout();
105 const int idx_width = get_data_layout_dimension_index(data_layout, DataLayoutDimension::WIDTH);
106 const int idx_height = get_data_layout_dimension_index(data_layout, DataLayoutDimension::HEIGHT);
107 const int idx_batch = get_data_layout_dimension_index(data_layout, DataLayoutDimension::BATCHES);
108
109 // Create kernel
110 CLBuildOptions build_opts;
111 build_opts.add_option("-DDATA_TYPE=" + get_cl_unsigned_type_from_element_size(data_size_from_type(input->info()->data_type())));
112 build_opts.add_option("-DWIDTH_OUT=" + support::cpp11::to_string(output->info()->dimension(idx_width)));
113 build_opts.add_option("-DHEIGHT_OUT=" + support::cpp11::to_string(output->info()->dimension(idx_height)));
114 build_opts.add_option("-DBATCH_SIZE=" + support::cpp11::to_string(output->info()->dimension(idx_batch)));
115 build_opts.add_option("-DWIDTH_IN=" + support::cpp11::to_string(input->info()->dimension(idx_width)));
116 build_opts.add_option("-DHEIGHT_IN=" + support::cpp11::to_string(input->info()->dimension(idx_height)));
117 build_opts.add_option("-DBATCH_IN=" + support::cpp11::to_string(input->info()->dimension(idx_batch)));
118 _kernel = create_kernel(compile_context, "space_to_batch_" + lower_string(string_from_data_layout(input->info()->data_layout())), build_opts.options());
119
120 // Configure kernel window
121 Window win = calculate_max_window(*output->info(), Steps());
122 ICLKernel::configure_internal(win);
123 }
124
configure(const ICLTensor * input,const int block_shape_x,const int block_shape_y,const Size2D & padding_left,const Size2D & padding_right,ICLTensor * output)125 void CLSpaceToBatchLayerKernel::configure(const ICLTensor *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right,
126 ICLTensor *output)
127 {
128 configure(CLKernelLibrary::get().get_compile_context(), input, block_shape_x, block_shape_y, padding_left, padding_right, output);
129 }
130
configure(const CLCompileContext & compile_context,const ICLTensor * input,const int block_shape_x,const int block_shape_y,const Size2D & padding_left,const Size2D & padding_right,ICLTensor * output)131 void CLSpaceToBatchLayerKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left,
132 const Size2D &padding_right,
133 ICLTensor *output)
134 {
135 ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
136
137 TensorShape output_shape = misc::shape_calculator::compute_space_to_batch_shape(input->info(), block_shape_x, block_shape_y, padding_left, padding_right);
138 auto_init_if_empty(*output->info(), output_shape, 1, input->info()->data_type(), input->info()->quantization_info());
139
140 ARM_COMPUTE_ERROR_THROW_ON(validate_arguments_static(input->info(), block_shape_x, block_shape_y, padding_left, padding_right, output->info()));
141
142 _input = input;
143 _output = output;
144
145 const DataLayout data_layout = input->info()->data_layout();
146 const int idx_width = get_data_layout_dimension_index(data_layout, DataLayoutDimension::WIDTH);
147 const int idx_height = get_data_layout_dimension_index(data_layout, DataLayoutDimension::HEIGHT);
148 const int idx_batch = get_data_layout_dimension_index(data_layout, DataLayoutDimension::BATCHES);
149
150 // Create kernel
151 CLBuildOptions build_opts;
152 build_opts.add_option("-DDATA_TYPE=" + get_cl_unsigned_type_from_element_size(data_size_from_type(input->info()->data_type())));
153 build_opts.add_option("-DWIDTH_OUT=" + support::cpp11::to_string(output->info()->dimension(idx_width)));
154 build_opts.add_option("-DHEIGHT_OUT=" + support::cpp11::to_string(output->info()->dimension(idx_height)));
155 build_opts.add_option("-DBATCH_SIZE=" + support::cpp11::to_string(output->info()->dimension(idx_batch)));
156 build_opts.add_option("-DWIDTH_IN=" + support::cpp11::to_string(input->info()->dimension(idx_width)));
157 build_opts.add_option("-DHEIGHT_IN=" + support::cpp11::to_string(input->info()->dimension(idx_height)));
158 build_opts.add_option("-DBATCH_IN=" + support::cpp11::to_string(input->info()->dimension(idx_batch)));
159 build_opts.add_option("-DBLOCK_SHAPE_X=" + support::cpp11::to_string(block_shape_x));
160 build_opts.add_option("-DBLOCK_SHAPE_Y=" + support::cpp11::to_string(block_shape_y));
161 build_opts.add_option("-DPAD_LEFT_X=" + support::cpp11::to_string(padding_left.x()));
162 build_opts.add_option("-DPAD_RIGHT_X=" + support::cpp11::to_string(padding_right.x()));
163 build_opts.add_option("-DPAD_LEFT_Y=" + support::cpp11::to_string(padding_left.y()));
164 build_opts.add_option("-DPAD_RIGHT_Y=" + support::cpp11::to_string(padding_right.y()));
165 _kernel = create_kernel(compile_context, "space_to_batch_static_" + lower_string(string_from_data_layout(input->info()->data_layout())), build_opts.options());
166
167 // Configure kernel window
168 Window win = calculate_max_window(*output->info(), Steps());
169 ICLKernel::configure_internal(win);
170 }
171
validate(const ITensorInfo * input,const ITensorInfo * block_shape,const ITensorInfo * paddings,const ITensorInfo * output)172 Status CLSpaceToBatchLayerKernel::validate(const ITensorInfo *input, const ITensorInfo *block_shape, const ITensorInfo *paddings, const ITensorInfo *output)
173 {
174 ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, block_shape, paddings, output));
175 return Status{};
176 }
validate(const ITensorInfo * input,const int block_shape_x,const int block_shape_y,const Size2D & padding_left,const Size2D & padding_right,const ITensorInfo * output)177 Status CLSpaceToBatchLayerKernel::validate(const ITensorInfo *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right,
178 const ITensorInfo *output)
179 {
180 ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments_static(input, block_shape_x, block_shape_y, padding_left, padding_right, output));
181 return Status{};
182 }
183
run(const Window & window,cl::CommandQueue & queue)184 void CLSpaceToBatchLayerKernel::run(const Window &window, cl::CommandQueue &queue)
185 {
186 ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
187 ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
188
189 Window slice_out = window.first_slice_window_3D();
190
191 Window slice_in = window.first_slice_window_4D();
192 slice_in.set(Window::DimX, Window::Dimension(0, 0, 0));
193 slice_in.set(Window::DimY, Window::Dimension(0, 0, 0));
194 slice_in.set(Window::DimZ, Window::Dimension(0, 0, 0));
195 slice_in.set(3, Window::Dimension(0, 0, 0));
196
197 Window vector_slice = window.first_slice_window_1D();
198 vector_slice.set(Window::DimX, Window::Dimension(0, 0, 0));
199
200 Window padding_slice = window.first_slice_window_2D();
201 padding_slice.set(Window::DimX, Window::Dimension(0, 0, 0));
202 padding_slice.set(Window::DimY, Window::Dimension(0, 0, 0));
203
204 int batch_id = 0;
205 do
206 {
207 unsigned int idx = 0;
208 const bool cond = (_paddings != nullptr && _block_shape != nullptr);
209 add_4D_tensor_argument(idx, _input, slice_in);
210 add_2D_tensor_argument_if(cond, idx, _paddings, padding_slice);
211 add_1D_tensor_argument_if(cond, idx, _block_shape, vector_slice);
212
213 add_argument(idx, batch_id);
214 add_3D_tensor_argument(idx, _output, slice_out);
215 enqueue(queue, *this, slice_out, lws_hint());
216 ++batch_id;
217 }
218 while(window.slide_window_slice_3D(slice_out));
219 }
220 } // namespace arm_compute
221