1 /*
2 * Copyright (c) 2018-2021 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 "arm_compute/runtime/CL/functions/CLReduceMean.h"
25
26 #include "arm_compute/core/CL/ICLTensor.h"
27 #include "arm_compute/core/Error.h"
28 #include "arm_compute/core/Types.h"
29 #include "arm_compute/core/utils/misc/ShapeCalculator.h"
30 #include "src/core/CL/CLValidate.h"
31 #include "src/core/CL/kernels/CLFillBorderKernel.h"
32 #include "src/core/CL/kernels/CLReductionOperationKernel.h"
33 #include "src/core/helpers/AutoConfiguration.h"
34
35 #include "src/common/utils/Log.h"
36
37 namespace arm_compute
38 {
39 namespace
40 {
validate_config(const ITensorInfo * input,const Coordinates & reduction_axis,bool keep_dims,const ITensorInfo * output)41 Status validate_config(const ITensorInfo *input, const Coordinates &reduction_axis, bool keep_dims, const ITensorInfo *output)
42 {
43 ARM_COMPUTE_UNUSED(keep_dims);
44 ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
45 ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
46 ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::QASYMM8_SIGNED, DataType::F16, DataType::F32);
47 ARM_COMPUTE_RETURN_ERROR_ON(reduction_axis.num_dimensions() < 1);
48 ARM_COMPUTE_RETURN_ERROR_ON(reduction_axis.num_dimensions() > input->num_dimensions());
49
50 const unsigned int reduction_ops = reduction_axis.num_dimensions();
51 const int input_dims = input->num_dimensions();
52 Coordinates axis_local = reduction_axis;
53
54 for(unsigned int i = 0; i < axis_local.num_dimensions(); ++i)
55 {
56 //axis: The dimensions to reduce. Must be in the range [-rank(input_tensor), rank(input_tensor)).
57 ARM_COMPUTE_RETURN_ERROR_ON(axis_local[i] < (-static_cast<int>(input->num_dimensions())));
58 ARM_COMPUTE_RETURN_ERROR_ON(axis_local[i] >= static_cast<int>(input->num_dimensions()));
59 }
60
61 if(output->tensor_shape().total_size() != 0)
62 {
63 // Only validate if not using auto_init for the output tensor
64 TensorShape out_shape = input->tensor_shape();
65 // Validate output_shape only if not using auto_init
66 convert_negative_axis(axis_local, input_dims);
67 std::sort(axis_local.begin(), axis_local.begin() + reduction_ops);
68 for(unsigned int i = 0; i < reduction_ops; ++i)
69 {
70 ARM_COMPUTE_RETURN_ERROR_ON(axis_local[i] > 3);
71 ARM_COMPUTE_RETURN_ERROR_ON(static_cast<unsigned int>(axis_local[i]) > input->num_dimensions() - 1);
72 if(output->total_size() > 0 && keep_dims)
73 {
74 ARM_COMPUTE_RETURN_ERROR_ON(output->dimension(axis_local[i]) != 1);
75 }
76 if(keep_dims)
77 {
78 out_shape.set(axis_local[i], 1);
79 }
80 else
81 {
82 ARM_COMPUTE_RETURN_ERROR_ON(i > static_cast<unsigned int>(axis_local[i]));
83 const unsigned int remove_index = axis_local[i] - i;
84 ARM_COMPUTE_RETURN_ERROR_ON(remove_index >= out_shape.num_dimensions());
85 out_shape.remove_dimension(remove_index);
86 }
87 }
88 const TensorInfo out_info = input->clone()->set_tensor_shape(out_shape);
89 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(output, &out_info);
90 const bool requant = is_data_type_quantized(input->data_type()) && input->quantization_info() != output->quantization_info();
91 if(requant)
92 {
93 TensorInfo input_no_quant(input->clone()->set_data_type(DataType::F32));
94 CLDequantizationLayer::validate(input, &input_no_quant);
95 TensorInfo output_no_quant(output->clone()->set_data_type(DataType::F32));
96 CLQuantizationLayer::validate(&output_no_quant, output);
97 }
98 }
99 return Status{};
100 }
101 }
102
CLReduceMean(std::shared_ptr<IMemoryManager> memory_manager)103 CLReduceMean::CLReduceMean(std::shared_ptr<IMemoryManager> memory_manager)
104 : _memory_group(std::move(memory_manager)), _reduction_kernels(), _reduced_outs(), _reshape(), _dequant(), _requant(), _reduction_ops(), _keep_dims(), _do_requant(), _input_no_quant(),
105 _output_no_quant()
106 {
107 }
108
configure(ICLTensor * input,const Coordinates & reduction_axis,bool keep_dims,ICLTensor * output)109 void CLReduceMean::configure(ICLTensor *input, const Coordinates &reduction_axis, bool keep_dims, ICLTensor *output)
110 {
111 configure(CLKernelLibrary::get().get_compile_context(), input, reduction_axis, keep_dims, output);
112 }
113
configure(const CLCompileContext & compile_context,ICLTensor * input,const Coordinates & reduction_axis,bool keep_dims,ICLTensor * output)114 void CLReduceMean::configure(const CLCompileContext &compile_context, ICLTensor *input, const Coordinates &reduction_axis, bool keep_dims, ICLTensor *output)
115 {
116 // Perform validate step
117 ARM_COMPUTE_ERROR_THROW_ON(CLReduceMean::validate(input->info(), reduction_axis, keep_dims, output->info()));
118 ARM_COMPUTE_LOG_PARAMS(input, reduction_axis, keep_dims, output);
119
120 // Output auto inizialitation if not yet initialized
121 const TensorShape output_shape = arm_compute::misc::shape_calculator::calculate_reduce_mean_shape(input->info(), reduction_axis, keep_dims);
122 auto_init_if_empty(*output->info(), input->info()->clone()->set_tensor_shape(output_shape));
123
124 _do_requant = is_data_type_quantized(input->info()->data_type()) && input->info()->quantization_info() != output->info()->quantization_info();
125 _reduction_ops = reduction_axis.num_dimensions();
126 _reduction_kernels.resize(_reduction_ops);
127 _reduced_outs.resize(_reduction_ops - (keep_dims ? 1 : 0));
128 _keep_dims = keep_dims;
129
130 ICLTensor *tmp_input = input;
131 ICLTensor *tmp_output = output;
132 if(_do_requant)
133 {
134 _memory_group.manage(&_input_no_quant);
135 _memory_group.manage(&_output_no_quant);
136 TensorInfo output_no_quant_info = input->info()->clone()->set_tensor_shape(output_shape);
137 output_no_quant_info.set_data_type(DataType::F32);
138 auto_init_if_empty(*_output_no_quant.info(), output_no_quant_info);
139 auto_init_if_empty(*_input_no_quant.info(), input->info()->clone()->set_data_type(DataType::F32));
140 _dequant.configure(compile_context, input, &_input_no_quant);
141 tmp_input = &_input_no_quant;
142 tmp_output = &_output_no_quant;
143 }
144
145 Coordinates axis_local = reduction_axis;
146 const int input_dims = tmp_input->info()->num_dimensions();
147
148 convert_negative_axis(axis_local, input_dims);
149
150 // Perform reduction for every axis
151 for(int i = 0; i < _reduction_ops; ++i)
152 {
153 TensorShape out_shape = i == 0 ? tmp_input->info()->tensor_shape() : (&_reduced_outs[i - 1])->info()->tensor_shape();
154 out_shape.set(axis_local[i], 1);
155 auto in = (i == 0) ? tmp_input : (&_reduced_outs[i - 1]);
156
157 if(i == _reduction_ops - 1 && keep_dims)
158 {
159 _reduction_kernels[i].configure(compile_context, in, tmp_output, axis_local[i], ReductionOperation::MEAN_SUM);
160 }
161 else
162 {
163 _reduced_outs[i].allocator()->init(TensorInfo(out_shape, tmp_input->info()->num_channels(), tmp_input->info()->data_type(), tmp_input->info()->quantization_info()));
164 _memory_group.manage(&_reduced_outs[i]);
165 _reduction_kernels[i].configure(compile_context, in, &_reduced_outs[i], axis_local[i], ReductionOperation::MEAN_SUM);
166 }
167 }
168
169 // Allocate intermediate tensors
170 for(int i = 0; i < _reduction_ops - (keep_dims ? 1 : 0); ++i)
171 {
172 _reduced_outs[i].allocator()->allocate();
173 }
174
175 // Configure reshape layer if we want to drop the dimensions
176 if(!_keep_dims)
177 {
178 TensorShape out_shape = tmp_input->info()->tensor_shape();
179
180 // We have to sort the reduction axis vectors in order for remove_dimension
181 // to work properly
182 std::sort(axis_local.begin(), axis_local.begin() + _reduction_ops);
183 for(int i = 0; i < _reduction_ops; ++i)
184 {
185 out_shape.remove_dimension(axis_local[i] - i);
186 }
187 auto_init_if_empty(*tmp_output->info(), tmp_input->info()->clone()->set_tensor_shape(out_shape));
188 _reshape.configure(compile_context, &_reduced_outs[_reduction_ops - 1], tmp_output);
189 }
190 if(_do_requant)
191 {
192 _requant.configure(compile_context, &_output_no_quant, output);
193 _input_no_quant.allocator()->allocate();
194 _output_no_quant.allocator()->allocate();
195 }
196 }
197
validate(const ITensorInfo * input,const Coordinates & reduction_axis,bool keep_dims,const ITensorInfo * output)198 Status CLReduceMean::validate(const ITensorInfo *input, const Coordinates &reduction_axis, bool keep_dims, const ITensorInfo *output)
199 {
200 return validate_config(input, reduction_axis, keep_dims, output);
201 }
202
run()203 void CLReduceMean::run()
204 {
205 MemoryGroupResourceScope scope_mg(_memory_group);
206
207 if(_do_requant)
208 {
209 _dequant.run();
210 }
211 for(auto &kernel : _reduction_kernels)
212 {
213 kernel.run();
214 }
215 if(!_keep_dims)
216 {
217 _reshape.run();
218 }
219 if(_do_requant)
220 {
221 _requant.run();
222 }
223 }
224 } // namespace arm_compute
225