1 /*
2 * Copyright (c) 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/CLGEMMLowpQuantizeDownInt32ScaleKernel.h"
25
26 #include "arm_compute/core/CL/CLHelpers.h"
27 #include "arm_compute/core/CL/ICLTensor.h"
28 #include "arm_compute/core/Helpers.h"
29 #include "arm_compute/core/Validate.h"
30 #include "arm_compute/core/utils/quantization/AsymmHelpers.h"
31 #include "src/core/AccessWindowStatic.h"
32 #include "src/core/helpers/AutoConfiguration.h"
33 #include "src/core/helpers/WindowHelpers.h"
34 #include "support/StringSupport.h"
35
36 namespace arm_compute
37 {
38 namespace
39 {
validate_arguments(const ITensorInfo * input,const ITensorInfo * bias,const ITensorInfo * output,const GEMMLowpOutputStageInfo * output_stage)40 Status validate_arguments(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output, const GEMMLowpOutputStageInfo *output_stage)
41 {
42 ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::S32);
43 ARM_COMPUTE_RETURN_ERROR_ON((output_stage->output_data_type != DataType::QASYMM8) && (output_stage->output_data_type != DataType::QASYMM8_SIGNED));
44 ARM_COMPUTE_RETURN_ERROR_ON(output_stage->gemmlowp_max_bound > std::get<1>(quantization::get_min_max_values_from_quantized_data_type(output_stage->output_data_type)));
45 ARM_COMPUTE_RETURN_ERROR_ON(output_stage->gemmlowp_min_bound < std::get<0>(quantization::get_min_max_values_from_quantized_data_type(output_stage->output_data_type))
46 || output_stage->gemmlowp_min_bound > output_stage->gemmlowp_max_bound);
47
48 // Check biases if exist
49 if(bias != nullptr)
50 {
51 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, bias);
52 ARM_COMPUTE_RETURN_ERROR_ON(bias->num_dimensions() > 1);
53 ARM_COMPUTE_RETURN_ERROR_ON(input->dimension(0) != bias->dimension(0));
54 }
55
56 if(output->total_size() != 0)
57 {
58 ARM_COMPUTE_RETURN_ERROR_ON_MSG(output->data_type() != output_stage->output_data_type, "Mismatching output data type");
59 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(input, output);
60 }
61
62 return Status{};
63 }
64 } //namespace
65
CLGEMMLowpQuantizeDownInt32ScaleKernel()66 CLGEMMLowpQuantizeDownInt32ScaleKernel::CLGEMMLowpQuantizeDownInt32ScaleKernel()
67 : _input(nullptr), _bias(nullptr), _output(nullptr)
68 {
69 }
70
validate(const ITensorInfo * input,const ITensorInfo * bias,const ITensorInfo * output,const GEMMLowpOutputStageInfo * output_stage)71 Status CLGEMMLowpQuantizeDownInt32ScaleKernel::validate(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output, const GEMMLowpOutputStageInfo *output_stage)
72 {
73 ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
74 ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, bias, output, output_stage));
75
76 return Status{};
77 }
78
configure(const ICLTensor * input,const ICLTensor * bias,ICLTensor * output,const GEMMLowpOutputStageInfo * output_stage)79 void CLGEMMLowpQuantizeDownInt32ScaleKernel::configure(const ICLTensor *input, const ICLTensor *bias, ICLTensor *output, const GEMMLowpOutputStageInfo *output_stage)
80 {
81 configure(CLKernelLibrary::get().get_compile_context(), input, bias, output, output_stage);
82 }
83
configure(const CLCompileContext & compile_context,const ICLTensor * input,const ICLTensor * bias,ICLTensor * output,const GEMMLowpOutputStageInfo * output_stage)84 void CLGEMMLowpQuantizeDownInt32ScaleKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *bias, ICLTensor *output,
85 const GEMMLowpOutputStageInfo *output_stage)
86 {
87 // Perform validate step
88 ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
89
90 ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(),
91 (bias != nullptr) ? bias->info() : nullptr,
92 output->info(),
93 output_stage));
94
95 auto padding_info = get_padding_info({ input, bias, output });
96
97 // Output auto inizialitation if not yet initialized
98 auto_init_if_empty(*output->info(), input->info()->clone()->set_data_type(output_stage->output_data_type));
99
100 _input = input;
101 _bias = bias;
102 _output = output;
103
104 const unsigned int num_elems_processed_per_iteration = adjust_vec_size(4, input->info()->dimension(0));
105
106 // Set the arguments to pass at compile time
107 auto min = output_stage->gemmlowp_min_bound;
108 auto max = output_stage->gemmlowp_max_bound;
109 CLBuildOptions build_opts;
110 build_opts.add_option("-DVEC_SIZE=" + support::cpp11::to_string(num_elems_processed_per_iteration));
111 build_opts.add_option("-DVEC_SIZE_LEFTOVER=" + support::cpp11::to_string(input->info()->dimension(0) % num_elems_processed_per_iteration));
112 build_opts.add_option("-DRESULT_OFFSET=" + support::cpp11::to_string(output_stage->gemmlowp_offset));
113 build_opts.add_option("-DRESULT_MULT_INT=" + support::cpp11::to_string(output_stage->gemmlowp_multiplier));
114 build_opts.add_option("-DRESULT_SHIFT=" + support::cpp11::to_string(output_stage->gemmlowp_shift));
115 build_opts.add_option_if((min > std::get<0>(quantization::get_min_max_values_from_quantized_data_type(output_stage->output_data_type))) && (min != max),
116 "-DMIN_BOUND=" + support::cpp11::to_string(min));
117 build_opts.add_option_if((max < std::get<1>(quantization::get_min_max_values_from_quantized_data_type(output_stage->output_data_type))) && (min != max),
118 "-DMAX_BOUND=" + support::cpp11::to_string(max));
119 build_opts.add_option("-DOUTPUT_DATA_TYPE=" + get_cl_type_from_data_type(output->info()->data_type()));
120 build_opts.add_option_if(bias != nullptr, "-DADD_BIAS");
121
122 // Create kernel
123 _kernel = create_kernel(compile_context, "gemmlowp_output_stage_quantize_down", build_opts.options());
124
125 // Configure kernel window
126 Window win = calculate_max_window(*input->info(), Steps(num_elems_processed_per_iteration));
127 ICLKernel::configure_internal(win);
128
129 ARM_COMPUTE_ERROR_ON(has_padding_changed(padding_info));
130 }
131
run(const Window & window,cl::CommandQueue & queue)132 void CLGEMMLowpQuantizeDownInt32ScaleKernel::run(const Window &window, cl::CommandQueue &queue)
133 {
134 ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
135 ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
136
137 Window collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
138 Window slice = collapsed.first_slice_window_3D();
139
140 unsigned int idx1 = num_arguments_per_3D_tensor();
141 if(_bias != nullptr)
142 {
143 Window biases_slice(slice);
144 biases_slice.set(Window::DimY, Window::Dimension(0, 1, 1));
145 biases_slice.set(Window::DimZ, Window::Dimension(0, 1, 1));
146 add_1D_tensor_argument(idx1, _bias, biases_slice);
147 }
148
149 do
150 {
151 unsigned int idx = 0;
152 add_3D_tensor_argument(idx, _input, slice);
153 add_3D_tensor_argument(idx1, _output, slice);
154 enqueue(queue, *this, slice, lws_hint());
155 }
156 while(collapsed.slide_window_slice_3D(slice));
157 }
158 }