| /* |
| * Copyright (c) 2017-2020 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/GLES_COMPUTE/kernels/GCGEMMMatrixAdditionKernel.h" |
| |
| #include "arm_compute/core/Error.h" |
| #include "arm_compute/core/GLES_COMPUTE/GCHelpers.h" |
| #include "arm_compute/core/GLES_COMPUTE/GCKernelLibrary.h" |
| #include "arm_compute/core/GLES_COMPUTE/IGCTensor.h" |
| #include "arm_compute/core/GLES_COMPUTE/OpenGLES.h" |
| #include "arm_compute/core/Helpers.h" |
| #include "arm_compute/core/Types.h" |
| #include "arm_compute/core/Validate.h" |
| #include "arm_compute/core/Window.h" |
| #include "support/StringSupport.h" |
| |
| using namespace arm_compute; |
| |
| GCGEMMMatrixAdditionKernel::GCGEMMMatrixAdditionKernel() |
| : _input(nullptr), _output(nullptr) |
| { |
| } |
| |
| void GCGEMMMatrixAdditionKernel::configure(const IGCTensor *input, IGCTensor *output, float beta) |
| { |
| ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::F32); |
| ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input, output); |
| ARM_COMPUTE_ERROR_ON(input->info()->dimension(0) != output->info()->dimension(0)); |
| ARM_COMPUTE_ERROR_ON(input->info()->dimension(1) != output->info()->dimension(1)); |
| |
| _input = input; |
| _output = output; |
| const unsigned int num_elems_processed_per_iteration = max_gc_vector_width / data_size_from_type(input->info()->data_type()); |
| |
| std::set<std::string> build_opts; |
| std::string dt_name = (input->info()->data_type() == DataType::F32) ? "DATA_TYPE_FP32" : "DATA_TYPE_FP16"; |
| build_opts.emplace(("#define " + dt_name)); |
| build_opts.emplace("#define LOCAL_SIZE_X " + support::cpp11::to_string(1)); |
| build_opts.emplace("#define LOCAL_SIZE_Y " + support::cpp11::to_string(1)); |
| build_opts.emplace("#define LOCAL_SIZE_Z " + support::cpp11::to_string(1)); |
| build_opts.emplace("#define BETA " + float_to_string_with_full_precision(beta)); |
| |
| // Create kernel |
| build_opts.emplace("#define GEMM_MATRIXADDITION"); |
| std::string data_type_name = lower_string(string_from_data_type(input->info()->data_type())); |
| _kernel = GCKernelLibrary::get().create_kernel(("gemm_ma"), build_opts); |
| |
| // Configure kernel window |
| Window win = calculate_max_window(*_input->info(), Steps(num_elems_processed_per_iteration)); |
| |
| AccessWindowHorizontal input_access(input->info(), 0, num_elems_processed_per_iteration); |
| AccessWindowHorizontal output_access(output->info(), 0, num_elems_processed_per_iteration); |
| |
| update_window_and_padding(win, input_access, output_access); |
| |
| output_access.set_valid_region(win, input->info()->valid_region()); |
| |
| IGCKernel::configure(win); |
| } |
| |
| void GCGEMMMatrixAdditionKernel::run(const Window &window) |
| { |
| ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this); |
| ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(IGCKernel::window(), window); |
| |
| _kernel.use(); |
| |
| Window slice = window.first_slice_window_2D(); |
| |
| do |
| { |
| unsigned int idx = 0; |
| add_2D_tensor_argument(idx, _input, 1, slice); |
| add_2D_tensor_argument(idx, _output, 2, slice); |
| |
| _kernel.update_shader_params(); |
| |
| enqueue(*this, slice); |
| } |
| while(window.slide_window_slice_2D(slice)); |
| } |