diff options
Diffstat (limited to 'src/core/CL/kernels/CLReorgLayerKernel.cpp')
-rw-r--r-- | src/core/CL/kernels/CLReorgLayerKernel.cpp | 52 |
1 files changed, 33 insertions, 19 deletions
diff --git a/src/core/CL/kernels/CLReorgLayerKernel.cpp b/src/core/CL/kernels/CLReorgLayerKernel.cpp index 065e25ea41..9fd21943e8 100644 --- a/src/core/CL/kernels/CLReorgLayerKernel.cpp +++ b/src/core/CL/kernels/CLReorgLayerKernel.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018-2020 ARM Limited. + * Copyright (c) 2018-2021, 2023 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -21,17 +21,19 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include "arm_compute/core/CL/kernels/CLReorgLayerKernel.h" +#include "src/core/CL/kernels/CLReorgLayerKernel.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/Error.h" #include "arm_compute/core/Helpers.h" #include "arm_compute/core/TensorInfo.h" -#include "arm_compute/core/Validate.h" -#include "arm_compute/core/Window.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" +#include "arm_compute/core/utils/StringUtils.h" +#include "arm_compute/core/Validate.h" + +#include "src/core/helpers/AutoConfiguration.h" +#include "src/core/helpers/WindowHelpers.h" #include "support/StringSupport.h" #include <string> @@ -50,13 +52,16 @@ Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, i const size_t idx_height = get_data_layout_dimension_index(input->data_layout(), DataLayoutDimension::HEIGHT); ARM_COMPUTE_RETURN_ERROR_ON(stride <= 0); - ARM_COMPUTE_RETURN_ERROR_ON_MSG((input->tensor_shape()[idx_width] % stride) != 0, "The width of the input tensor must be a multiple of stride"); - ARM_COMPUTE_RETURN_ERROR_ON_MSG((input->tensor_shape()[idx_height] % stride) != 0, "The height of the input tensor must be a multiple of stride"); + ARM_COMPUTE_RETURN_ERROR_ON_MSG((input->tensor_shape()[idx_width] % stride) != 0, + "The width of the input tensor must be a multiple of stride"); + ARM_COMPUTE_RETURN_ERROR_ON_MSG((input->tensor_shape()[idx_height] % stride) != 0, + "The height of the input tensor must be a multiple of stride"); // Validate output if initialized - if(output->total_size() != 0) + if (output->total_size() != 0) { - const TensorInfo tensor_info_output = output->clone()->set_tensor_shape(misc::shape_calculator::compute_reorg_output_shape(*input, stride)); + const TensorInfo tensor_info_output = + output->clone()->set_tensor_shape(misc::shape_calculator::compute_reorg_output_shape(*input, stride)); ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(output, &tensor_info_output); ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output); } @@ -65,9 +70,9 @@ Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, i } } // namespace -CLReorgLayerKernel::CLReorgLayerKernel() - : _input(nullptr), _output(nullptr) +CLReorgLayerKernel::CLReorgLayerKernel() : _input(nullptr), _output(nullptr) { + _type = CLKernelType::ELEMENTWISE; } void CLReorgLayerKernel::configure(const ICLTensor *input, ICLTensor *output, int32_t stride) @@ -75,16 +80,22 @@ void CLReorgLayerKernel::configure(const ICLTensor *input, ICLTensor *output, in configure(CLKernelLibrary::get().get_compile_context(), input, output, stride); } -void CLReorgLayerKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, int32_t stride) +void CLReorgLayerKernel::configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + int32_t stride) { ARM_COMPUTE_ERROR_ON_NULLPTR(input, output); ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), output->info(), stride)); + auto padding_info = get_padding_info({input, output}); _input = input; _output = output; - std::string kernel_name = std::string("reorg_layer_") + lower_string(string_from_data_layout(input->info()->data_layout())); - const size_t idx_channel = get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::CHANNEL); + std::string kernel_name = + std::string("reorg_layer_") + lower_string(string_from_data_layout(input->info()->data_layout())); + const size_t idx_channel = + get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::CHANNEL); // Create kernel CLBuildOptions build_opts; @@ -95,12 +106,13 @@ void CLReorgLayerKernel::configure(const CLCompileContext &compile_context, cons // Configure window // auto inizialize the output tensor if not yet initialized - auto_init_if_empty(*output->info(), input->info()->clone()->set_tensor_shape(misc::shape_calculator::compute_reorg_output_shape(*input->info(), stride))); + auto_init_if_empty(*output->info(), + input->info()->clone()->set_tensor_shape( + misc::shape_calculator::compute_reorg_output_shape(*input->info(), stride))); Window win = calculate_max_window(*output->info(), Steps()); // The CLWeightsReshapeKernel doesn't need padding so update_window_and_padding() can be skipped - output->info()->set_valid_region(ValidRegion(Coordinates(), output->info()->tensor_shape())); ICLKernel::configure_internal(win); _config_id = kernel_name; @@ -114,9 +126,12 @@ void CLReorgLayerKernel::configure(const CLCompileContext &compile_context, cons _config_id += support::cpp11::to_string(input->info()->dimension(2)); _config_id += "_"; _config_id += support::cpp11::to_string(stride); + ARM_COMPUTE_ERROR_ON(has_padding_changed(padding_info)); } -Status CLReorgLayerKernel::validate(const arm_compute::ITensorInfo *input, const arm_compute::ITensorInfo *output, int32_t stride) +Status CLReorgLayerKernel::validate(const arm_compute::ITensorInfo *input, + const arm_compute::ITensorInfo *output, + int32_t stride) { ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output, stride)); @@ -136,7 +151,6 @@ void CLReorgLayerKernel::run(const Window &window, cl::CommandQueue &queue) add_3D_tensor_argument(idx, _input, slice); add_3D_tensor_argument(idx, _output, slice); enqueue(queue, *this, slice, lws_hint()); - } - while(window.slide_window_slice_3D(slice)); + } while (window.slide_window_slice_3D(slice)); } } // namespace arm_compute |