diff options
Diffstat (limited to 'src/core/NEON/kernels/NEReorgLayerKernel.cpp')
| -rw-r--r-- | src/core/NEON/kernels/NEReorgLayerKernel.cpp | 65 |
1 files changed, 36 insertions, 29 deletions
diff --git a/src/core/NEON/kernels/NEReorgLayerKernel.cpp b/src/core/NEON/kernels/NEReorgLayerKernel.cpp index f8a8732c64..227570405c 100644 --- a/src/core/NEON/kernels/NEReorgLayerKernel.cpp +++ b/src/core/NEON/kernels/NEReorgLayerKernel.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018-2019 ARM Limited. + * Copyright (c) 2018-2021 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -21,15 +21,18 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include "arm_compute/core/NEON/kernels/NEReorgLayerKernel.h" +#include "src/core/NEON/kernels/NEReorgLayerKernel.h" #include "arm_compute/core/Error.h" #include "arm_compute/core/Helpers.h" #include "arm_compute/core/ITensor.h" #include "arm_compute/core/TensorInfo.h" #include "arm_compute/core/Types.h" -#include "arm_compute/core/Validate.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" +#include "arm_compute/core/Validate.h" + +#include "src/core/helpers/AutoConfiguration.h" +#include "src/core/helpers/WindowHelpers.h" #include <cstddef> #include <cstdint> @@ -40,7 +43,7 @@ namespace { Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, int32_t stride) { - //Note: ARM_COMPUTE_RETURN_ERROR_ON_CPU_F16_UNSUPPORTED(input) is not needed here as this kernel doesn't use NEON FP16 instructions. + //Note: ARM_COMPUTE_RETURN_ERROR_ON_CPU_F16_UNSUPPORTED(input) is not needed here as this kernel doesn't use CPU FP16 instructions. ARM_COMPUTE_RETURN_ERROR_ON(input->data_type() == DataType::UNKNOWN); ARM_COMPUTE_RETURN_ERROR_ON(input->data_layout() == DataLayout::UNKNOWN); @@ -48,13 +51,16 @@ 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); } @@ -63,8 +69,7 @@ Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, i } } // namespace -NEReorgLayerKernel::NEReorgLayerKernel() - : _input(nullptr), _output(nullptr), _stride(1) +NEReorgLayerKernel::NEReorgLayerKernel() : _input(nullptr), _output(nullptr), _stride(1) { } @@ -84,7 +89,6 @@ void NEReorgLayerKernel::configure(const ITensor *input, ITensor *output, int32_ _stride = stride; // The NEReorgLayerKernel doesn't need padding so update_window_and_padding() can be skipped - output->info()->set_valid_region(ValidRegion(Coordinates(), output->info()->tensor_shape())); // Configure kernel window Window win = calculate_max_window(*output->info(), Steps()); @@ -120,23 +124,26 @@ void NEReorgLayerKernel::run(const Window &window, const ThreadInfo &info) Iterator out(_output, collapsed_window); // Perform reorg - execute_window_loop(collapsed_window, [&](const Coordinates & id) - { - // Get spatial coords and channels - const unsigned int w = id[idx_w]; - const unsigned int h = id[idx_h]; - const unsigned int c = id[idx_c]; - - // Calculate mapping - const unsigned int offset = c / out_c; - Coordinates map_coords = id; - map_coords.set(idx_w, w * stride + offset % stride); - map_coords.set(idx_h, h * stride + offset / stride); - map_coords.set(idx_c, c % out_c); - - // Perform mapping - std::memcpy(out.ptr(), in_ptr + _input->info()->offset_element_in_bytes(map_coords), _input->info()->element_size()); - }, - out); + execute_window_loop( + collapsed_window, + [&](const Coordinates &id) + { + // Get spatial coords and channels + const unsigned int w = id[idx_w]; + const unsigned int h = id[idx_h]; + const unsigned int c = id[idx_c]; + + // Calculate mapping + const unsigned int offset = c / out_c; + Coordinates map_coords = id; + map_coords.set(idx_w, w * stride + offset % stride); + map_coords.set(idx_h, h * stride + offset / stride); + map_coords.set(idx_c, c % out_c); + + // Perform mapping + std::memcpy(out.ptr(), in_ptr + _input->info()->offset_element_in_bytes(map_coords), + _input->info()->element_size()); + }, + out); } } // namespace arm_compute |