diff options
Diffstat (limited to 'src/gpu/cl/kernels/ClHeightConcatenateKernel.cpp')
| -rw-r--r-- | src/gpu/cl/kernels/ClHeightConcatenateKernel.cpp | 24 |
1 files changed, 14 insertions, 10 deletions
diff --git a/src/gpu/cl/kernels/ClHeightConcatenateKernel.cpp b/src/gpu/cl/kernels/ClHeightConcatenateKernel.cpp index 719201d1fe..2e1cefc6e7 100644 --- a/src/gpu/cl/kernels/ClHeightConcatenateKernel.cpp +++ b/src/gpu/cl/kernels/ClHeightConcatenateKernel.cpp @@ -30,10 +30,10 @@ #include "arm_compute/core/utils/helpers/AdjustVecSize.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/core/utils/StringUtils.h" + #include "src/core/CL/CLValidate.h" #include "src/core/helpers/WindowHelpers.h" #include "support/Cast.h" - #include "support/StringSupport.h" namespace arm_compute @@ -52,7 +52,7 @@ Status validate_arguments(const ITensorInfo *src, unsigned int height_offset, co ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(Window::DimY) + height_offset > dst->dimension(Window::DimY)); ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(0) != dst->dimension(0)); - for(size_t i = 2; i < Coordinates::num_max_dimensions; ++i) + for (size_t i = 2; i < Coordinates::num_max_dimensions; ++i) { ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(i) != dst->dimension(i)); } @@ -62,8 +62,7 @@ Status validate_arguments(const ITensorInfo *src, unsigned int height_offset, co } } // namespace -ClHeightConcatenateKernel::ClHeightConcatenateKernel() - : _height_offset(0) +ClHeightConcatenateKernel::ClHeightConcatenateKernel() : _height_offset(0) { _type = CLKernelType::ELEMENTWISE; } @@ -74,12 +73,15 @@ Status ClHeightConcatenateKernel::validate(const ITensorInfo *src, unsigned int return Status{}; } -void ClHeightConcatenateKernel::configure(const CLCompileContext &compile_context, ITensorInfo *src, unsigned int height_offset, ITensorInfo *dst) +void ClHeightConcatenateKernel::configure(const CLCompileContext &compile_context, + ITensorInfo *src, + unsigned int height_offset, + ITensorInfo *dst) { ARM_COMPUTE_ERROR_ON_NULLPTR(src, dst); ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(src, height_offset, dst)); - auto padding_info = get_padding_info({ src, dst }); + auto padding_info = get_padding_info({src, dst}); _height_offset = height_offset; @@ -90,9 +92,10 @@ void ClHeightConcatenateKernel::configure(const CLCompileContext &compile_contex build_opts.add_option("-DDATA_TYPE=" + get_cl_unsigned_type_from_element_size(src->element_size())); build_opts.add_option("-DVEC_SIZE=" + support::cpp11::to_string(num_elems_processed_per_iteration)); build_opts.add_option("-DHEIGHT_OFFSET=" + support::cpp11::to_string(_height_offset)); - build_opts.add_option("-DVEC_SIZE_LEFTOVER=" + support::cpp11::to_string(src->dimension(0) % num_elems_processed_per_iteration)); + build_opts.add_option("-DVEC_SIZE_LEFTOVER=" + + support::cpp11::to_string(src->dimension(0) % num_elems_processed_per_iteration)); - if(is_data_type_quantized_asymmetric(src->data_type()) && src->quantization_info() != dst->quantization_info()) + if (is_data_type_quantized_asymmetric(src->data_type()) && src->quantization_info() != dst->quantization_info()) { const UniformQuantizationInfo iq_info = src->quantization_info().uniform(); const UniformQuantizationInfo oq_info = dst->quantization_info().uniform(); @@ -125,8 +128,9 @@ void ClHeightConcatenateKernel::run_op(ITensorPack &tensors, const Window &windo ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this); ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window); - const auto src = utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC)); - auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST)); + const auto src = + utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC)); + auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST)); unsigned int idx = 0; add_4D_tensor_argument(idx, src, window); |