From 4646d2e026a0fa92085fcba2f4aec5ec148956aa Mon Sep 17 00:00:00 2001 From: Michele Di Giorgio Date: Wed, 19 Jun 2019 12:28:47 +0100 Subject: COMPMID-2072: Use a constexpr for num_elems_processed_per_iteration where possible Change-Id: I26cb699ae3a77003ef7d05ac30d3ed518214e25f Signed-off-by: Michele Di Giorgio Reviewed-on: https://review.mlplatform.org/c/1375 Comments-Addressed: Arm Jenkins Reviewed-by: Manuel Bottini Tested-by: Arm Jenkins Reviewed-by: Pablo Marquez --- src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp | 4 ++-- src/core/NEON/kernels/NEHistogramKernel.cpp | 14 ++++---------- .../kernels/NELocallyConnectedMatrixMultiplyKernel.cpp | 8 +++----- src/core/NEON/kernels/NEStridedSliceKernel.cpp | 4 +--- src/core/NEON/kernels/NEThresholdKernel.cpp | 6 ++---- 5 files changed, 12 insertions(+), 24 deletions(-) (limited to 'src/core/NEON') diff --git a/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp b/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp index 86bea849e4..8ee46eae42 100644 --- a/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp +++ b/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp @@ -36,6 +36,8 @@ namespace arm_compute { namespace { +constexpr unsigned int num_elems_processed_per_iteration = 16; + Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, float beta) { ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output); @@ -134,7 +136,6 @@ void NEGEMMMatrixAdditionKernel::configure(const ITensor *input, ITensor *output } // Configure kernel window - constexpr unsigned int num_elems_processed_per_iteration = 16; INESimpleKernel::configure(input, output, num_elems_processed_per_iteration); _beta = beta; @@ -142,7 +143,6 @@ void NEGEMMMatrixAdditionKernel::configure(const ITensor *input, ITensor *output Status NEGEMMMatrixAdditionKernel::validate(const ITensorInfo *input, const ITensorInfo *output, float beta) { - constexpr unsigned int num_elems_processed_per_iteration = 16; ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output, beta)); ARM_COMPUTE_RETURN_ON_ERROR(INESimpleKernel::validate(input->clone().get(), output->clone().get(), num_elems_processed_per_iteration)); return Status{}; diff --git a/src/core/NEON/kernels/NEHistogramKernel.cpp b/src/core/NEON/kernels/NEHistogramKernel.cpp index 02de566f6a..b088a232a8 100644 --- a/src/core/NEON/kernels/NEHistogramKernel.cpp +++ b/src/core/NEON/kernels/NEHistogramKernel.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -35,12 +35,9 @@ #include #include -using namespace arm_compute; - namespace arm_compute { class Coordinates; -} // namespace arm_compute inline void NEHistogramKernel::merge_histogram(uint32_t *global_hist, const uint32_t *local_hist, size_t bins) { @@ -217,9 +214,7 @@ void NEHistogramKernel::configure(const IImage *input, IDistribution1D *output, // Set appropriate function _func = &NEHistogramKernel::histogram_U8; - constexpr unsigned int num_elems_processed_per_iteration = 1; - - Window win = calculate_max_window(*input->info(), Steps(num_elems_processed_per_iteration)); + Window win = calculate_max_window(*input->info(), Steps()); INEKernel::configure(win); } @@ -236,9 +231,7 @@ void NEHistogramKernel::configure(const IImage *input, IDistribution1D *output) // Set appropriate function _func = &NEHistogramKernel::histogram_fixed_U8; - constexpr unsigned int num_elems_processed_per_iteration = 1; - - Window win = calculate_max_window(*input->info(), Steps(num_elems_processed_per_iteration)); + Window win = calculate_max_window(*input->info(), Steps()); INEKernel::configure(win); } @@ -251,3 +244,4 @@ void NEHistogramKernel::run(const Window &window, const ThreadInfo &info) (this->*_func)(window, info); } +} // namespace arm_compute diff --git a/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp b/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp index 46b7913223..467546a5d3 100644 --- a/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp +++ b/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017-2018 ARM Limited. + * Copyright (c) 2017-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -40,12 +40,9 @@ #include #include -using namespace arm_compute; - namespace arm_compute { class Coordinates; -} // namespace arm_compute namespace { @@ -317,7 +314,7 @@ Status validate_arguments(const ITensorInfo *input0, const ITensorInfo *input1, std::tuple validate_and_configure_window(ITensorInfo *input0, ITensorInfo *input1, ITensorInfo *output) { - const unsigned int num_elems_processed_per_iteration_x = 16; + constexpr unsigned int num_elems_processed_per_iteration_x = 16; Window win = calculate_max_window(*output, Steps(num_elems_processed_per_iteration_x)); @@ -389,3 +386,4 @@ void NELocallyConnectedMatrixMultiplyKernel::run(const Window &window, const Thr } } } +} // namespace arm_compute diff --git a/src/core/NEON/kernels/NEStridedSliceKernel.cpp b/src/core/NEON/kernels/NEStridedSliceKernel.cpp index 2ae029b64c..ece291e0a3 100644 --- a/src/core/NEON/kernels/NEStridedSliceKernel.cpp +++ b/src/core/NEON/kernels/NEStridedSliceKernel.cpp @@ -86,9 +86,7 @@ std::pair validate_and_configure_window(ITensorInfo *input, ITen auto_init_if_empty(*output, input->clone()->set_tensor_shape(output_shape)); // Create window - const unsigned int num_elems_processed_per_iteration = 1; - - Window win = calculate_max_window(*output, Steps(num_elems_processed_per_iteration)); + Window win = calculate_max_window(*output, Steps()); output->set_valid_region(ValidRegion(Coordinates(), output->tensor_shape())); return std::make_pair(Status{}, win); diff --git a/src/core/NEON/kernels/NEThresholdKernel.cpp b/src/core/NEON/kernels/NEThresholdKernel.cpp index ae9c62bc92..5c3b2a7540 100644 --- a/src/core/NEON/kernels/NEThresholdKernel.cpp +++ b/src/core/NEON/kernels/NEThresholdKernel.cpp @@ -30,12 +30,9 @@ #include -using namespace arm_compute; - namespace arm_compute { class Coordinates; -} // namespace arm_compute NEThresholdKernel::NEThresholdKernel() : _func(nullptr), _input(nullptr), _output(nullptr), _threshold(0), _false_value(0), _true_value(0), _upper(0) @@ -67,7 +64,7 @@ void NEThresholdKernel::configure(const ITensor *input, ITensor *output, uint8_t break; } - const unsigned int num_elems_processed_per_iteration = 16; + constexpr unsigned int num_elems_processed_per_iteration = 16; Window win = calculate_max_window(*input->info(), Steps(num_elems_processed_per_iteration)); AccessWindowHorizontal output_access(output->info(), 0, num_elems_processed_per_iteration); @@ -128,3 +125,4 @@ void NEThresholdKernel::run(const Window &window, const ThreadInfo &info) (this->*_func)(window); } +} // namespace arm_compute -- cgit v1.2.1