aboutsummaryrefslogtreecommitdiff
path: root/src/cpu/kernels/CpuPool3dKernel.cpp
diff options
context:
space:
mode:
authorFelix Thomasmathibalan <felixjohnny.thomasmathibalan@arm.com>2023-09-27 17:46:17 +0100
committerfelixjohnny.thomasmathibalan <felixjohnny.thomasmathibalan@arm.com>2023-09-28 12:08:05 +0000
commitafd38f0c617d6f89b2b4532c6c44f116617e2b6f (patch)
tree03bc7d5a762099989b16a656fa8d397b490ed70e /src/cpu/kernels/CpuPool3dKernel.cpp
parentbdcb4c148ee2fdeaaddf4cf1e57bbb0de02bb894 (diff)
downloadComputeLibrary-afd38f0c617d6f89b2b4532c6c44f116617e2b6f.tar.gz
Apply clang-format on repository
Code is formatted as per a revised clang format configuration file(not part of this delivery). Version 14.0.6 is used. Exclusion List: - files with .cl extension - files that are not strictly C/C++ (e.g. Android.bp, Sconscript ...) And the following directories - compute_kernel_writer/validation/ - tests/ - include/ - src/core/NEON/kernels/convolution/ - src/core/NEON/kernels/arm_gemm/ - src/core/NEON/kernels/arm_conv/ - data/ There will be a follow up for formatting of .cl files and the files under tests/ and compute_kernel_writer/validation/. Signed-off-by: Felix Thomasmathibalan <felixjohnny.thomasmathibalan@arm.com> Change-Id: Ib7eb1fcf4e7537b9feaefcfc15098a804a3fde0a Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10391 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Diffstat (limited to 'src/cpu/kernels/CpuPool3dKernel.cpp')
-rw-r--r--src/cpu/kernels/CpuPool3dKernel.cpp75
1 files changed, 35 insertions, 40 deletions
diff --git a/src/cpu/kernels/CpuPool3dKernel.cpp b/src/cpu/kernels/CpuPool3dKernel.cpp
index 4504f3f7c9..8b484d4e0b 100644
--- a/src/cpu/kernels/CpuPool3dKernel.cpp
+++ b/src/cpu/kernels/CpuPool3dKernel.cpp
@@ -25,8 +25,9 @@
#include "arm_compute/core/TensorInfo.h"
#include "arm_compute/core/utils/misc/ShapeCalculator.h"
-#include "src/core/CPP/Validate.h"
+
#include "src/core/common/Registrars.h"
+#include "src/core/CPP/Validate.h"
#include "src/core/helpers/AutoConfiguration.h"
#include "src/core/helpers/WindowHelpers.h"
#include "src/cpu/kernels/pool3d/list.h"
@@ -41,39 +42,28 @@ namespace
{
using namespace misc::shape_calculator;
-static const std::vector<CpuPool3dKernel::Pooling3dKernel> available_kernels =
-{
- {
- "neon_qu8_ndhwc_poolMxNxD",
- [](const DataTypeISASelectorData & data) { return (data.dt == DataType::QASYMM8); },
- REGISTER_QASYMM8_NEON(arm_compute::cpu::neon_q8_pool3d)
- },
- {
- "neon_qs8_ndhwc_poolMxNxD",
- [](const DataTypeISASelectorData & data) { return (data.dt == DataType::QASYMM8_SIGNED); },
- REGISTER_QASYMM8_SIGNED_NEON(arm_compute::cpu::neon_q8_signed_pool3d)
- },
- {
- "neon_fp16_ndhwc_poolMxNxD",
- [](const DataTypeISASelectorData & data) { return (data.dt == DataType::F16 && data.isa.fp16); },
- REGISTER_FP16_NEON(arm_compute::cpu::neon_fp16_pool3d)
- },
- {
- "neon_fp32_ndhwc_poolMxNxD",
- [](const DataTypeISASelectorData & data) { return (data.dt == DataType::F32); },
- REGISTER_FP32_NEON(arm_compute::cpu::neon_fp32_pool3d)
- }
-};
+static const std::vector<CpuPool3dKernel::Pooling3dKernel> available_kernels = {
+ {"neon_qu8_ndhwc_poolMxNxD", [](const DataTypeISASelectorData &data) { return (data.dt == DataType::QASYMM8); },
+ REGISTER_QASYMM8_NEON(arm_compute::cpu::neon_q8_pool3d)},
+ {"neon_qs8_ndhwc_poolMxNxD",
+ [](const DataTypeISASelectorData &data) { return (data.dt == DataType::QASYMM8_SIGNED); },
+ REGISTER_QASYMM8_SIGNED_NEON(arm_compute::cpu::neon_q8_signed_pool3d)},
+ {"neon_fp16_ndhwc_poolMxNxD",
+ [](const DataTypeISASelectorData &data) { return (data.dt == DataType::F16 && data.isa.fp16); },
+ REGISTER_FP16_NEON(arm_compute::cpu::neon_fp16_pool3d)},
+ {"neon_fp32_ndhwc_poolMxNxD", [](const DataTypeISASelectorData &data) { return (data.dt == DataType::F32); },
+ REGISTER_FP32_NEON(arm_compute::cpu::neon_fp32_pool3d)}};
Status validate_arguments(const ITensorInfo *src, const ITensorInfo *dst, const Pooling3dLayerInfo &pool_info)
{
ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(src, dst);
ARM_COMPUTE_RETURN_ERROR_ON_MSG(src->data_layout() != DataLayout::NDHWC, "Only NDHWC layout supported");
ARM_COMPUTE_RETURN_ERROR_ON_CPU_F16_UNSUPPORTED(src);
- ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(src, 1, DataType::F16, DataType::F32, DataType::QASYMM8, DataType::QASYMM8_SIGNED);
+ ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(src, 1, DataType::F16, DataType::F32, DataType::QASYMM8,
+ DataType::QASYMM8_SIGNED);
- ARM_COMPUTE_RETURN_ERROR_ON_MSG((!is_data_type_float(src->data_type())) && (!pool_info.exclude_padding
- && (pool_info.pool_type == PoolingType::AVG)),
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG((!is_data_type_float(src->data_type())) &&
+ (!pool_info.exclude_padding && (pool_info.pool_type == PoolingType::AVG)),
"Exclude padding is unsupported for non-float types for Avg op");
const auto data_layout = src->data_layout();
@@ -97,21 +87,26 @@ Status validate_arguments(const ITensorInfo *src, const ITensorInfo *dst, const
int output_height = 0;
int output_depth = 0;
- ARM_COMPUTE_RETURN_ERROR_ON_MSG(is_pool_3d_region_entirely_outside_input(pool_info), "Pooling region that is entirely outside input tensor is unsupported");
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG(is_pool_3d_region_entirely_outside_input(pool_info),
+ "Pooling region that is entirely outside input tensor is unsupported");
- std::tie(output_width, output_height, output_depth) = scaled_3d_dimensions_signed(src->tensor_shape()[idx_width], src->tensor_shape()[idx_height], src->tensor_shape()[idx_depth],
- pool_size_x, pool_size_y, pool_size_z, pool_info);
- ARM_COMPUTE_RETURN_ERROR_ON_MSG((output_width < 1 || output_height < 1 || output_depth < 1), "Calculated output dimension size is invalid");
+ std::tie(output_width, output_height, output_depth) =
+ scaled_3d_dimensions_signed(src->tensor_shape()[idx_width], src->tensor_shape()[idx_height],
+ src->tensor_shape()[idx_depth], pool_size_x, pool_size_y, pool_size_z, pool_info);
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG((output_width < 1 || output_height < 1 || output_depth < 1),
+ "Calculated output dimension size is invalid");
- if(dst->total_size() != 0)
+ if (dst->total_size() != 0)
{
ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(src, dst);
ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_LAYOUT(src, dst);
- TensorInfo out_info(TensorInfo(compute_pool3d_shape(src->tensor_shape(), pool_info), 1, dst->data_type(), DataLayout::NDHWC));
+ TensorInfo out_info(
+ TensorInfo(compute_pool3d_shape(src->tensor_shape(), pool_info), 1, dst->data_type(), DataLayout::NDHWC));
ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(dst, &out_info);
}
- const auto *uk = CpuPool3dKernel::get_implementation(DataTypeISASelectorData{ src->data_type(), CPUInfo::get().get_isa() });
+ const auto *uk =
+ CpuPool3dKernel::get_implementation(DataTypeISASelectorData{src->data_type(), CPUInfo::get().get_isa()});
ARM_COMPUTE_RETURN_ERROR_ON(uk == nullptr || uk->ukernel == nullptr);
return Status{};
@@ -136,12 +131,12 @@ void CpuPool3dKernel::configure(const ITensorInfo *src, ITensorInfo *dst, const
// Update pool size in case of global pooling
const bool is_global_pooling = pool_info.is_global_pooling;
- const Size3D pool_size(
- is_global_pooling ? src->dimension(idx_width) : pool_info.pool_size.width,
- is_global_pooling ? src->dimension(idx_height) : pool_info.pool_size.height,
- is_global_pooling ? src->dimension(idx_depth) : pool_info.pool_size.depth);
+ const Size3D pool_size(is_global_pooling ? src->dimension(idx_width) : pool_info.pool_size.width,
+ is_global_pooling ? src->dimension(idx_height) : pool_info.pool_size.height,
+ is_global_pooling ? src->dimension(idx_depth) : pool_info.pool_size.depth);
- const auto *uk = CpuPool3dKernel::get_implementation(DataTypeISASelectorData{ src->data_type(), CPUInfo::get().get_isa() });
+ const auto *uk =
+ CpuPool3dKernel::get_implementation(DataTypeISASelectorData{src->data_type(), CPUInfo::get().get_isa()});
ARM_COMPUTE_ERROR_ON(uk == nullptr);
// Set instance variables
@@ -188,4 +183,4 @@ const std::vector<CpuPool3dKernel::Pooling3dKernel> &CpuPool3dKernel::get_availa
} // namespace kernels
} // namespace cpu
-} // namespace arm_compute \ No newline at end of file
+} // namespace arm_compute