aboutsummaryrefslogtreecommitdiff
path: root/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.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/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.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/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp')
-rw-r--r--src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp34
1 files changed, 19 insertions, 15 deletions
diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp
index c07fac0e0d..c3b1b3c8bc 100644
--- a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp
+++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp
@@ -24,16 +24,18 @@
#include "GpuCkwActivation.h"
#include "arm_compute/core/Error.h"
-#include "arm_compute/core/Validate.h"
#include "arm_compute/core/utils/helpers/AdjustVecSize.h"
+#include "arm_compute/core/Validate.h"
#include "ckw/TensorTileSampler.h"
+
#include "src/core/helpers/WindowHelpers.h"
-#include "src/dynamic_fusion/sketch/gpu/GpuKernelArgument.h"
-#include "src/dynamic_fusion/sketch/gpu/GpuKernelComponentGroup.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/WriterHelper.h"
#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwKernelWriter.h"
#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwScopedKernelWriter.h"
#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwVariableTable.h"
-#include "src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/WriterHelper.h"
+#include "src/dynamic_fusion/sketch/gpu/GpuKernelArgument.h"
+#include "src/dynamic_fusion/sketch/gpu/GpuKernelComponentGroup.h"
+
#include <string>
using namespace ckw;
@@ -87,24 +89,25 @@ inline TensorTileSampler create_sampler(GpuCkwScopedKernelWriter &writer, int32_
GpuCkwActivation::GpuCkwActivation(ComponentId id,
const ArgumentPack<ITensorInfo> &tensors,
const Attributes &attributes)
- : IGpuCkwComponentDriver{ id, tensors },
- _src{},
- _dst{},
- _attributes{ attributes }
+ : IGpuCkwComponentDriver{id, tensors}, _src{}, _dst{}, _attributes{attributes}
{
_src = this->tensors().get_const_tensor(TensorType::ACL_SRC_0);
_dst = this->tensors().get_const_tensor(TensorType::ACL_DST_0);
ARM_COMPUTE_ERROR_ON_NULLPTR(_src, _dst);
}
-void GpuCkwActivation::write_component_code(const ComponentGroup &comp_group, GpuCkwVariableTable &vtable, GpuCkwScopedKernelWriter writer) const
+void GpuCkwActivation::write_component_code(const ComponentGroup &comp_group,
+ GpuCkwVariableTable &vtable,
+ GpuCkwScopedKernelWriter writer) const
{
const auto root_window = comp_group.get_root_component()->ckw_component_driver()->get_window();
const unsigned int n0 = root_window.x().step();
const unsigned int m0 = root_window.y().step();
- GpuCkwComponentArgument *src = vtable.declare_variable(comp_group, writer, _src, TensorStorageType::ClBufferUint8Ptr, "src");
- GpuCkwComponentArgument *dst = vtable.declare_variable(comp_group, writer, _dst, TensorStorageType::ClBufferUint8Ptr, "dst");
+ GpuCkwComponentArgument *src =
+ vtable.declare_variable(comp_group, writer, _src, TensorStorageType::ClBufferUint8Ptr, "src");
+ GpuCkwComponentArgument *dst =
+ vtable.declare_variable(comp_group, writer, _dst, TensorStorageType::ClBufferUint8Ptr, "dst");
load_src_dst_tiles_and_prepare_sampler(writer, src, dst, m0, n0, create_sampler);
@@ -119,7 +122,7 @@ void GpuCkwActivation::write_component_code(const ComponentGroup &comp_group, Gp
const auto &constant_B = writer->declare_tile("B_VAL", _attributes.b());
// Perform the operation.
- switch(_attributes.activation())
+ switch (_attributes.activation())
{
case ActivationLayerInfo::ActivationFunction::LOGISTIC:
{
@@ -179,9 +182,10 @@ Window GpuCkwActivation::get_window() const
// Collapse Dim 1 (W) and Dim 2 (H) together, leave Dim 0 (C) unchanged
// This is in line with the collapsing convention used by operators like Conv2d
output_shape.collapse(2U, 1U);
- constexpr unsigned int vector_size_byte_opencl = 16;
- const unsigned int num_elems_processed_per_iteration = adjust_vec_size(vector_size_byte_opencl / _dst->element_size(), _dst->dimension(0));
- Window win = calculate_max_window(output_shape, Steps(num_elems_processed_per_iteration));
+ constexpr unsigned int vector_size_byte_opencl = 16;
+ const unsigned int num_elems_processed_per_iteration =
+ adjust_vec_size(vector_size_byte_opencl / _dst->element_size(), _dst->dimension(0));
+ Window win = calculate_max_window(output_shape, Steps(num_elems_processed_per_iteration));
return win;
}