diff options
author | Felix Thomasmathibalan <felixjohnny.thomasmathibalan@arm.com> | 2023-09-27 17:46:17 +0100 |
---|---|---|
committer | felixjohnny.thomasmathibalan <felixjohnny.thomasmathibalan@arm.com> | 2023-09-28 12:08:05 +0000 |
commit | afd38f0c617d6f89b2b4532c6c44f116617e2b6f (patch) | |
tree | 03bc7d5a762099989b16a656fa8d397b490ed70e /src/gpu/cl/operators/ClIndirectConv2d.cpp | |
parent | bdcb4c148ee2fdeaaddf4cf1e57bbb0de02bb894 (diff) | |
download | ComputeLibrary-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/gpu/cl/operators/ClIndirectConv2d.cpp')
-rw-r--r-- | src/gpu/cl/operators/ClIndirectConv2d.cpp | 52 |
1 files changed, 31 insertions, 21 deletions
diff --git a/src/gpu/cl/operators/ClIndirectConv2d.cpp b/src/gpu/cl/operators/ClIndirectConv2d.cpp index b900974574..777fc9e5e1 100644 --- a/src/gpu/cl/operators/ClIndirectConv2d.cpp +++ b/src/gpu/cl/operators/ClIndirectConv2d.cpp @@ -27,16 +27,15 @@ #include "arm_compute/core/Types.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/runtime/CL/CLScheduler.h" + +#include "src/common/utils/Log.h" +#include "src/core/helpers/MemoryHelpers.h" #include "src/gpu/cl/kernels/ClIndirectConv2dAddressPrecalculationKernel.h" #include "src/gpu/cl/kernels/ClIndirectConv2dKernel.h" +#include "src/gpu/cl/utils/ClAuxTensorHandler.h" #include "src/runtime/heuristics/indirect_conv/ClIndirectConvKernelConfig.h" #include "src/runtime/heuristics/indirect_conv/IClIndirectConvKernelConfig.h" -#include "src/core/helpers/MemoryHelpers.h" -#include "src/gpu/cl/utils/ClAuxTensorHandler.h" - -#include "src/common/utils/Log.h" - using namespace arm_compute::cl_indirect_conv; namespace arm_compute @@ -47,7 +46,8 @@ using namespace arm_compute::experimental; namespace { -DirectConvComputeKernelInfo config_indirect_convolution_nhwc(const ITensorInfo *src, const ITensorInfo *weights, const PadStrideInfo &conv_info) +DirectConvComputeKernelInfo +config_indirect_convolution_nhwc(const ITensorInfo *src, const ITensorInfo *weights, const PadStrideInfo &conv_info) { // Get GPU target GPUTarget gpu_target = CLScheduler::get().target(); @@ -59,8 +59,13 @@ DirectConvComputeKernelInfo config_indirect_convolution_nhwc(const ITensorInfo * } // namespace -void ClIndirectConv2d::configure(const CLCompileContext &compile_context, ITensorInfo *src, ITensorInfo *weights, ITensorInfo *biases, ITensorInfo *dst, - const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info) +void ClIndirectConv2d::configure(const CLCompileContext &compile_context, + ITensorInfo *src, + ITensorInfo *weights, + ITensorInfo *biases, + ITensorInfo *dst, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info) { ARM_COMPUTE_ERROR_ON_NULLPTR(src); ARM_COMPUTE_LOG_PARAMS(src, weights, biases, dst, conv_info, act_info); @@ -86,25 +91,29 @@ void ClIndirectConv2d::configure(const CLCompileContext &compile_context, ITenso CLScheduler::get().tune_kernel_static(*_indirect_conv_kernel); // Request memory for the indirect buffer - _aux_mem[IndirectBuffer] = MemoryInfo(offset_int_vec(IndirectBuffer), MemoryLifetime::Persistent, _indirect_buffer.total_size()); + _aux_mem[IndirectBuffer] = + MemoryInfo(offset_int_vec(IndirectBuffer), MemoryLifetime::Persistent, _indirect_buffer.total_size()); } -Status ClIndirectConv2d::validate(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *dst, - const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info) +Status ClIndirectConv2d::validate(const ITensorInfo *src, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *dst, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info) { // Initialize the direct convolution descriptor const DirectConvComputeKernelInfo desc = config_indirect_convolution_nhwc(src, weights, conv_info); - TensorShape ind_buffer_shape = misc::shape_calculator::compute_indirect_buffer_shape(src->tensor_shape(), - src->data_layout(), - weights->tensor_shape(), - conv_info, - desc); + TensorShape ind_buffer_shape = misc::shape_calculator::compute_indirect_buffer_shape( + src->tensor_shape(), src->data_layout(), weights->tensor_shape(), conv_info, desc); TensorInfo indirect_buffer(ind_buffer_shape, 1, DataType::S32); - ARM_COMPUTE_RETURN_ON_ERROR(kernels::ClIndirectConv2dAddressPrecalculationKernel::validate(src, weights, &indirect_buffer, conv_info, desc)); - ARM_COMPUTE_RETURN_ON_ERROR(kernels::ClIndirectConv2dKernel::validate(src, weights, biases, &indirect_buffer, dst, conv_info, act_info, desc)); + ARM_COMPUTE_RETURN_ON_ERROR(kernels::ClIndirectConv2dAddressPrecalculationKernel::validate( + src, weights, &indirect_buffer, conv_info, desc)); + ARM_COMPUTE_RETURN_ON_ERROR(kernels::ClIndirectConv2dKernel::validate(src, weights, biases, &indirect_buffer, dst, + conv_info, act_info, desc)); return Status{}; } @@ -124,9 +133,10 @@ void ClIndirectConv2d::run(ITensorPack &tensors) void ClIndirectConv2d::prepare(ITensorPack &constants) { - if(!_is_prepared) + if (!_is_prepared) { - ICLTensor *indirect_buffer_aux = utils::cast::polymorphic_downcast<ICLTensor *>(constants.get_tensor(offset_int_vec(IndirectBuffer))); + ICLTensor *indirect_buffer_aux = + utils::cast::polymorphic_downcast<ICLTensor *>(constants.get_tensor(offset_int_vec(IndirectBuffer))); ARM_COMPUTE_ERROR_ON(indirect_buffer_aux == nullptr); ARM_COMPUTE_LOG_INFO_WITH_FUNCNAME_ACL("Preparing indirect buffer"); @@ -134,7 +144,7 @@ void ClIndirectConv2d::prepare(ITensorPack &constants) CLAuxTensorHandler indirect_buffer(_indirect_buffer, *indirect_buffer_aux); ARM_COMPUTE_ERROR_ON(indirect_buffer.get()->cl_buffer().get() == nullptr); - ITensorPack indirect_buffer_pack{ { ACL_DST, indirect_buffer.get() } }; + ITensorPack indirect_buffer_pack{{ACL_DST, indirect_buffer.get()}}; CLScheduler::get().enqueue_op(*_addr_precalculation_kernel, indirect_buffer_pack, true); _is_prepared = true; |