diff options
author | telsoa01 <telmo.soares@arm.com> | 2018-08-31 09:22:23 +0100 |
---|---|---|
committer | telsoa01 <telmo.soares@arm.com> | 2018-08-31 09:22:23 +0100 |
commit | c577f2c6a3b4ddb6ba87a882723c53a248afbeba (patch) | |
tree | bd7d4c148df27f8be6649d313efb24f536b7cf34 /src/armnn/backends/NeonWorkloads/NeonAdditionFloat32Workload.cpp | |
parent | 4c7098bfeab1ffe1cdc77f6c15548d3e73274746 (diff) | |
download | armnn-c577f2c6a3b4ddb6ba87a882723c53a248afbeba.tar.gz |
Release 18.08
Diffstat (limited to 'src/armnn/backends/NeonWorkloads/NeonAdditionFloat32Workload.cpp')
-rw-r--r-- | src/armnn/backends/NeonWorkloads/NeonAdditionFloat32Workload.cpp | 20 |
1 files changed, 18 insertions, 2 deletions
diff --git a/src/armnn/backends/NeonWorkloads/NeonAdditionFloat32Workload.cpp b/src/armnn/backends/NeonWorkloads/NeonAdditionFloat32Workload.cpp index d1fb64093d..f26e42aff9 100644 --- a/src/armnn/backends/NeonWorkloads/NeonAdditionFloat32Workload.cpp +++ b/src/armnn/backends/NeonWorkloads/NeonAdditionFloat32Workload.cpp @@ -4,14 +4,30 @@ // #include "NeonAdditionFloat32Workload.hpp" +#include "backends/ArmComputeTensorUtils.hpp" #include "backends/CpuTensorHandle.hpp" namespace armnn { +arm_compute::Status NeonAdditionWorkloadValidate(const TensorInfo& input0, + const TensorInfo& input1, + const TensorInfo& output) +{ + const arm_compute::TensorInfo aclInput0 = armcomputetensorutils::BuildArmComputeTensorInfo(input0); + const arm_compute::TensorInfo aclInput1 = armcomputetensorutils::BuildArmComputeTensorInfo(input1); + const arm_compute::TensorInfo aclOutput = armcomputetensorutils::BuildArmComputeTensorInfo(output); + + return arm_compute::NEArithmeticAddition::validate(&aclInput0, + &aclInput1, + &aclOutput, + arm_compute::ConvertPolicy::SATURATE); +} + + NeonAdditionFloat32Workload::NeonAdditionFloat32Workload(const AdditionQueueDescriptor& descriptor, const WorkloadInfo& info) - : Float32Workload<AdditionQueueDescriptor>(descriptor, info) + : FloatWorkload<AdditionQueueDescriptor>(descriptor, info) { m_Data.ValidateInputsOutputs("NeonAdditionFloat32Workload", 2, 1); @@ -24,7 +40,7 @@ NeonAdditionFloat32Workload::NeonAdditionFloat32Workload(const AdditionQueueDesc void NeonAdditionFloat32Workload::Execute() const { - ARMNN_SCOPED_PROFILING_EVENT(Compute::CpuAcc, "NeonAdditionFloat32Workload_Execute"); + ARMNN_SCOPED_PROFILING_EVENT_NEON("NeonAdditionFloat32Workload_Execute"); m_AddLayer.run(); } |