ClFullyConnectedWorkload Class Reference

#include <ClFullyConnectedWorkload.hpp>

Inheritance diagram for ClFullyConnectedWorkload:

Public Member Functions
	ClFullyConnectedWorkload (const armnn::FullyConnectedQueueDescriptor &descriptor, const armnn::WorkloadInfo &info, std::shared_ptr< arm_compute::MemoryManagerOnDemand > &memoryManager)
void	Execute () const override
Public Member Functions inherited from BaseWorkload< armnn::FullyConnectedQueueDescriptor >
	BaseWorkload (const armnn::FullyConnectedQueueDescriptor &descriptor, const WorkloadInfo &info)
void	PostAllocationConfigure () override
const armnn::FullyConnectedQueueDescriptor &	GetData () const
profiling::ProfilingGuid	GetGuid () const final
Public Member Functions inherited from IWorkload
virtual	~IWorkload ()
virtual void	RegisterDebugCallback (const DebugCallbackFunction &)

Additional Inherited Members
Protected Attributes inherited from BaseWorkload< armnn::FullyConnectedQueueDescriptor >
const armnn::FullyConnectedQueueDescriptor	m_Data
const profiling::ProfilingGuid	m_Guid

Detailed Description

Definition at line 25 of file ClFullyConnectedWorkload.hpp.

Constructor & Destructor Documentation

ClFullyConnectedWorkload()

ClFullyConnectedWorkload	(	const armnn::FullyConnectedQueueDescriptor &	descriptor,
		const armnn::WorkloadInfo &	info,
		std::shared_ptr< arm_compute::MemoryManagerOnDemand > &	memoryManager
	)

Definition at line 48 of file ClFullyConnectedWorkload.cpp.

    : BaseWorkload<FullyConnectedQueueDescriptor>(descriptor, info)
    , m_FullyConnectedLayer(memoryManager)
{
    m_WeightsTensor = std::make_unique<arm_compute::CLTensor>();
    BuildArmComputeTensor(*m_WeightsTensor, m_Data.m_Weight->GetTensorInfo());

    if (m_Data.m_Parameters.m_BiasEnabled)
    {
        m_BiasesTensor = std::make_unique<arm_compute::CLTensor>();
        BuildArmComputeTensor(*m_BiasesTensor, m_Data.m_Bias->GetTensorInfo());
    }

    m_Data.ValidateInputsOutputs("ClFullyConnectedWorkload", 1, 1);

    arm_compute::ICLTensor& input  = static_cast<IClTensorHandle*>(m_Data.m_Inputs[0])->GetTensor();
    arm_compute::ICLTensor& output = static_cast<IClTensorHandle*>(m_Data.m_Outputs[0])->GetTensor();

    const arm_compute::ActivationLayerInfo activationInfo = ConvertAdditionalInfoToAclActivationLayerInfo(descriptor);

    arm_compute::FullyConnectedLayerInfo fc_info =
            ConvertFullyConnectedDescriptorToAclFullyConnectedLayerInfo(descriptor.m_Parameters, activationInfo);

    m_FullyConnectedLayer.configure(&input, m_WeightsTensor.get(), m_BiasesTensor.get(), &output, fc_info);

    InitializeArmComputeClTensorData(*m_WeightsTensor, m_Data.m_Weight);

    if (m_BiasesTensor)
    {
        InitializeArmComputeClTensorData(*m_BiasesTensor, m_Data.m_Bias);
    }

    // Force Compute Library to perform the necessary copying and reshaping, after which
    // delete all the input tensors that will no longer be needed
    m_FullyConnectedLayer.prepare();
    FreeUnusedTensors();
}

Member Function Documentation

Execute()

void Execute ( ) const

overridevirtual

Implements IWorkload.

Definition at line 87 of file ClFullyConnectedWorkload.cpp.

References ARMNN_SCOPED_PROFILING_EVENT_CL, CHECK_LOCATION, and armnn::RunClFunction().

{
    ARMNN_SCOPED_PROFILING_EVENT_CL("ClFullyConnectedWorkload_Execute");
    RunClFunction(m_FullyConnectedLayer, CHECK_LOCATION());
}

---

The documentation for this class was generated from the following files:

• src/backends/cl/workloads/ClFullyConnectedWorkload.hpp
• src/backends/cl/workloads/ClFullyConnectedWorkload.cpp