NeonMultiplicationWorkload Class Reference

#include <NeonMultiplicationWorkload.hpp>

Inheritance diagram for NeonMultiplicationWorkload:

Collaboration diagram for NeonMultiplicationWorkload:

Public Member Functions
	NeonMultiplicationWorkload (const MultiplicationQueueDescriptor &descriptor, const WorkloadInfo &info)
virtual void	Execute () const override
Public Member Functions inherited from NeonBaseWorkload< MultiplicationQueueDescriptor >
	NeonBaseWorkload (const MultiplicationQueueDescriptor &descriptor, const WorkloadInfo &info)
void	ReplaceInputTensorHandle (ITensorHandle *tensorHandle, unsigned int slot) override
void	ReplaceOutputTensorHandle (ITensorHandle *tensorHandle, unsigned int slot) override
Public Member Functions inherited from BaseWorkload< MultiplicationQueueDescriptor >
	BaseWorkload (const MultiplicationQueueDescriptor &descriptor, const WorkloadInfo &info)
virtual const std::string &	GetName () const override
void	ExecuteAsync (ExecutionData &executionData) override
void	PostAllocationConfigure () override
const MultiplicationQueueDescriptor &	GetData () const
arm::pipe::ProfilingGuid	GetGuid () const final
virtual bool	SupportsTensorHandleReplacement () const override
Public Member Functions inherited from IWorkload
virtual	~IWorkload ()
virtual arm::pipe::ProfilingGuid	GetGuid () const =0
virtual bool	SupportsTensorHandleReplacement () const =0
virtual const std::string &	GetName () const =0
virtual void	RegisterDebugCallback (const DebugCallbackFunction &)
virtual armnn::Optional< armnn::MemoryRequirements >	GetMemoryRequirements ()

Additional Inherited Members
Protected Member Functions inherited from NeonBaseWorkload< MultiplicationQueueDescriptor >
virtual void	Reconfigure ()
Protected Attributes inherited from BaseWorkload< MultiplicationQueueDescriptor >
MultiplicationQueueDescriptor	m_Data
const arm::pipe::ProfilingGuid	m_Guid
const std::string	m_Name

Detailed Description

Definition at line 23 of file NeonMultiplicationWorkload.hpp.

Constructor & Destructor Documentation

NeonMultiplicationWorkload()

NeonMultiplicationWorkload	(	const MultiplicationQueueDescriptor &	descriptor,
		const WorkloadInfo &	info
	)

Definition at line 47 of file NeonMultiplicationWorkload.cpp.

    : NeonBaseWorkload<MultiplicationQueueDescriptor>(descriptor, info)
{
    m_Data.ValidateInputsOutputs("NeonMultiplicationWorkload", 2, 1);
 
    arm_compute::ITensor& input1 = PolymorphicDowncast<IAclTensorHandle*>(m_Data.m_Inputs[0])->GetTensor();
    arm_compute::ITensor& input2 = PolymorphicDowncast<IAclTensorHandle*>(m_Data.m_Inputs[1])->GetTensor();
    arm_compute::ITensor& output = PolymorphicDowncast<IAclTensorHandle*>(m_Data.m_Outputs[0])->GetTensor();
 
    auto convertPolicy = (IsQuantizedType(info.m_InputTensorInfos[0].GetDataType()) ||
                          IsQuantizedType(info.m_InputTensorInfos[1].GetDataType())) ?
                          arm_compute::ConvertPolicy::SATURATE :
                          arm_compute::ConvertPolicy::WRAP;
 
    const arm_compute::ActivationLayerInfo activationInfo = ConvertAdditionalInfoToAclActivationLayerInfo(descriptor);
 
    // At the time of writing, configure() will fail if a rounding policy other than TO_ZERO is supplied to it,
    // when providing a scale of 1.0 for F32 tensors, even though the provided rounding policy appears to be
    // ignored for F32 tensors.
    auto layer = std::make_unique<arm_compute::NEPixelWiseMultiplication>();
    layer->configure(&input1,
                     &input2,
                     &output,
                     1.0f,
                     convertPolicy,
                     arm_compute::RoundingPolicy::TO_ZERO,
                     activationInfo);
    m_PixelWiseMultiplication.reset(layer.release());
}

References armnn::ConvertAdditionalInfoToAclActivationLayerInfo(), armnn::info, armnn::IsQuantizedType(), BaseWorkload< MultiplicationQueueDescriptor >::m_Data, QueueDescriptor::m_Inputs, QueueDescriptor::m_Outputs, and QueueDescriptor::ValidateInputsOutputs().

Member Function Documentation

Execute()

void Execute ( ) const

overridevirtual

Implements IWorkload.

Definition at line 78 of file NeonMultiplicationWorkload.cpp.

{
    ARMNN_SCOPED_PROFILING_EVENT_NEON_NAME_GUID("NeonMultiplicationWorkload_Execute");
    m_PixelWiseMultiplication->run();
}

References ARMNN_SCOPED_PROFILING_EVENT_NEON_NAME_GUID.

---

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

• src/backends/neon/workloads/NeonMultiplicationWorkload.hpp
• src/backends/neon/workloads/NeonMultiplicationWorkload.cpp