ConvertFp32ToFp16QueueDescriptor Struct Reference

#include <WorkloadData.hpp>

Inheritance diagram for ConvertFp32ToFp16QueueDescriptor:

Public Member Functions
void	Validate (const WorkloadInfo &workloadInfo) const
Public Member Functions inherited from QueueDescriptor
virtual	~QueueDescriptor ()=default
void	ValidateTensorNumDimensions (const TensorInfo &tensor, std::string const &descName, unsigned int numDimensions, std::string const &tensorName) const
void	ValidateTensorNumDimNumElem (const TensorInfo &tensorInfo, unsigned int numDimension, unsigned int numElements, std::string const &tensorName) const
void	ValidateInputsOutputs (const std::string &descName, unsigned int numExpectedIn, unsigned int numExpectedOut) const
template<typename T >
const T *	GetAdditionalInformation () const

Additional Inherited Members
Public Attributes inherited from QueueDescriptor
std::vector< ITensorHandle * >	m_Inputs
std::vector< ITensorHandle * >	m_Outputs
void *	m_AdditionalInfoObject
bool	m_AllowExpandedDims = false
Protected Member Functions inherited from QueueDescriptor
	QueueDescriptor ()
	QueueDescriptor (QueueDescriptor const &)=default
QueueDescriptor &	operator= (QueueDescriptor const &)=default

Detailed Description

Definition at line 499 of file WorkloadData.hpp.

Member Function Documentation

Validate()

void Validate

(

const WorkloadInfo &

workloadInfo

)

const

Definition at line 2318 of file WorkloadData.cpp.

References armnn::Float16, armnn::Float32, TensorInfo::GetDataType(), WorkloadInfo::m_InputTensorInfos, and WorkloadInfo::m_OutputTensorInfos.

{
    const std::string descriptorName{"ConvertFp32ToFp16QueueDescriptor"};

    ValidateNumInputs(workloadInfo,  descriptorName, 1);
    ValidateNumOutputs(workloadInfo, descriptorName, 1);

    const TensorInfo& inputTensorInfo  = workloadInfo.m_InputTensorInfos[0];
    const TensorInfo& outputTensorInfo = workloadInfo.m_OutputTensorInfos[0];

    if (inputTensorInfo.GetDataType() != DataType::Float32)
    {
        throw InvalidArgumentException(descriptorName + ": Input tensor type must be Float32.");
    }

    if (outputTensorInfo.GetDataType() != DataType::Float16)
    {
        throw InvalidArgumentException(descriptorName + ": Output tensor type must be Float16.");
    }

    ValidateTensorShapesMatch(inputTensorInfo, outputTensorInfo, descriptorName, "input", "output");
}

---

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

• include/armnn/backends/WorkloadData.hpp
• src/backends/backendsCommon/WorkloadData.cpp