ArmNN
 22.11
Public Member Functions | List of all members
NeonConvertFp16ToFp32Workload Class Reference

#include <NeonConvertFp16ToFp32Workload.hpp>

Inheritance diagram for NeonConvertFp16ToFp32Workload:
MultiTypedWorkload< QueueDescriptor, InputDataType, OutputDataType > BaseWorkload< QueueDescriptor > IWorkload

Public Member Functions

 NeonConvertFp16ToFp32Workload (const ConvertFp16ToFp32QueueDescriptor &descriptor, const WorkloadInfo &info)
 
virtual void Execute () const override
 
void ReplaceInputTensorHandle (ITensorHandle *tensorHandle, unsigned int slot) override
 
void ReplaceOutputTensorHandle (ITensorHandle *tensorHandle, unsigned int slot) override
 
- Public Member Functions inherited from MultiTypedWorkload< QueueDescriptor, InputDataType, OutputDataType >
 MultiTypedWorkload (const QueueDescriptor &descriptor, const WorkloadInfo &info)
 
- Public Member Functions inherited from BaseWorkload< QueueDescriptor >
 BaseWorkload (const QueueDescriptor &descriptor, const WorkloadInfo &info)
 
void ExecuteAsync (ExecutionData &executionData) override
 
void PostAllocationConfigure () override
 
const QueueDescriptor & GetData () const
 
arm::pipe::ProfilingGuid GetGuid () const final
 
virtual bool SupportsTensorHandleReplacement () const override
 
- Public Member Functions inherited from IWorkload
virtual ~IWorkload ()
 
virtual void RegisterDebugCallback (const DebugCallbackFunction &)
 
virtual armnn::Optional< armnn::MemoryRequirements > GetMemoryRequirements ()
 

Additional Inherited Members

- Protected Attributes inherited from BaseWorkload< QueueDescriptor >
QueueDescriptor m_Data
 
const arm::pipe::ProfilingGuid m_Guid
 

Detailed Description

Definition at line 15 of file NeonConvertFp16ToFp32Workload.hpp.

Constructor & Destructor Documentation

◆ NeonConvertFp16ToFp32Workload()

NeonConvertFp16ToFp32Workload ( const ConvertFp16ToFp32QueueDescriptor &  descriptor,
const WorkloadInfo &  info 
)

Definition at line 17 of file NeonConvertFp16ToFp32Workload.cpp.

References armnn::GatherTensorHandlePairs(), BaseWorkload< QueueDescriptor >::m_Data, and QueueDescriptor::ValidateInputsOutputs().

19  : Float16ToFloat32Workload<ConvertFp16ToFp32QueueDescriptor>(descriptor, info)
20 {
21  this->m_Data.ValidateInputsOutputs("NeonConvertFp16ToFp32Workload", 1, 1);
22  GatherTensorHandlePairs(descriptor, m_TensorHandlePairs);
23 }
armnn::QueueDescriptor::ValidateInputsOutputs
void ValidateInputsOutputs(const std::string &descName, unsigned int numExpectedIn, unsigned int numExpectedOut) const
Definition: WorkloadData.cpp:475
armnn::BaseWorkload::m_Data
QueueDescriptor m_Data
Definition: Workload.hpp:83
armnn::GatherTensorHandlePairs
void GatherTensorHandlePairs(const DescriptorType &descriptor, std::vector< std::pair< SrcTensorHandleType *, DstTensorHandleType *>> &tensorHandlePairs)
Definition: WorkloadUtils.hpp:189

Member Function Documentation

◆ Execute()

void Execute ( ) const
overridevirtual

Implements IWorkload.

Definition at line 25 of file NeonConvertFp16ToFp32Workload.cpp.

References ARMNN_SCOPED_PROFILING_EVENT_NEON_GUID, FloatingPointConverter::ConvertFloat16To32(), armnn::CopyTensorContentsGeneric(), and BaseWorkload< QueueDescriptor >::GetGuid().

26 {
27  ARMNN_SCOPED_PROFILING_EVENT_NEON_GUID("NeonConvertFp16ToFp32Workload_Execute", this->GetGuid());
28 
29  auto convertFunc = [](uint8_t* dst, const uint8_t* src, size_t size)
30  {
31  auto input = reinterpret_cast<const Half*>(src);
32  auto output = reinterpret_cast<float*>(dst);
33  size_t numElements = size/2; // 2 bytes per fp16
34  armnnUtils::FloatingPointConverter::ConvertFloat16To32(input, numElements, output);
35  };
36 
37  for (const auto& pair : m_TensorHandlePairs)
38  {
39  CopyTensorContentsGeneric(pair.first, pair.second, convertFunc);
40  }
41 }
armnn::BaseWorkload::GetGuid
arm::pipe::ProfilingGuid GetGuid() const final
Definition: Workload.hpp:61
armnn::Half
half_float::half Half
Definition: Half.hpp:22
armnnUtils::FloatingPointConverter::ConvertFloat16To32
static void ConvertFloat16To32(const void *srcFloat16Buffer, size_t numElements, float *dstFloat32Buffer)
Definition: FloatingPointConverter.cpp:31
armnn::CopyTensorContentsGeneric
void CopyTensorContentsGeneric(const ITensorHandle *srcTensor, ITensorHandle *dstTensor, CopyFunc copy)
Definition: WorkloadUtils.hpp:46
ARMNN_SCOPED_PROFILING_EVENT_NEON_GUID
#define ARMNN_SCOPED_PROFILING_EVENT_NEON_GUID(name, guid)
Definition: NeonWorkloadUtils.hpp:24

◆ ReplaceInputTensorHandle()

void ReplaceInputTensorHandle ( ITensorHandle *  tensorHandle,
unsigned int  slot 
)
overridevirtual

Reimplemented from BaseWorkload< QueueDescriptor >.

Definition at line 43 of file NeonConvertFp16ToFp32Workload.cpp.

References BaseWorkload< QueueDescriptor >::m_Data, and QueueDescriptor::m_Inputs.

44 {
45  ITensorHandle* backupHandle = this->m_Data.m_Inputs[slot];
46  this->m_Data.m_Inputs[slot] = tensorHandle;
47  try
48  {
49  Reconfigure();
50  }
51  catch(armnn::UnimplementedException& e)
52  {
53  // Cannot reconfigure, revert the slot back and throw the exception.
54  this->m_Data.m_Inputs[slot] = backupHandle;
55  throw e;
56  }
57 }
armnn::BaseWorkload::m_Data
QueueDescriptor m_Data
Definition: Workload.hpp:83
armnn::QueueDescriptor::m_Inputs
std::vector< ITensorHandle * > m_Inputs
Definition: WorkloadData.hpp:26

◆ ReplaceOutputTensorHandle()

void ReplaceOutputTensorHandle ( ITensorHandle *  tensorHandle,
unsigned int  slot 
)
overridevirtual

Reimplemented from BaseWorkload< QueueDescriptor >.

Definition at line 60 of file NeonConvertFp16ToFp32Workload.cpp.

References BaseWorkload< QueueDescriptor >::m_Data, and QueueDescriptor::m_Inputs.

61 {
62  ITensorHandle* backupHandle = this->m_Data.m_Inputs[slot];
63  this->m_Data.m_Inputs[slot] = tensorHandle;
64  try
65  {
66  Reconfigure();
67  }
68  catch(armnn::UnimplementedException& e)
69  {
70  // Cannot reconfigure, revert the slot back and throw the exception.
71  this->m_Data.m_Inputs[slot] = backupHandle;
72  throw e;
73  }
74 }
armnn::BaseWorkload::m_Data
QueueDescriptor m_Data
Definition: Workload.hpp:83
armnn::QueueDescriptor::m_Inputs
std::vector< ITensorHandle * > m_Inputs
Definition: WorkloadData.hpp:26
The documentation for this class was generated from the following files: