ArmNN
 22.05.01
Public Member Functions | List of all members
NeonConvertFp32ToBf16Workload Class Reference

#include <NeonConvertFp32ToBf16Workload.hpp>

Inheritance diagram for NeonConvertFp32ToBf16Workload:
MultiTypedWorkload< QueueDescriptor, InputDataType, OutputDataType > BaseWorkload< QueueDescriptor > IWorkload

Public Member Functions

 NeonConvertFp32ToBf16Workload (const ConvertFp32ToBf16QueueDescriptor &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 (WorkingMemDescriptor &workingMemDescriptor) 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 &)
 

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 NeonConvertFp32ToBf16Workload.hpp.

Constructor & Destructor Documentation

◆ NeonConvertFp32ToBf16Workload()

NeonConvertFp32ToBf16Workload ( const ConvertFp32ToBf16QueueDescriptor &  descriptor,
const WorkloadInfo &  info 
)

Definition at line 18 of file NeonConvertFp32ToBf16Workload.cpp.

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

20  : Float32ToBFloat16Workload<ConvertFp32ToBf16QueueDescriptor>(descriptor, info)
21 {
22  this->m_Data.ValidateInputsOutputs("NeonConvertFp32ToBf16Workload", 1, 1);
23  GatherTensorHandlePairs(descriptor, m_TensorHandlePairs);
24 }
armnn::QueueDescriptor::ValidateInputsOutputs
void ValidateInputsOutputs(const std::string &descName, unsigned int numExpectedIn, unsigned int numExpectedOut) const
Definition: WorkloadData.cpp:474
armnn::BaseWorkload::m_Data
QueueDescriptor m_Data
Definition: Workload.hpp:81
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 26 of file NeonConvertFp32ToBf16Workload.cpp.

References ARMNN_SCOPED_PROFILING_EVENT_NEON_GUID, FloatingPointConverter::ConvertFloat32ToBFloat16(), armnn::CopyTensorContentsGeneric(), and BaseWorkload< QueueDescriptor >::GetGuid().

27 {
28  ARMNN_SCOPED_PROFILING_EVENT_NEON_GUID("NeonConvertFp32ToBf16Workload_Execute", this->GetGuid());
29 
30  auto convertFunc = [](uint8_t* dst, const uint8_t* src, size_t size)
31  {
32  auto input = reinterpret_cast<const float*>(src);
33  auto output = reinterpret_cast<BFloat16*>(dst);
34  size_t numElements = size/2; // 2 bytes per bf16
35  armnnUtils::FloatingPointConverter::ConvertFloat32ToBFloat16(input, numElements, output);
36  };
37 
38  for (const auto& pair : m_TensorHandlePairs)
39  {
40  CopyTensorContentsGeneric(pair.first, pair.second, convertFunc);
41  }
42 }
armnn::BaseWorkload::GetGuid
arm::pipe::ProfilingGuid GetGuid() const final
Definition: Workload.hpp:59
armnnUtils::FloatingPointConverter::ConvertFloat32ToBFloat16
static void ConvertFloat32ToBFloat16(const float *srcFloat32Buffer, size_t numElements, void *dstBFloat16Buffer)
Definition: FloatingPointConverter.cpp:46
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 44 of file NeonConvertFp32ToBf16Workload.cpp.

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

45 {
46  ITensorHandle* backupHandle = this->m_Data.m_Inputs[slot];
47  this->m_Data.m_Inputs[slot] = tensorHandle;
48  try
49  {
50  Reconfigure();
51  }
52  catch(armnn::UnimplementedException& e)
53  {
54  // Cannot reconfigure, revert the slot back and throw the exception.
55  this->m_Data.m_Inputs[slot] = backupHandle;
56  throw e;
57  }
58 }
armnn::BaseWorkload::m_Data
QueueDescriptor m_Data
Definition: Workload.hpp:81
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 61 of file NeonConvertFp32ToBf16Workload.cpp.

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

62 {
63  ITensorHandle* backupHandle = this->m_Data.m_Inputs[slot];
64  this->m_Data.m_Inputs[slot] = tensorHandle;
65  try
66  {
67  Reconfigure();
68  }
69  catch(armnn::UnimplementedException& e)
70  {
71  // Cannot reconfigure, revert the slot back and throw the exception.
72  this->m_Data.m_Inputs[slot] = backupHandle;
73  throw e;
74  }
75 }
armnn::BaseWorkload::m_Data
QueueDescriptor m_Data
Definition: Workload.hpp:81
armnn::QueueDescriptor::m_Inputs
std::vector< ITensorHandle * > m_Inputs
Definition: WorkloadData.hpp:26
The documentation for this class was generated from the following files: