ArmNN
 20.08
Public Member Functions | List of all members
NeonSoftmaxWorkload Class Reference

#include <NeonSoftmaxWorkload.hpp>

Inheritance diagram for NeonSoftmaxWorkload:
BaseWorkload< SoftmaxQueueDescriptor > IWorkload

Public Member Functions

 NeonSoftmaxWorkload (const SoftmaxQueueDescriptor &descriptor, const WorkloadInfo &info, std::shared_ptr< arm_compute::MemoryManagerOnDemand > &memoryManager)
 
virtual void Execute () const override
 
- Public Member Functions inherited from BaseWorkload< SoftmaxQueueDescriptor >
 BaseWorkload (const SoftmaxQueueDescriptor &descriptor, const WorkloadInfo &info)
 
void PostAllocationConfigure () override
 
const SoftmaxQueueDescriptor & 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< SoftmaxQueueDescriptor >
const SoftmaxQueueDescriptor m_Data
 
const profiling::ProfilingGuid m_Guid
 

Detailed Description

Definition at line 24 of file NeonSoftmaxWorkload.hpp.

Constructor & Destructor Documentation

◆ NeonSoftmaxWorkload()

NeonSoftmaxWorkload ( const SoftmaxQueueDescriptor &  descriptor,
const WorkloadInfo &  info,
std::shared_ptr< arm_compute::MemoryManagerOnDemand > &  memoryManager 
)

Definition at line 34 of file NeonSoftmaxWorkload.cpp.

References armnn::ComputeAclAxis(), armnn::ComputePositiveAxis(), SoftmaxDescriptor::m_Axis, SoftmaxDescriptor::m_Beta, BaseWorkload< SoftmaxQueueDescriptor >::m_Data, QueueDescriptor::m_Inputs, WorkloadInfo::m_InputTensorInfos, QueueDescriptor::m_Outputs, QueueDescriptorWithParameters< LayerDescriptor >::m_Parameters, and QueueDescriptor::ValidateInputsOutputs().

36  : BaseWorkload<SoftmaxQueueDescriptor>(descriptor, info)
37 {
38  m_Data.ValidateInputsOutputs("NeonSoftmaxWorkload", 1, 1);
39 
40  // The ArmCompute softmax layer uses 2D input/output tensors, so flatten the first three dimensions.
41  arm_compute::ITensor& input = PolymorphicDowncast<IAclTensorHandle*>(m_Data.m_Inputs[0])->GetTensor();
42  arm_compute::ITensor& output = PolymorphicDowncast<IAclTensorHandle*>(m_Data.m_Outputs[0])->GetTensor();
43 
44  auto layer = std::make_unique<arm_compute::NESoftmaxLayer>(memoryManager);
45  int aclAxis_int = ComputeAclAxis(m_Data.m_Parameters.m_Axis, info.m_InputTensorInfos[0]);
46  unsigned int aclAxis = ComputePositiveAxis(aclAxis_int, info.m_InputTensorInfos[0]);
47  layer->configure(&input, &output, m_Data.m_Parameters.m_Beta, static_cast<int>(aclAxis));
48  m_SoftmaxLayer.reset(layer.release());
49 }
armnn::ComputeAclAxis
int ComputeAclAxis(const int &armnnAxis, const armnn::TensorInfo &tensor)
Function to convert ArmNN axis (left to right) to ACL axis (right to left) ranging from [-rank...
Definition: ArmComputeUtils.hpp:193
armnn::SoftmaxDescriptor::m_Axis
int m_Axis
Scalar, defaulted to the last index (-1), specifying the dimension the activation will be performed o...
Definition: Descriptors.hpp:146
armnn::BaseWorkload< SoftmaxQueueDescriptor >::m_Data
const SoftmaxQueueDescriptor m_Data
Definition: Workload.hpp:46
armnn::SoftmaxDescriptor::m_Beta
float m_Beta
Exponentiation value.
Definition: Descriptors.hpp:144
armnn::QueueDescriptor::ValidateInputsOutputs
void ValidateInputsOutputs(const std::string &descName, unsigned int numExpectedIn, unsigned int numExpectedOut) const
Definition: WorkloadData.cpp:460
armnn::ComputePositiveAxis
unsigned int ComputePositiveAxis(const int &axis, const armnn::TensorInfo &tensor)
Function to convert axis to its positive equivalent value.
Definition: ArmComputeUtils.hpp:209
armnn::QueueDescriptor::m_Outputs
std::vector< ITensorHandle * > m_Outputs
Definition: WorkloadData.hpp:31
armnn::QueueDescriptor::m_Inputs
std::vector< ITensorHandle * > m_Inputs
Definition: WorkloadData.hpp:30

Member Function Documentation

◆ Execute()

void Execute ( ) const
overridevirtual

Implements IWorkload.

Definition at line 51 of file NeonSoftmaxWorkload.cpp.

References ARMNN_SCOPED_PROFILING_EVENT_NEON.

52 {
53  ARMNN_SCOPED_PROFILING_EVENT_NEON("NeonSoftmaxWorkload_Execute");
54  m_SoftmaxLayer->run();
55 }
ARMNN_SCOPED_PROFILING_EVENT_NEON
#define ARMNN_SCOPED_PROFILING_EVENT_NEON(name)
Definition: NeonWorkloadUtils.hpp:17
The documentation for this class was generated from the following files: