23 #include <boost/polymorphic_cast.hpp> 35 std::string& outReasonIfUnsupported)
46 : m_MemoryManager(memoryManager)
52 unsigned int const* subTensorOrigin)
const 54 const arm_compute::TensorShape shape = armcomputetensorutils::BuildArmComputeTensorShape(subTensorShape);
62 coords.set(i, boost::numeric_cast<int>(subTensorOrigin[revertedIndex]));
65 const arm_compute::TensorShape parentShape = armcomputetensorutils::BuildArmComputeTensorShape(parent.
GetShape());
66 if (!::arm_compute::error_on_invalid_subtensor(__func__, __FILE__, __LINE__, parentShape, coords, shape))
71 return std::make_unique<NeonSubTensorHandle>(
72 boost::polymorphic_downcast<IAclTensorHandle*>(&parent), shape, coords);
76 const bool IsMemoryManaged)
const 78 auto tensorHandle = std::make_unique<NeonTensorHandle>(tensorInfo);
81 tensorHandle->SetMemoryGroup(m_MemoryManager->GetInterLayerMemoryGroup());
88 const bool IsMemoryManaged)
const 90 auto tensorHandle = std::make_unique<NeonTensorHandle>(tensorInfo, dataLayout);
93 tensorHandle->SetMemoryGroup(m_MemoryManager->GetInterLayerMemoryGroup());
112 return std::make_unique<NeonActivationWorkload>(descriptor,
info);
118 return std::make_unique<NeonAdditionWorkload>(descriptor,
info);
124 return std::make_unique<NeonArgMinMaxWorkload>(descriptor,
info);
130 return std::make_unique<NeonBatchNormalizationWorkload>(descriptor,
info);
136 return std::make_unique<NeonBatchToSpaceNdWorkload>(descriptor,
info);
148 return MakeWorkloadHelper<NeonGreaterFloat32Workload, NeonGreaterUint8Workload>(greaterQueueDescriptor,
info);
150 return MakeWorkloadHelper<NullWorkload, NullWorkload>(descriptor,
info);
156 return std::make_unique<NeonConcatWorkload>(descriptor,
info);
162 return std::make_unique<NeonConstantWorkload>(descriptor,
info);
169 return std::make_unique<NeonConvertFp16ToFp32Workload>(descriptor,
info);
176 return std::make_unique<NeonConvertFp32ToFp16Workload>(descriptor,
info);
182 return std::make_unique<NeonConvolution2dWorkload>(descriptor,
info,
183 m_MemoryManager->GetIntraLayerManager());
189 return MakeWorkloadHelper<NullWorkload, NullWorkload>(descriptor,
info);
195 return std::make_unique<NeonDepthToSpaceWorkload>(descriptor,
info);
201 return std::make_unique<NeonDepthwiseConvolutionWorkload>(descriptor,
info);
207 return std::make_unique<NeonDequantizeWorkload>(descriptor,
info);
213 return MakeWorkloadHelper<NullWorkload, NullWorkload>(descriptor,
info);
219 return std::make_unique<NeonDivisionWorkload>(descriptor,
info);
232 return std::make_unique<NeonAbsWorkload>(absQueueDescriptor,
info);
240 return std::make_unique<NeonRsqrtWorkload>(rsqrtQueueDescriptor,
info);
242 return MakeWorkloadHelper<NullWorkload, NullWorkload>(descriptor,
info);
259 return MakeWorkloadHelper<NeonFloorFloatWorkload, NullWorkload>(descriptor,
info);
265 return MakeWorkloadHelper<NeonFullyConnectedWorkload, NeonFullyConnectedWorkload>(
266 descriptor,
info, m_MemoryManager->GetIntraLayerManager());
272 return MakeWorkloadHelper<NullWorkload, NullWorkload>(descriptor,
info);
289 return std::make_unique<CopyMemGenericWorkload>(descriptor,
info);
296 return std::make_unique<NeonInstanceNormalizationWorkload>(descriptor,
info);
302 return MakeWorkloadHelper<NeonL2NormalizationFloatWorkload, NullWorkload>(descriptor,
info,
303 m_MemoryManager->GetIntraLayerManager());
309 return MakeWorkloadHelper<NeonLstmFloatWorkload, NullWorkload>(descriptor,
info);
315 return std::make_unique<NeonMaximumWorkload>(descriptor,
info);
321 return std::make_unique<NeonMeanWorkload>(descriptor,
info);
332 return MakeWorkloadHelper<CopyMemGenericWorkload, CopyMemGenericWorkload>(descriptor,
info);
343 return std::make_unique<ImportMemGenericWorkload>(descriptor,
info);
355 return std::make_unique<NeonMinimumWorkload>(descriptor,
info);
361 return std::make_unique<NeonMultiplicationWorkload>(descriptor,
info);
367 return MakeWorkloadHelper<NeonNormalizationFloatWorkload, NullWorkload>(descriptor,
info,
368 m_MemoryManager->GetIntraLayerManager());
374 return std::make_unique<CopyMemGenericWorkload>(descriptor,
info);
380 return std::make_unique<NeonPadWorkload>(descriptor,
info);
386 return std::make_unique<NeonPermuteWorkload>(descriptor,
info);
392 return std::make_unique<NeonPooling2dWorkload>(descriptor,
info);
398 return MakeWorkloadHelper<NullWorkload, NullWorkload>(descriptor,
info);
404 return std::make_unique<NeonPreluWorkload>(descriptor,
info);
410 return std::make_unique<NeonQuantizeWorkload>(descriptor,
info);
416 return std::make_unique<NeonQuantizedLstmWorkload>(descriptor,
info);
422 return std::make_unique<NeonReshapeWorkload>(descriptor,
info);
428 return std::make_unique<NeonResizeWorkload>(descriptor,
info);
460 return std::make_unique<NeonSliceWorkload>(descriptor,
info);
466 return MakeWorkloadHelper<NeonSoftmaxFloatWorkload, NeonSoftmaxUint8Workload>(
467 descriptor,
info, m_MemoryManager->GetIntraLayerManager());
473 return std::make_unique<NeonSpaceToBatchNdWorkload>(descriptor,
info);
479 return std::make_unique<NeonSpaceToDepthWorkload>(descriptor,
info);
485 return std::make_unique<NeonSplitterWorkload>(descriptor,
info);
491 return std::make_unique<NeonStackWorkload>(descriptor,
info);
497 return std::make_unique<NeonStridedSliceWorkload>(descriptor,
info);
503 return std::make_unique<NeonSubtractionWorkload>(descriptor,
info);
509 return std::make_unique<NeonTransposeWorkload>(descriptor,
info);
516 return std::make_unique<NeonTransposeConvolution2dWorkload>(descriptor,
info,
517 m_MemoryManager->GetIntraLayerManager());
std::unique_ptr< IWorkload > CreateDepthToSpace(const DepthToSpaceQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateDebug(const DebugQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateLstm(const LstmQueueDescriptor &descriptor, const WorkloadInfo &info) const override
UnaryOperation m_Operation
Specifies the elementwiseUnary operation to execute.
std::unique_ptr< IWorkload > CreateGather(const GatherQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateNormalization(const NormalizationQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateFullyConnected(const FullyConnectedQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateComparison(const ComparisonQueueDescriptor &descriptor, const WorkloadInfo &Info) const override
std::unique_ptr< IWorkload > CreateRsqrt(const RsqrtQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreatePad(const PadQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateTransposeConvolution2d(const TransposeConvolution2dQueueDescriptor &descriptor, const WorkloadInfo &info) const override
A ComparisonDescriptor for the ComparisonLayer.
uint32_t m_TargetWidth
Target width value.
DataLayout m_DataLayout
The data layout to be used (NCHW, NHWC).
std::unique_ptr< IWorkload > CreateGreater(const GreaterQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateMemImport(const MemImportQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::array< unsigned int, MaxNumOfTensorDimensions > Coordinates
constexpr const char * NeonBackendId()
std::unique_ptr< IWorkload > CreateL2Normalization(const L2NormalizationQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateConvertFp16ToFp32(const ConvertFp16ToFp32QueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateStridedSlice(const StridedSliceQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateMinimum(const MinimumQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateMultiplication(const MultiplicationQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateQuantize(const QuantizeQueueDescriptor &descriptor, const WorkloadInfo &info) const override
Copyright (c) 2020 ARM Limited.
void IgnoreUnused(Ts &&...)
std::unique_ptr< ITensorHandle > CreateSubTensorHandle(ITensorHandle &parent, TensorShape const &subTensorShape, unsigned int const *subTensorOrigin) const override
std::unique_ptr< IWorkload > CreateConvolution2d(const Convolution2dQueueDescriptor &descriptor, const WorkloadInfo &info) const override
LayerDescriptor m_Parameters
NeonWorkloadFactory(const std::shared_ptr< NeonMemoryManager > &memoryManager)
std::unique_ptr< IWorkload > CreateReshape(const ReshapeQueueDescriptor &descriptor, const WorkloadInfo &info) const override
static bool IsLayerSupported(const Layer &layer, Optional< DataType > dataType, std::string &outReasonIfUnsupported)
std::unique_ptr< IWorkload > CreateAbs(const AbsQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateConvertFp32ToFp16(const ConvertFp32ToFp16QueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateSoftmax(const SoftmaxQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateResizeBilinear(const ResizeBilinearQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateEqual(const EqualQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateSlice(const SliceQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateQuantizedLstm(const QuantizedLstmQueueDescriptor &descriptor, const WorkloadInfo &info) const override
uint32_t m_TargetWidth
Target width value.
const BackendId & GetBackendId() const override
std::unique_ptr< IWorkload > CreateSplitter(const SplitterQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateDequantize(const DequantizeQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateMemCopy(const MemCopyQueueDescriptor &descriptor, const WorkloadInfo &info) const override
static bool IsLayerSupported(const BackendId &backendId, const IConnectableLayer &layer, Optional< DataType > dataType, std::string &outReasonIfUnsupported)
virtual TensorShape GetShape() const =0
Get the number of elements for each dimension ordered from slowest iterating dimension to fastest ite...
std::unique_ptr< IWorkload > CreateInstanceNormalization(const InstanceNormalizationQueueDescriptor &descriptor, const WorkloadInfo &info) const override
uint32_t m_TargetHeight
Target height value.
uint32_t m_TargetHeight
Target height value.
std::unique_ptr< IWorkload > CreatePreCompiled(const PreCompiledQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateOutput(const OutputQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateConstant(const ConstantQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateStack(const StackQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateFloor(const FloorQueueDescriptor &descriptor, const WorkloadInfo &info) const override
ComparisonOperation m_Operation
Specifies the comparison operation to execute.
std::unique_ptr< IWorkload > CreateActivation(const ActivationQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateDepthwiseConvolution2d(const DepthwiseConvolution2dQueueDescriptor &descriptor, const WorkloadInfo &info) const override
A ElementwiseUnaryDescriptor for the ElementwiseUnaryLayer.
std::vector< ITensorHandle * > m_Outputs
std::unique_ptr< IWorkload > CreateMerger(const MergerQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateDivision(const DivisionQueueDescriptor &descriptor, const WorkloadInfo &info) const override
unsigned int GetNumDimensions() const
std::unique_ptr< IWorkload > CreateInput(const InputQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateBatchNormalization(const BatchNormalizationQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateElementwiseUnary(const ElementwiseUnaryQueueDescriptor &descriptor, const WorkloadInfo &Info) const override
Contains information about inputs and outputs to a layer.
std::unique_ptr< IWorkload > CreateResize(const ResizeQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::vector< ITensorHandle * > m_Inputs
std::unique_ptr< IWorkload > CreateConcat(const ConcatQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreatePooling2d(const Pooling2dQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateBatchToSpaceNd(const BatchToSpaceNdQueueDescriptor &descriptor, const WorkloadInfo &Info) const override
std::unique_ptr< IWorkload > CreateAddition(const AdditionQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateDetectionPostProcess(const DetectionPostProcessQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateSubtraction(const SubtractionQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateMaximum(const MaximumQueueDescriptor &descriptor, const WorkloadInfo &info) const override
DataLayout m_DataLayout
The data layout to be used (NCHW, NHWC).
std::unique_ptr< IWorkload > CreateMean(const MeanQueueDescriptor &descriptor, const WorkloadInfo &Info) const override
std::unique_ptr< IWorkload > CreateArgMinMax(const ArgMinMaxQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreatePrelu(const PreluQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateTranspose(const TransposeQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateSpaceToBatchNd(const SpaceToBatchNdQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreateSpaceToDepth(const SpaceToDepthQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< IWorkload > CreatePermute(const PermuteQueueDescriptor &descriptor, const WorkloadInfo &info) const override
std::unique_ptr< ITensorHandle > CreateTensorHandle(const TensorInfo &tensorInfo, const bool IsMemoryManaged=true) const override