Atlas 300 Application (Model 3000)
Atlas 300 Application (Model 3000)
All results for "
" in this service
All results for "
" in this service
Development Environment Deployment Guides
DDK Installation Guide (Ubuntu, ARM)
Introduction
Environment Preparation
Preparing the Server Environment
Setting a Common User
Setting the Source
Installing Dependency
Preparing Software Packages
Verifying the Software Package Integrity
Installing the DDK
Installation Procedure
Common Operations
Uninstalling the DDK
Querying the DDK Version
FAQs
What Do I Do If a Message Is Displayed Indicating pip2 or pip Unavailability During Mind Studio or DDK Installation?
Change History
DDK Installation Guide (EulerOS)
Introduction
Obtaining Software Packages
Verifying the Software Package Integrity
Environment Preparation
Installing the DDK
Installation
Common Operations
Uninstalling the DDK
Querying the DDK Version
FAQs
What Do I Do If the Message "Warning: DDK functions except TE have been successfully installed." Is Displayed During DDK Installation?
Overview of Software Packages
Open Source Third-Party Libraries
Change History
DDK Installation Guide (CentOS, ARM)
Introduction
Obtaining Software Packages
Verifying the Software Package Integrity
Environment Preparation
Installing the DDK
Installation
Common Operations
Uninstalling the DDK
Querying the DDK Version
Configuring the OpenPGP Public Key
FAQs
What Do I Do If the Message "Import Error:cannot import name main" Is Displayed When the pip3/pip3.5 -V Command Is Run after pip3/pip3.5 Upgrade?
What Do I Do If epel-release Installation Fails?
What Do I Do If the Message "Warning: DDK functions except TE have been successfully installed." Is Displayed During DDK Installation?
Overview of Software Packages
Open Source Third-Party Libraries
Change History
Model Conversion Guide
Introduction
Caffe/TensorFlow Mode Conversion
Constraints and Parameters
Model Conversion Using OMG
AIPP Configuration
Overview
Configuration File Template
CSC Configuration
Crop/Padding Configuration
AIPP Configurations According to Model Inputs
AIPP Verification of the Model Input Size
Parameter Structure for Dynamic AIPP
Quantization Configuration
Overview
Configuration File Template
Description of the Configuration Parameters
Parameter Tuning
FAQs
What Do I Do If No Log Is Output to the Screen When the DDK Installation User Is HwHiAiUser?
What Do I Do If Model Conversion Takes Too Long When the OS and Architecture Configuration of the DDK Server is Arm (aarch64)?
What Do I Do If the Error "Unrecognized layer:xxx, layer type xxx" Is Reported During Model Conversion?
What Do I Do If the Error "It is recommended to convert layers-structure to layer-structure by caffe tool" Is Reported During Model Conversion?
Appendix
Change History
Tuning Guide
Overview
Key Points
Memory Management
Memory Management APIs Provided by the Native Language
Memory Management APIs Provided by the Matrix Module
Host-Device Data Transmission
Usage of DVPP
Image and Video Encoding and Decoding
Image Cropping and Resizing
Model Conversion Pre-Processing Configuration
Batch and Timeout
Processing the Input and Output Data of Algorithm Inference
Optimization of Data Backhaul
Operator Usage Suggestions
Samples
Data Flow
Zero-Copy
FAQs
What Do I Do If an Exception Occurs Because a Smart Pointer Is Used to Automatically Free Memory but the Destructor Is Not Empty?
What Do I Do If an Exception Occurs Because a Smart Pointer Is Used to Manually Free Memory but the Destructor Is Not HIAI_DFree?
Appendix
Change History
API Reference
Matrix API Reference
Overview
Process Orchestration APIs
Process Connection APIs (C Language)
HIAI_Init (Common with C++)
HIAI_CreateGraph
HIAI_DestroyGraph
Process Connection APIs (C++ Language)
Graph::GetInstance
Graph::CreateGraph (Creating a Graph Based on the Configuration File)
Graph::CreateGraph (Creating a Graph Based on the Configuration File and Writing the Generated Graph Back to the List)
Graph::CreateGraph (Creating a Graph Based on the Protobuf Data Format)
Graph::DestroyGraph
Graph::UpdateEngineConfig
Data Sending APIs (C Language)
HIAI_SetDataRecvFunctor
HIAI_C_SendData
Data Sending APIs (C++ Language)
Graph::SetDataRecvFunctor
Graph::SendData
HIAI_SendData
API Usage Example
Example of Using the C APIs
Example of Using C++ APIs (Creating a Graph Based on the Configuration File)
Usage Example for C++ APIs (Creating a Graph Based on the Configuration File and Writing the Generated Graph to the List)
Engine Implementation APIs (C++ Language)
Constructor and Destructor Functions of an Engine
Engine::Init
Macro: HIAI_DEFINE_PROCESS
Macro: HIAI_IMPL_ENGINE_PROCESS
Engine::SetDataRecvFunctor
Engine::SendData
Calling Example
Model Manager APIs (C++ Language)
Offline Model Manager
AIModelManager::Init
AIModelManager::SetListener
AIModelManager::Process
AIModelManager::CreateOutputTensor
AIModelManager::CreateInputTensor
AIModelManager::IsPreAllocateOutputMem
AIModelManager::GetModelIOTensorDim
AIModelManager::GetMaxUsedMemory
AISimpleTensor::SetBuffer
AITensorFactory::CreateTensor
Calling Example
AIPP Configuration APIs
Overview
SetDynamicInputIndex
SetDynamicInputEdgeIndex
SetInputFormat
SetCscParams (Setting the Default Parameter Value)
SetCscParams (Setting a Parameter Value as Required)
SetRbuvSwapSwitch
SetAxSwapSwitch
SetSrcImageSize
SetCropParams
SetPaddingParams
SetDtcPixelMean
SetDtcPixelMin
SetPixelVarReci
SetInputDynamicAIPP
GetDynamicInputIndex
GetDynamicInputEdgeIndex
Calling Example
Data Types
AIConfigItem
AIConfig
AITensorParaDescription
AITensorDescription
AITensorDescriptionList
AIModelDescription
AINNNodeDescription
AINNNodeDescriptionList
AIAPIDescription
AIAPIDescriptionList
AIOPDescription
AIOPDescriptionList
NodeDesc
EngineDesc
GraphInitDesc
GeneralFileBuffer
AIContext
TensorDimension
IAIListener
IAITensor
Other Compilation Dependent APIs
AIAlgAPIFactory
AIAlgAPIRegisterar
REGISTER_ALG_API_UNIQUE
REGISTER_ALG_API
AIModelManager
getModelInfo
IAINNNode
AINNNodeFactory
AINNNodeRegisterar
REGISTER_NN_NODE
AITensorGetBytes
AITensorFactory
REGISTER_TENSOR_CREATER_UNIQUE
REGISTER_TENSOR_CREATER
AISimpleTensor
AINeuralNetworkBuffer
AIImageBuffer
HIAILog
HIAI_ENGINE_LOG
Exception Handling
Auxiliary APIs
Data Receiving APIs (C++ Language)
Obtaining the Number of Devices
Obtaining the ID of the First Device
Obtaining the ID of the First Graph
Obtaining the ID of the Next Device
Obtaining the ID of the Next Graph
Obtaining the Engine Pointer
Obtaining the Graph ID
Obtaining the Device ID of a Graph
Obtaining the Graph ID of an Engine
Obtaining the Maximum Queue Size of an Engine
Obtaining the Current Queue Size of a Specified Port of the Engine
Parsing the Matrix Configuration File
Obtaining the PCIe Information
Obtaining the Version Number
Obtaining the OamConfig Smart Pointer
Data Type Serialization and Deserialization (C++ Language)
Macro: HIAI_REGISTER_DATA_TYPE
Macro: HIAI_REGISTER_TEMPLATE_DATA_TYPE
Macro: HIAI_REGISTER_SERIALIZE_FUNC
Graph::ReleaseDataBuffer
API Usage Example
Memory Management (C Language)
HIAI_DMalloc
HIAI_DFree
HIAI_DVPP_DMalloc
HIAI_DVPP_DFree
Memory Management (C++ Language)
HIAIMemory::HIAI_DMalloc
HIAIMemory::HIAI_DFree
HIAIMemory:: IsDMalloc
HIAIMemory::HIAI_DVPP_DMalloc
HIAIMemory::HIAI_DVPP_DFree
Logs (C++)
Error Code Registration
HIAI_DEF_ERROR_CODE
API Usage Example
Log Printing
Log Printing Format 1
Log Printing Format 2
Log Printing Format 3
Log Printing Format 4
Log Printing Format 5
Log Printing Format 6
Log Printing Format 7
Log Printing Format 8
MultiTypeQueue API for Queue Management (C++ Language)
MultiTypeQueue Constructor
PushData
FrontData
PopData
PopAllData
API Usage Example
Event Registration API (C++ Language)
Graph::RegisterEventHandle
Others (C++ Language)
DataRecvInterface::RecvData
Graph::SetPublicKeyForSignatureEncryption
Python APIs
Appendix
APIs to Be Deleted in Later Versions
Low-Power APIs (C++ Language)
PowerState::SubscribePowerStateEvent
PowerState::UnsubscribePowerStateEvent
Data Types
Enumeration: PowerState::DEVICE_POWER_STATE
PowerState::PowerStateNotifyCallbackT
Matrix Data Types
Protobuf Data Types
Matrix Custom Data Types
Data Structures Registered in the Matrix
Example
Orchestration Configuration Example
Example of Data Transmission Based on Optimized Performance
Change History
DVPP API Reference
Overview
API Introduction
API List
VPC Function
JPEGE Function
JPEGD Function
PNGD Function
VDEC Function
VENC Function
Input and Output Memory
VPC/JPEGE/JPEGD/PNGD Function Interfaces
CreateDvppApi
DvppCtl
API Description
VPC Parameters
JPEGE Parameters
JPEGD Parameters
PNGD Parameters
Parameters for Querying the DVPP Engine
DestroyDvppApi
DvppGetOutParameter
VDEC Function Interfaces
General Description
CreateVdecApi
VdecCtl
DestroyVdecApi
VENC Function Interfaces
General Description
CreateVenc
RunVenc
DestroyVenc
Compatibility with the Functions of Earlier Versions
Compatibility
VPC Functions and Parameters
CMDLIST Functions and Parameters
VENC Functions and Parameters
Calling Example
Implementing the VPC Function
Implementing the JPEGE Function
Implementing the JPEGD Function
Implementing the PNGD Function
Implementing the VDEC Function
Implementing the VENC Function
Data Types
Structures in VpcUserImageConfigure
Structures and Classes in vdec_in_msg
Structures in IMAGE_CONFIG
Structures in dvpp_engine_capability_stru
Structures in vpc_in_msg
RDMACHANNEL Structure
VpcTurningReverse Structure
Appendix
Auxiliary Function APIs
List of APIs that Are Not Recommended
Sample Usage of the DVPP Executor
Environment Preparation
Sample Input Parameters of the DVPP Executor
VPC Instructions
VPC Basic Function 1
VPC Basic Function 2
VDEC Usage
Usage of the VENC
Usage of the JPEGE
Usage of the JPEGD
Usage of PNGD
JPEGD+VPC+JPEGE Cascading
Exception Handling
Acronyms
Change History
Framework API Reference
Introduction
Non-Class Member Functions
custom_op_run
custom_op_compare
GetContext
AutoMappingFn
aicpu_run_func
Class Member Functions
Class StatusFactory
Instance
RegisterErrorNo
GetErrDesc
StatusFactory
~StatusFactory
Class ErrorNoRegisterar
ErrorNoRegisterar
~ErrorNoRegisterar
Class OpTypeContainer
Instance
Register
IsExisting
OpTypeContainer
Class OpTypeRegistrar
OpTypeRegistrar
~OpTypeRegistrar
Class OpRegistrationData
OpRegistrationData
~OpRegistrationData
FrameworkType
OriginOpType
ParseParamsFn
InferShapeAndTypeFn
UpdateOpDescFn
GetWorkspaceSizeFn
TEBinBuildFn
ImplyType
Formats
WeightFormats
Finalize
Class OpRegistry
Instance
Register
GetImplyType
GetOpTypeByImplyType
GetFormats
GetWeightFormats
GetParseParamFunc
GetInferShapeFunc
GetGetWorkspaceSizeFunc
GetUpdateOpDescFunc
GetBuildTeBinFunc
GetTransWeightFunc
Class OpReceiver
OpReceiver
~OpReceiver
Macro Definitions
Macro for Generating Error Codes
Macro for Generating Error Codes of the Common Module
Macro for Generating Error Codes of the OMG Module
Macro for Generating Error Codes of the OME Module
Macro for Generating Error Codes of the CALIBRATION Module
Macro for Obtaining Error Code Description.
Macro for Registering the Operator Type
Macro Indicating Whether an Operator Type Exists
Macro for Registering an Operator
REGISTER_CUSTOM_OP
Macros for Marking External APIs
FMK_FUNC_HOST_VISIBILITY
FMK_FUNC_DEV_VISIBILITY
Appendix
Change History
GE API Reference
Introduction
Attribute APIs
SetAttr
GetAttr
HasAttr
DelAttr
SetName
GetName
GetItem
GetValueType
IsEmpty
Copy
MutableTensor
MutableListTensor
InitDefault
Buffer APIs
ClearBuffer
GetData
GetSize
CopyFrom
Graph APIs
SetInputs
SetOutputs
IsValid
AddOp
FindOpByName
CheckOpByName
GetAllOpName
Model APIs
SetName
GetName
SetVersion
GetVersion
SetPlatformVersion
GetPlatformVersion
GetGraph
SetGraph
Save
Load
IsValid
Operator APIs
Operator Constructor
GetName
SetInput (srcOprt Has Only One Output)
SetInput (srcOprt Has Multiple Outputs)
GetInputDesc
TryGetInputDesc
UpdateInputDesc
GetOutputDesc
UpdateOutputDesc
GetDynamicInputDesc
UpdateDynamicInputDesc
GetDynamicOutputDesc
UpdateDynamicOutputDesc
SetAttr
GetAttr
Shape APIs
GetDimNum
GetDim
GetDims
SetDim
GetShapeSize
Tensor APIs
GetTensorDesc
MutableTensorDesc
SetTensorDesc
GetData
MutableData
SetData
Clone
TensorDesc APIs
Update
GetShape
MutableShape
SetShape
GetFormat
SetFormat
GetDataType
SetDataType
Clone
IsValid
Operator Registration APIs
REG_OP
ATTR
REQUIRED_ATTR
INPUT
OPTIONAL_INPUT
DYNAMIC_INPUT
OUTPUT
DYNAMIC_OUTPUT
INFER_SHAPE_AND_TYPE
ATTR_ALL_VERIFY
OP_END
List of Built-in Operators
Model Building APIs
CreateModelBuff
BuildIRModel
ReleaseModelBuff
Appendix
Change History
TE API Reference
Introduction
Description
Compute APIs
te.lang.cce.vadd(lhs, rhs)
te.lang.cce.vsub(lhs, rhs)
te.lang.cce.vmul(lhs, rhs)
te.lang.cce.vmin(lhs, rhs)
te.lang.cce.vmax(lhs, rhs)
te.lang.cce.vor(lhs, rhs)
te.lang.cce.vand(lhs, rhs)
te.lang.cce.vadds(raw_tensor, scalar)
te.lang.cce.vmuls(raw_tensor, scalar)
te.lang.cce.vlog(raw_tensor)
te.lang.cce.vexp(raw_tensor)
te.lang.cce.vabs(raw_tensor)
te.lang.cce.vrec(raw_tensor)
te.lang.cce.cast_to(data, dtype, f1628IntegerFlag=False)
te.lang.cce.vrelu(raw_tensor)
te.lang.cce.vnot(raw_tensor)
te.lang.cce.vaxpy(lhs, rhs, scalar)
te.lang.cce.vmla(x, y, z)
te.lang.cce.vmadd(x, y, z)
te.lang.cce.vmaddrelu(x, y, z)
te.lang.cce.ceil(raw_tensor)
te.lang.cce.floor(raw_tensor)
te.lang.cce.round(raw_tensor)
te.lang.cce.sum(raw_tensor, axis, keepdims=False)
te.lang.cce.reduce_min(raw_tensor, axis, keepdims=False)
te.lang.cce.reduce_max(raw_tensor, axis, keepdims=False)
te.lang.cce.reduce_prod(raw_tensor, axis, keepdims=False)
te.lang.cce.broadcast(var, shape, output_dtype=None)
te.lang.cce.unsorted_segment_sum(tensor, segment_ids, num_segments, init_value=0)
te.lang.cce.unsorted_segment_mean(tensor, segment_ids, num_segments, init_value=0)
te.lang.cce.unsorted_segment_prod(tensor, segment_ids, num_segments, init_value=0)
te.lang.cce.unsorted_segment_min(tensor, segment_ids, num_segments, init_value=0)
te.lang.cce.unsorted_segment_max(tensor, segment_ids, num_segments, init_value=0)
te.lang.cce.concat(raw_tensors, axis)
te.lang.cce.conv(*args)
te.lang.cce.compute_four2five(input, raw_shape_4D)
te.lang.cce.compute_five2four(input, raw_shape_4D)
te.lang.cce.matmul(tensor_a, tensor_b, trans_a=False, trans_b=False, alpha_num=1.0, beta_num=0.0, tensor_c=None)
Build APIs
te.lang.cce.auto_schedule(outs)
te.lang.cce.cce_build_code(sch, config_map = {})
Convergence API
bool BuildTeCustomOp(std::string ddkVer, std::string opName, std::string opPath, std::string opFuncName, const char *format, ...)
APIs that Compilation Depends on
Instructions
Usage Example
Exception Handling
Appendix
Change History
Operator Development Guides
Operator List
Parameter Description
Operator Boundaries
General Restrictions
Caffe Operator Boundaries
TensorFlow Operator Boundaries
Appendix
Change History
TE Custom Operator Development Guide
Before You Start
Overview
Intended Audience
Precautions for Operator Development
Environment Preparation
Overall Development Process
Setting Environment Variables
Creating a Custom Operator Development Project
Developing a Custom Operator
Operator Basics
Implementing an Operator
Procedure
Importing Python Modules
Implementing an Operator
Scheduling and Building an Operator
Running an Operator
Code Examples
Operator Running Verification
Compiling an Operator
Building an Input Data File
Running a Single Operator
caffe.proto File Operator Definition (Optional)
Developing the Plug-In of a Custom Operator
Implementing a Plug-In
Implementing the Plug-In
Including Header Files
Parsing an Operator
Inferring the Output Tensor Description of an Operator
Building an Operator
Registering an Operator
Plug-In Code Example
Compiling the Operator Plug-In
Loading the Plug-In for Model Conversion
Function Reference for Operator Plug-In Development
ParseParamsFn
InferShapeAndTypeFn
TEBinBuildFn
WeightFormats
Registration Macro of the Operator Building Function
Appendix
Change History
IDE-daemon-client Command Reference
Introduction
Sending a Command to the Device by Using IDE Daemon
Sending a Command to the Device by Using IDE Daemon
Transferring a File to the Device by Using IDE Daemon
Transferring a File to the Host by Using IDE Daemon
Obtaining a File from the Host by Using IDE Daemon
Detecting the Heartbeat Between IDE Daemon and the Host
Setting Log Levels
Sending Profiling Items to the Host by Using IDE Daemon
Collecting Dump Information Recorded by the Black Box on the Host
Obtaining the Information About the Device Connected to the Host
Generating a Character String Encryption Files
Synchronizing the Time Between IDE Daemon and the Host
Viewing the Help Information
Appendix
Change History
General Reference
Glossary
Service Level Agreement
White Papers
Endpoints
Permissions