NeuralNeworkRuntime

Overview

Provides APIs for accelerating the Neural Network Runtime (NNRt) model inference.

Since: 9

System capability: @Syscap SystemCapability.Ai.NeuralNetworkRuntime

Summary

File

Name	Description
neural_network_core.h	Defines APIs for the Neural Network Core module. The AI inference framework uses the native interfaces provided by Neural Network Core to build models and perform inference and computing on acceleration device.
neural_network_runtime.h	Defines APIs for NNRt. The AI inference framework uses the native APIs provided by the NNRt to construct and build models.
neural_network_runtime_type.h	Defines the structure and enums for NNRt.

Structs

Name	Description
OH_NN_UInt32Array	Used to store a 32-bit unsigned integer array.
OH_NN_QuantParam	Used to define the quantization information.
OH_NN_Tensor	Used to define the tensor structure.
OH_NN_Memory	Used to define the memory structure.

Types

Name	Description
OH_NNModel	Model handle.
OH_NNCompilation	Compiler handle.
OH_NNExecutor	Executor handle.
NN_QuantParam	Quantization parameter handle.
NN_TensorDesc	Tensor description handle.
NN_Tensor	Tensor handle.
(NN_OnRunDone) (void userData, OH_NN_ReturnCode errCode, void *outputTensor[], int32_t outputCount)	Handle of the callback processing function invoked when the asynchronous inference ends.
(NN_OnServiceDied) (void userData)	Handle of the callback processing function invoked when the device driver service terminates unexpectedly during asynchronous inference.
OH_NN_UInt32Array	Used to store a 32-bit unsigned integer array.
OH_NN_QuantParam	Used to define the quantization information.
OH_NN_Tensor	Used to define the tensor structure.
OH_NN_Memory	Used to define the memory structure.

Enums

Name	Description
OH_NN_PerformanceMode { OH_NN_PERFORMANCE_NONE = 0, OH_NN_PERFORMANCE_LOW = 1, OH_NN_PERFORMANCE_MEDIUM = 2, OH_NN_PERFORMANCE_HIGH = 3, OH_NN_PERFORMANCE_EXTREME = 4 }	Performance modes of the device.
OH_NN_Priority { OH_NN_PRIORITY_NONE = 0, OH_NN_PRIORITY_LOW = 1, OH_NN_PRIORITY_MEDIUM = 2, OH_NN_PRIORITY_HIGH = 3 }	Priorities of a model inference task.
OH_NN_ReturnCode { OH_NN_SUCCESS = 0, OH_NN_FAILED = 1, OH_NN_INVALID_PARAMETER = 2, OH_NN_MEMORY_ERROR = 3,OH_NN_OPERATION_FORBIDDEN = 4, OH_NN_NULL_PTR = 5, OH_NN_INVALID_FILE = 6, OH_NN_UNAVALIDABLE_DEVICE = 7, OH_NN_INVALID_PATH = 8, OH_NN_TIMEOUT = 9, OH_NN_UNSUPPORTED = 10, OH_NN_CONNECTION_EXCEPTION = 11,OH_NN_SAVE_CACHE_EXCEPTION = 12, OH_NN_DYNAMIC_SHAPE = 13, OH_NN_UNAVAILABLE_DEVICE = 14 }	Error codes for NNRt.
OH_NN_FuseType : int8_t { OH_NN_FUSED_NONE = 0, OH_NN_FUSED_RELU = 1, OH_NN_FUSED_RELU6 = 2 }	Activation function types in the fusion operator for NNRt.
OH_NN_Format { OH_NN_FORMAT_NONE = 0, OH_NN_FORMAT_NCHW = 1, OH_NN_FORMAT_NHWC = 2, OH_NN_FORMAT_ND = 3 }	Formats of tensor data.
OH_NN_DeviceType { OH_NN_OTHERS = 0, OH_NN_CPU = 1, OH_NN_GPU = 2, OH_NN_ACCELERATOR = 3 }	Device types supported by NNRt.
OH_NN_DataType { OH_NN_UNKNOWN = 0, OH_NN_BOOL = 1, OH_NN_INT8 = 2, OH_NN_INT16 = 3,OH_NN_INT32 = 4, OH_NN_INT64 = 5, OH_NN_UINT8 = 6, OH_NN_UINT16 = 7,OH_NN_UINT32 = 8, OH_NN_UINT64 = 9, OH_NN_FLOAT16 = 10, OH_NN_FLOAT32 = 11,OH_NN_FLOAT64 = 12 }	Data types supported by NNRt.
OH_NN_OperationType { OH_NN_OPS_ADD = 1, OH_NN_OPS_AVG_POOL = 2, OH_NN_OPS_BATCH_NORM = 3, OH_NN_OPS_BATCH_TO_SPACE_ND = 4, OH_NN_OPS_BIAS_ADD = 5, OH_NN_OPS_CAST = 6, OH_NN_OPS_CONCAT = 7, OH_NN_OPS_CONV2D = 8, OH_NN_OPS_CONV2D_TRANSPOSE = 9, OH_NN_OPS_DEPTHWISE_CONV2D_NATIVE = 10, OH_NN_OPS_DIV = 11, OH_NN_OPS_ELTWISE = 12, OH_NN_OPS_EXPAND_DIMS = 13, OH_NN_OPS_FILL = 14, OH_NN_OPS_FULL_CONNECTION = 15, OH_NN_OPS_GATHER = 16,OH_NN_OPS_HSWISH = 17, OH_NN_OPS_LESS_EQUAL = 18, OH_NN_OPS_MATMUL = 19, OH_NN_OPS_MAXIMUM = 20, OH_NN_OPS_MAX_POOL = 21, OH_NN_OPS_MUL = 22, OH_NN_OPS_ONE_HOT = 23, OH_NN_OPS_PAD = 24, OH_NN_OPS_POW = 25, OH_NN_OPS_SCALE = 26, OH_NN_OPS_SHAPE = 27, OH_NN_OPS_SIGMOID = 28, OH_NN_OPS_SLICE = 29, OH_NN_OPS_SOFTMAX = 30, OH_NN_OPS_SPACE_TO_BATCH_ND = 31, OH_NN_OPS_SPLIT = 32,OH_NN_OPS_SQRT = 33, OH_NN_OPS_SQUARED_DIFFERENCE = 34, OH_NN_OPS_SQUEEZE = 35, OH_NN_OPS_STACK = 36, OH_NN_OPS_STRIDED_SLICE = 37, OH_NN_OPS_SUB = 38, OH_NN_OPS_TANH = 39, OH_NN_OPS_TILE = 40, OH_NN_OPS_TRANSPOSE = 41, OH_NN_OPS_REDUCE_MEAN = 42, OH_NN_OPS_RESIZE_BILINEAR = 43, OH_NN_OPS_RSQRT = 44,OH_NN_OPS_RESHAPE = 45, OH_NN_OPS_PRELU = 46, OH_NN_OPS_RELU = 47, OH_NN_OPS_RELU6 = 48, OH_NN_OPS_LAYER_NORM = 49, OH_NN_OPS_REDUCE_PROD = 50, OH_NN_OPS_REDUCE_ALL = 51, OH_NN_OPS_QUANT_DTYPE_CAST = 52, OH_NN_OPS_TOP_K = 53, OH_NN_OPS_ARG_MAX = 54, OH_NN_OPS_UNSQUEEZE = 55, OH_NN_OPS_GELU = 56 }	Operator types supported by NNRt.
OH_NN_TensorType { OH_NN_TENSOR = 0, OH_NN_ADD_ACTIVATIONTYPE = 1, OH_NN_AVG_POOL_KERNEL_SIZE = 2, OH_NN_AVG_POOL_STRIDE = 3, OH_NN_AVG_POOL_PAD_MODE = 4, OH_NN_AVG_POOL_PAD = 5, OH_NN_AVG_POOL_ACTIVATION_TYPE = 6, OH_NN_BATCH_NORM_EPSILON = 7, OH_NN_BATCH_TO_SPACE_ND_BLOCKSIZE = 8, OH_NN_BATCH_TO_SPACE_ND_CROPS = 9, OH_NN_CONCAT_AXIS = 10, OH_NN_CONV2D_STRIDES = 11, OH_NN_CONV2D_PAD = 12, OH_NN_CONV2D_DILATION = 13, OH_NN_CONV2D_PAD_MODE = 14, OH_NN_CONV2D_ACTIVATION_TYPE = 15, OH_NN_CONV2D_GROUP = 16, OH_NN_CONV2D_TRANSPOSE_STRIDES = 17, OH_NN_CONV2D_TRANSPOSE_PAD = 18, OH_NN_CONV2D_TRANSPOSE_DILATION = 19, OH_NN_CONV2D_TRANSPOSE_OUTPUT_PADDINGS = 20, OH_NN_CONV2D_TRANSPOSE_PAD_MODE = 21, OH_NN_CONV2D_TRANSPOSE_ACTIVATION_TYPE = 22, OH_NN_CONV2D_TRANSPOSE_GROUP = 23, OH_NN_DEPTHWISE_CONV2D_NATIVE_STRIDES = 24, OH_NN_DEPTHWISE_CONV2D_NATIVE_PAD = 25, OH_NN_DEPTHWISE_CONV2D_NATIVE_DILATION = 26, OH_NN_DEPTHWISE_CONV2D_NATIVE_PAD_MODE = 27, OH_NN_DEPTHWISE_CONV2D_NATIVE_ACTIVATION_TYPE = 28, OH_NN_DIV_ACTIVATIONTYPE = 29, OH_NN_ELTWISE_MODE = 30, OH_NN_FULL_CONNECTION_AXIS = 31, OH_NN_FULL_CONNECTION_ACTIVATIONTYPE = 32, OH_NN_MATMUL_TRANSPOSE_A = 33, OH_NN_MATMUL_TRANSPOSE_B = 34, OH_NN_MATMUL_ACTIVATION_TYPE = 35, OH_NN_MAX_POOL_KERNEL_SIZE = 36, OH_NN_MAX_POOL_STRIDE = 37, OH_NN_MAX_POOL_PAD_MODE = 38, OH_NN_MAX_POOL_PAD = 39,OH_NN_MAX_POOL_ACTIVATION_TYPE = 40, OH_NN_MUL_ACTIVATION_TYPE = 41, OH_NN_ONE_HOT_AXIS = 42, OH_NN_PAD_CONSTANT_VALUE = 43, OH_NN_SCALE_ACTIVATIONTYPE = 44, OH_NN_SCALE_AXIS = 45, OH_NN_SOFTMAX_AXIS = 46, OH_NN_SPACE_TO_BATCH_ND_BLOCK_SHAPE = 47, OH_NN_SPACE_TO_BATCH_ND_PADDINGS = 48, OH_NN_SPLIT_AXIS = 49, OH_NN_SPLIT_OUTPUT_NUM = 50, OH_NN_SPLIT_SIZE_SPLITS = 51, OH_NN_SQUEEZE_AXIS = 52, OH_NN_STACK_AXIS = 53, OH_NN_STRIDED_SLICE_BEGIN_MASK = 54, OH_NN_STRIDED_SLICE_END_MASK = 55, OH_NN_STRIDED_SLICE_ELLIPSIS_MASK = 56, OH_NN_STRIDED_SLICE_NEW_AXIS_MASK = 57, OH_NN_STRIDED_SLICE_SHRINK_AXIS_MASK = 58, OH_NN_SUB_ACTIVATIONTYPE = 59, OH_NN_REDUCE_MEAN_KEEP_DIMS = 60, OH_NN_RESIZE_BILINEAR_NEW_HEIGHT = 61, OH_NN_RESIZE_BILINEAR_NEW_WIDTH = 62, OH_NN_RESIZE_BILINEAR_PRESERVE_ASPECT_RATIO = 63, OH_NN_RESIZE_BILINEAR_COORDINATE_TRANSFORM_MODE = 64, OH_NN_RESIZE_BILINEAR_EXCLUDE_OUTSIDE = 65, OH_NN_LAYER_NORM_BEGIN_NORM_AXIS = 66, OH_NN_LAYER_NORM_EPSILON = 67, OH_NN_LAYER_NORM_BEGIN_PARAM_AXIS = 68, OH_NN_LAYER_NORM_ELEMENTWISE_AFFINE = 69, OH_NN_REDUCE_PROD_KEEP_DIMS = 70, OH_NN_REDUCE_ALL_KEEP_DIMS = 71, OH_NN_QUANT_DTYPE_CAST_SRC_T = 72, OH_NN_QUANT_DTYPE_CAST_DST_T = 73, OH_NN_TOP_K_SORTED = 74, OH_NN_ARG_MAX_AXIS = 75, OH_NN_ARG_MAX_KEEPDIMS = 76, OH_NN_UNSQUEEZE_AXIS = 77 }	Tensor types.

Functions

Name	Description
OH_NNCompilation_Construct (const OH_NNModel model)	Creates a model building instance of the OH_NNCompilation type.
OH_NNCompilation_ConstructWithOfflineModelFile (const char modelPath)	Creates a model building instance based on an offline model file.
OH_NNCompilation_ConstructWithOfflineModelBuffer (const void modelBuffer, size_t modelSize)	Creates a model building instance based on the offline model buffer.
*OH_NNCompilation_ConstructForCache ()	Creates an empty model building instance for later recovery from the model cache.
OH_NNCompilation_ExportCacheToBuffer (OH_NNCompilation compilation, const void buffer, size_t length, size_t *modelSize)	Writes the model cache to the specified buffer.
OH_NNCompilation_ImportCacheFromBuffer (OH_NNCompilation compilation, const void buffer, size_t modelSize)	Reads the model cache from the specified buffer.
OH_NNCompilation_AddExtensionConfig (OH_NNCompilation compilation, const char configName, const void *configValue, const size_t configValueSize)	Adds extended configurations for custom device attributes.
OH_NNCompilation_SetDevice (OH_NNCompilation *compilation, size_t deviceID)	Sets the device for model building and computing.
OH_NNCompilation_SetCache (OH_NNCompilation compilation, const char cachePath, uint32_t version)	Sets the cache directory and version for model building.
OH_NNCompilation_SetPerformanceMode (OH_NNCompilation *compilation, OH_NN_PerformanceMode performanceMode)	Sets the performance mode for model computing.
OH_NNCompilation_SetPriority (OH_NNCompilation *compilation, OH_NN_Priority priority)	Sets the priority for model computing.
OH_NNCompilation_EnableFloat16 (OH_NNCompilation *compilation, bool enableFloat16)	Enables float16 for computing.
OH_NNCompilation_Build (OH_NNCompilation *compilation)	Performs model building.
OH_NNCompilation_Destroy (OH_NNCompilation **compilation)	Destroys a model building instance of the OH_NNCompilation type.
*OH_NNTensorDesc_Create ()	Creates an NN_TensorDesc instance.
OH_NNTensorDesc_Destroy (NN_TensorDesc **tensorDesc)	Releases an NN_TensorDesc instance.
OH_NNTensorDesc_SetName (NN_TensorDesc tensorDesc, const char name)	Sets the name of an NN_TensorDesc instance.
OH_NNTensorDesc_GetName (const NN_TensorDesc tensorDesc, const char *name)	Obtains the name of an NN_TensorDesc instance.
OH_NNTensorDesc_SetDataType (NN_TensorDesc *tensorDesc, OH_NN_DataType dataType)	Sets the data type of an NN_TensorDesc instance.
OH_NNTensorDesc_GetDataType (const NN_TensorDesc tensorDesc, OH_NN_DataType dataType)	Obtains the data type of an NN_TensorDesc instance.
OH_NNTensorDesc_SetShape (NN_TensorDesc tensorDesc, const int32_t shape, size_t shapeLength)	Sets the data shape of an NN_TensorDesc instance.
OH_NNTensorDesc_GetShape (const NN_TensorDesc tensorDesc, int32_t shape, size_t shapeLength)	Obtains the shape of an NN_TensorDesc instance.
OH_NNTensorDesc_SetFormat (NN_TensorDesc *tensorDesc, OH_NN_Format format)	Sets the data format of an NN_TensorDesc instance.
OH_NNTensorDesc_GetFormat (const NN_TensorDesc tensorDesc, OH_NN_Format format)	Obtains the data format of an NN_TensorDesc instance.
OH_NNTensorDesc_GetElementCount (const NN_TensorDesc tensorDesc, size_t elementCount)	Obtains the number of elements in an NN_TensorDesc instance.
OH_NNTensorDesc_GetByteSize (const NN_TensorDesc tensorDesc, size_t byteSize)	Obtains the number of bytes occupied by the tensor data obtained through calculation based on the shape and data type of an NN_TensorDesc instance.
OH_NNTensor_Create (size_t deviceID, NN_TensorDesc tensorDesc)	Creates an NN_Tensor instance from NN_TensorDesc.
OH_NNTensor_CreateWithSize (size_t deviceID, NN_TensorDesc tensorDesc, size_t size)	Creates an NN_Tensor instance based on the specified memory size and NN_TensorDesc instance.
OH_NNTensor_CreateWithFd (size_t deviceID, NN_TensorDesc tensorDesc, int fd, size_t size, size_t offset)	Creates an {@Link NN_Tensor} instance based on the specified file descriptor of the shared memory and NN_TensorDesc instance.
OH_NNTensor_Destroy (NN_Tensor **tensor)	Destroys an NN_Tensor instance.
OH_NNTensor_GetTensorDesc (const NN_Tensor tensor)	Obtains an NN_TensorDesc instance of NN_Tensor.
OH_NNTensor_GetDataBuffer (const NN_Tensor tensor)	Obtains the memory address of NN_Tensor data.
OH_NNTensor_GetFd (const NN_Tensor tensor, int fd)	Obtains the file descriptor of the shared memory where NN_Tensor data is stored.
OH_NNTensor_GetSize (const NN_Tensor tensor, size_t size)	Obtains the size of the shared memory where the NN_Tensor data is stored.
OH_NNTensor_GetOffset (const NN_Tensor tensor, size_t offset)	Obtains the offset of NN_Tensor data in the shared memory.
OH_NNExecutor_Construct (OH_NNCompilation compilation)	Creates an OH_NNExecutor instance.
OH_NNExecutor_GetOutputShape (OH_NNExecutor executor, uint32_t outputIndex, int32_t shape, uint32_t shapeLength)	Obtains the dimension information about the output tensor.
OH_NNExecutor_Destroy (OH_NNExecutor **executor)	Destroys an executor instance to release the memory occupied by it.
OH_NNExecutor_GetInputCount (const OH_NNExecutor executor, size_t inputCount)	Obtains the number of input tensors.
OH_NNExecutor_GetOutputCount (const OH_NNExecutor executor, size_t outputCount)	Obtains the number of output tensors.
OH_NNExecutor_CreateInputTensorDesc (const OH_NNExecutor executor, size_t index)	Creates the description of an input tensor based on the specified index value.
OH_NNExecutor_CreateOutputTensorDesc (const OH_NNExecutor executor, size_t index)	Creates the description of an output tensor based on the specified index value.
OH_NNExecutor_GetInputDimRange (const OH_NNExecutor executor, size_t index, size_t minInputDims, size_t maxInputDims, size_t shapeLength)	Obtains the dimension range of all input tensors.
OH_NNExecutor_SetOnRunDone (OH_NNExecutor *executor, NN_OnRunDone onRunDone)	Sets the callback processing function invoked when the asynchronous inference ends.
OH_NNExecutor_SetOnServiceDied (OH_NNExecutor *executor, NN_OnServiceDied onServiceDied)	Sets the callback processing function invoked when the device driver service terminates unexpectedly during asynchronous inference.
OH_NNExecutor_RunSync (OH_NNExecutor executor, NN_Tensor inputTensor[], size_t inputCount, NN_Tensor *outputTensor[], size_t outputCount)	Performs synchronous inference.
OH_NNExecutor_RunAsync (OH_NNExecutor executor, NN_Tensor inputTensor[], size_t inputCount, NN_Tensor outputTensor[], size_t outputCount, int32_t timeout, void userData)	Performs asynchronous inference.
OH_NNDevice_GetAllDevicesID (const size_t *allDevicesID, uint32_t deviceCount)	Obtains the ID of the device connected to NNRt.
OH_NNDevice_GetName (size_t deviceID, const char **name)	Obtains the name of the specified device.
OH_NNDevice_GetType (size_t deviceID, OH_NN_DeviceType *deviceType)	Obtains the type of the specified device.
*OH_NNQuantParam_Create ()	Creates an NN_QuantParam instance.
OH_NNQuantParam_SetScales (NN_QuantParam quantParams, const double scales, size_t quantCount)	Sets the scaling coefficient for an NN_QuantParam instance.
OH_NNQuantParam_SetZeroPoints (NN_QuantParam quantParams, const int32_t zeroPoints, size_t quantCount)	Sets the zero point for an NN_QuantParam instance.
OH_NNQuantParam_SetNumBits (NN_QuantParam quantParams, const uint32_t numBits, size_t quantCount)	Sets the number of quantization bits for an NN_QuantParam instance.
OH_NNQuantParam_Destroy (NN_QuantParam **quantParams)	Destroys an NN_QuantParam instance.
*OH_NNModel_Construct (void)	Creates a model instance of the OH_NNModel type and constructs a model instance by using the APIs provided by OH_NNModel.
OH_NNModel_AddTensorToModel (OH_NNModel model, const NN_TensorDesc tensorDesc)	Adds a tensor to a model instance.
OH_NNModel_SetTensorData (OH_NNModel model, uint32_t index, const void dataBuffer, size_t length)	Sets the tensor value.
OH_NNModel_SetTensorQuantParams (OH_NNModel model, uint32_t index, NN_QuantParam quantParam)	Sets the quantization parameters of a tensor. For details, see NN_QuantParam.
OH_NNModel_SetTensorType (OH_NNModel *model, uint32_t index, OH_NN_TensorType tensorType)	Sets the tensor type. For details, see OH_NN_TensorType.
OH_NNModel_AddOperation (OH_NNModel model, OH_NN_OperationType op, const OH_NN_UInt32Array paramIndices, const OH_NN_UInt32Array inputIndices, const OH_NN_UInt32Array outputIndices)	Adds an operator to a model instance.
OH_NNModel_SpecifyInputsAndOutputs (OH_NNModel model, const OH_NN_UInt32Array inputIndices, const OH_NN_UInt32Array *outputIndices)	Sets an index value for the input and output tensors of a model.
OH_NNModel_Finish (OH_NNModel *model)	Completes model composition.
OH_NNModel_Destroy (OH_NNModel **model)	Destroys a model instance.
OH_NNModel_GetAvailableOperations (OH_NNModel model, size_t deviceID, const bool isSupported, uint32_t opCount)	Checks whether all operators in a model are supported by the device. The result is indicated by a Boolean value.
OH_NNModel_AddTensor (OH_NNModel model, const OH_NN_Tensor tensor)	Adds a tensor to a model instance.
OH_NNExecutor_SetInput (OH_NNExecutor executor, uint32_t inputIndex, const OH_NN_Tensor tensor, const void *dataBuffer, size_t length)	Sets the data for a single model input.
OH_NNExecutor_SetOutput (OH_NNExecutor executor, uint32_t outputIndex, void dataBuffer, size_t length)	Sets the memory for a single model output.
OH_NNExecutor_Run (OH_NNExecutor *executor)	Executes model inference.
OH_NNExecutor_AllocateInputMemory (OH_NNExecutor executor, uint32_t inputIndex, size_t length)	Applies for shared memory for a single model input on the device.
OH_NNExecutor_AllocateOutputMemory (OH_NNExecutor executor, uint32_t outputIndex, size_t length)	Applies for shared memory for a single model output on the device.
OH_NNExecutor_DestroyInputMemory (OH_NNExecutor executor, uint32_t inputIndex, OH_NN_Memory *memory)	Releases the input memory pointed by the OH_NN_Memory instance.
OH_NNExecutor_DestroyOutputMemory (OH_NNExecutor executor, uint32_t outputIndex, OH_NN_Memory *memory)	Releases the output memory pointed by the OH_NN_Memory instance.
OH_NNExecutor_SetInputWithMemory (OH_NNExecutor executor, uint32_t inputIndex, const OH_NN_Tensor tensor, const OH_NN_Memory *memory)	Shared memory pointed by the OH_NN_Memory instance for a single model input.
OH_NNExecutor_SetOutputWithMemory (OH_NNExecutor executor, uint32_t outputIndex, const OH_NN_Memory memory)	Shared memory pointed by the OH_NN_Memory instance for a single model output.

Type Description

NN_OnRunDone

typedef void(*NN_OnRunDone) (void *userData, OH_NN_ReturnCode errCode, void *outputTensor[], int32_t outputCount)

Description

Handle of the callback processing function invoked when the asynchronous inference ends.

Use the userData parameter to specify the asynchronous inference to query. The value of userData is the same as that passed to OH_NNExecutor_RunAsync. Use the errCode parameter to obtain the return result (defined by OH_NN_ReturnCode of the asynchronous inference.

Since: 11

Parameters

Name	Description
userData	Identifier of asynchronous inference. The value is the same as the userData parameter passed to OH_NNExecutor_RunAsync.
errCode	Return result (defined by OH_NN_ReturnCode of the asynchronous inference.
outputTensor	Output tensor for asynchronous inference. The value is the same as the outputTensor parameter passed to OH_NNExecutor_RunAsync.
outputCount	Number of output tensors for asynchronous inference. The value is the same as the outputCount parameter passed to OH_NNExecutor_RunAsync.

NN_OnServiceDied

typedef void(*NN_OnServiceDied) (void *userData)

Description

Handle of the callback processing function invoked when the device driver service terminates unexpectedly during asynchronous inference.

You need to rebuild the model if the callback is invoked.

You can use the userData parameter to specify the asynchronous inference to query. The value of userData is the same as that passed to OH_NNExecutor_RunAsync.

Since: 11

Parameters

Name	Description
userData	Identifier of asynchronous inference. The value is the same as the userData parameter passed to OH_NNExecutor_RunAsync.

NN_QuantParam

typedef struct NN_QuantParam NN_QuantParam

Description

Quantization parameter handle.

Since: 11

NN_Tensor

typedef struct NN_Tensor NN_Tensor

Description

Tensor handle.

Since: 11

NN_TensorDesc

typedef struct NN_TensorDesc NN_TensorDesc

Description

Tensor description handle.

Since: 11

OH_NN_Memory^(deprecated)

typedef struct OH_NN_Memory OH_NN_Memory

Description

Used to define the memory structure.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: You are advised to use NN_Tensor.

OH_NN_QuantParam^(deprecated)

typedef struct OH_NN_QuantParam OH_NN_QuantParam

Description

Used to define the quantization information.

In quantization scenarios, the 32-bit floating-point data type is quantized into the fixed-point data type according to the following formula:

zh-cn_formulaimage_0000001460519245

where, s and z are quantization parameters, which are stored by scale and zeroPoint in OH_NN_QuanParam. r is a floating point number, q is the quantization result, q_min is the lower bound of the quantization result, and q_max is the upper bound of the quantization result. The calculation method is as follows:

zh-cn_formulaimage_0000001460759081

zh-cn_formulaimage_0000001460639441

The clamp function is defined as follows:

zh-cn_formulaimage_0000001410679330

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: You are advised to use NN_QuantParam.

OH_NN_Tensor^(deprecated)

typedef struct OH_NN_Tensor OH_NN_Tensor

Description

Used to define the tensor structure.

It is usually used to construct data nodes and operator parameters in a model diagram. When constructing a tensor, you need to specify the data type, number of dimensions, dimension information, and quantization information.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: You are advised to use NN_TensorDesc.

OH_NN_UInt32Array

typedef struct OH_NN_UInt32Array OH_NN_UInt32Array

Description

Defines the structure for storing 32-bit unsigned integer arrays.

Since: 9

OH_NNCompilation

typedef struct OH_NNCompilation OH_NNCompilation

Description

Compiler handle.

Since: 9

OH_NNExecutor

typedef struct OH_NNExecutor OH_NNExecutor

Description

Executor handle.

Since: 9

OH_NNModel

typedef struct OH_NNModel OH_NNModel

Description

Model handle.

Since: 9

Enum Description

OH_NN_DataType

enum OH_NN_DataType

Description

Data types supported by NNRt.

Since: 9

Value	Description
OH_NN_UNKNOWN	Unknown type.
OH_NN_BOOL	bool type.
OH_NN_INT8	int8 type.
OH_NN_INT16	int16 type.
OH_NN_INT32	int32 type.
OH_NN_INT64	int64 type.
OH_NN_UINT8	uint8 type.
OH_NN_UINT16	uint16 type.
OH_NN_UINT32	uint32 type.
OH_NN_UINT64	uint64 type.
OH_NN_FLOAT16	float16 type.
OH_NN_FLOAT32	float32 type.
OH_NN_FLOAT64	float64 type.

OH_NN_DeviceType

enum OH_NN_DeviceType

Description

Device types supported by NNRt.

Since: 9

Value	Description
OH_NN_OTHERS	Devices that are not CPU, GPU, or dedicated accelerator.
OH_NN_CPU	CPU.
OH_NN_GPU	GPU.
OH_NN_ACCELERATOR	Dedicated device accelerator.

OH_NN_Format

enum OH_NN_Format

Description

Formats of tensor data.

Since: 9

Value	Description
OH_NN_FORMAT_NONE	The tensor does not have a specific arrangement type (such as scalar or vector).
OH_NN_FORMAT_NCHW	The tensor arranges data in NCHW format.
OH_NN_FORMAT_NHWC	The tensor arranges data in NHWC format.
OH_NN_FORMAT_ND¹¹⁺	The tensor arranges data in ND format. This API is supported since API version 11.

OH_NN_FuseType

enum OH_NN_FuseType : int8_t

Description

Activation function types in the fusion operator for NNRt.

Since: 9

Value	Description
OH_NN_FUSED_NONE	The fusion activation function is not specified.
OH_NN_FUSED_RELU	Fusion relu activation function.
OH_NN_FUSED_RELU6	Fusion relu6 activation function.

OH_NN_OperationType

enum OH_NN_OperationType

Description

Operator types supported by NNRt.

Since: 9

Value	Description
OH_NN_OPS_ADD	Returns the tensor of the sum of the elements corresponding to two input tensors. Input: - input1: first input tensor, which is of the Boolean or number type. - input2: second input tensor, whose data type must be the same as that of the first tensor. Parameters: - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: output: sum of input1 and input2. The data shape is the same as that of the input after broadcasting, and the data type is the same as that of the input with a higher precision.
OH_NN_OPS_AVG_POOL	Applies 2D average pooling to the input tensor, which must be in the NHWC format. The int8 quantization input is supported. If the input contains the padMode parameter: Input: - input: a tensor. Parameters: - kernelSize: average kernel size. It is an int array whose length is 2. It is in the format of [kernel_height, kernel_weight], where the first number indicates the kernel height, and the second number indicates the kernel width. - strides: kernel moving stride. It is an int array whose length is 2. It is in the format of [stride_height, stride_weight], where the first number indicates the moving stride in height, and the second number indicates the moving stride in width. - padMode: padding mode, which is optional. It is an int value, which can be 0 (same) or 1 (valid). The nearest neighbor value is used for padding. 0 (same): The height and width of the output are the same as those of the input. The total padding quantity is calculated horizontally and vertically and evenly distributed to the top, bottom, left, and right if possible. Otherwise, the last additional padding will be completed from the bottom and right. 1 (valid): The possible maximum height and width of the output will be returned in case of no padding. The excessive pixels will be discarded. - activationType: integer constant contained in FuseType. The specified activation function is called before output. If the input contains the padList parameter: Input: - input: a tensor. Parameters: - kernelSize: average kernel size. It is an int array whose length is 2. It is in the format of [kernel_height, kernel_weight], where the first number indicates the kernel height, and the second number indicates the kernel width. - strides: kernel moving stride. It is an int array whose length is 2. It is in the format of [stride_height, stride_weight], where the first number indicates the moving stride in height, and the second number indicates the moving stride in width. padList: padding around input. It is an int array in the format of [top, bottom, left, right], and the nearest neighbor values are used for padding. - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: output: average pooling result of the input.
OH_NN_OPS_BATCH_NORM	Performs batch normalization on a tensor to scale and shift tensor elements, relieving potential covariate shift in a batch of data. Input: - input: n-dimensional tensor in the shape of [N, ..., C]. The nth dimension is the number of channels. - scale: 1D tensor of the scaling factor used to scale the first normalized tensor. - offset: 1D tensor used to move to the first normalized tensor. - mean: 1D tensor of the overall mean value. It is used only for inference. In case of training, this parameter must be left empty. - variance: 1D tensor used for the overall variance. It is used only for inference. In case of training, this parameter must be left empty. Parameters: epsilon: fixed small additional value. Output: output: n-dimensional output tensor whose shape and data type are the same as those of the input.
OH_NN_OPS_BATCH_TO_SPACE_ND	Divides batch dimension of a 4D tensor into small blocks by block_shape and interleaves them into the spatial dimension. Parameters: - input: input tensor. The dimension will be divided into small blocks, and these blocks will be interleaved into the spatial dimension. Output: blockSize: size of blocks to be interleaved into the spatial dimension. The value is an array in the format of [height_block, weight_block]. crops: elements truncated from the spatial dimension of the output. The value is a 2D array in the format of [[crop0_start, crop0_end], [crop1_start, crop1_end]] with the shape of (2, 2). Output: - output. If the shape of input is (n,h,w,c), the shape of output is (n',h',w',c'), where n' = n / (block_shape[0] * block_shape[1]), h' = h * block_shape[0] - crops[0][0] - crops[0][1], w' = w * block_shape[1] - crops[1][0] - crops[1][1], and c'= c.
OH_NN_OPS_BIAS_ADD	Offsets the data in each dimension of the input tensor. Input: input: input tensor, which can have two to five dimensions. bias: offset of the number of input dimensions. Output: output: sum of the input tensor and the bias in each dimension.
OH_NN_OPS_CAST	Converts the data type in the input tensor. Input: - input: input tensor. - type: converted data type. Output: output: converted tensor.
OH_NN_OPS_CONCAT	Connects tensors in a specified dimension. Input: - input: n input tensors. Parameters: - axis: dimension for connecting tensors. Output: output: result of connecting n tensors along the axis.
OH_NN_OPS_CONV2D	Sets a 2D convolutional layer. If the input contains the padMode parameter: Input: - input: input tensor. - weight: convolution weight in the format of [outChannel, kernelHeight, kernelWidth, inChannel/group]. The value of inChannel must be exactly divided by the value of group. - bias: bias of the convolution. It is an array with a length of [outChannel]. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. Parameters: - stride: moving stride of the convolution kernel in height and weight. It is an int array in the format of [strideHeight, strideWidth]. - dilation: dilation size of the convolution kernel in height and weight. It is an int array in the format of [dilationHeight, dilationWidth]. The value must be greater than or equal to 1 and cannot exceed the height and width of input. - padMode: padding mode of input. The value is of the int type and can be 0 (same) or 1 (valid). 0 (same): The height and width of the output are the same as those of the input. The total padding quantity is calculated horizontally and vertically and evenly distributed to the top, bottom, left, and right if possible. Otherwise, the last additional padding will be completed from the bottom and right. 1 (valid): The possible maximum height and width of the output will be returned in case of no padding. The excessive pixels will be discarded. - group: number of groups in which the input is divided by in_channel. The value is of the int type. If group is 1, it is a conventional convolution. If group is greater than 1 and less than or equal to in_channel, it is a group convolution. - activationType: integer constant contained in FuseType. The specified activation function is called before output. If the input contains the padList parameter: Input: - input: input tensor. - weight: convolution weight in the format of [outChannel, kernelHeight, kernelWidth, inChannel/group]. The value of inChannel must be exactly divided by the value of group. - bias: bias of the convolution. It is an array with a length of [outChannel]. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. Parameters: - stride: moving stride of the convolution kernel in height and weight. It is an int array in the format of [strideHeight, strideWidth]. - dilation: dilation size of the convolution kernel in height and weight. It is an int array in the format of [dilationHeight, dilationWidth]. The value must be greater than or equal to 1 and cannot exceed the height and width of input. - padList: padding around input. It is an int array in the format of [top, bottom, left, right]. - group: number of groups in which the input is divided by in_channel. The value is of the int type. If group is 1, it is a conventional convolution. If group is in_channel, it is depthwiseConv2d. In this case, group==in_channel==out_channel. If group is greater than 1 and less than in_channel, it is a group convolution. In this case, out_channel==group. - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: convolution computing result.
OH_NN_OPS_CONV2D_TRANSPOSE	Sets 2D convolution transposition. If the input contains the padMode parameter: Input: - input: input tensor. - weight: convolution weight in the format of [outChannel, kernelHeight, kernelWidth, inChannel/group]. The value of inChannel must be exactly divided by the value of group. - bias: bias of the convolution. It is an array with a length of [outChannel]. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. - stride: moving stride of the convolution kernel in height and weight. It is an int array in the format of [strideHeight, strideWidth]. Parameters: - dilation: dilation size of the convolution kernel in height and weight. It is an int array in the format of [dilationHeight, dilationWidth]. The value must be greater than or equal to 1 and cannot exceed the height and width of input. - padMode: padding mode of input. The value is of the int type and can be 0 (same) or 1 (valid). 0 (same): The height and width of the output are the same as those of the input. The total padding quantity is calculated horizontally and vertically and evenly distributed to the top, bottom, left, and right if possible. Otherwise, the last additional padding will be completed from the bottom and right. 1 (valid): The possible maximum height and width of the output will be returned in case of no padding. The excessive pixels will be discarded. - group: number of groups in which the input is divided by in_channel. The value is of the int type. If group is 1, it is a conventional convolution. If group is greater than 1 and less than or equal to in_channel, it is a group convolution. - outputPads: padding along the height and width of the output tensor. The value is an int number, a tuple, or a list of two integers. It can be a single integer to specify the same value for all spatial dimensions. The amount of output padding along a dimension must be less than the stride along this dimension. - activationType: integer constant contained in FuseType. The specified activation function is called before output. If the input contains the padList parameter: Input: - input: input tensor. - weight: convolution weight in the format of [outChannel, kernelHeight, kernelWidth, inChannel/group]. The value of inChannel must be exactly divided by the value of group. - bias: bias of the convolution. It is an array with a length of [outChannel]. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. Parameters: - stride: moving stride of the convolution kernel in height and weight. It is an int array in the format of [strideHeight, strideWidth]. - dilation: dilation size of the convolution kernel in height and weight. It is an int array in the format of [dilationHeight, dilationWidth]. The value must be greater than or equal to 1 and cannot exceed the height and width of input. - padList: padding around input. It is an int array in the format of [top, bottom, left, right]. - group: number of groups in which the input is divided by in_channel. The value is of the int type. If group is 1, it is a conventional convolution. If group is greater than 1 and less than or equal to in_channel, it is a group convolution. - outputPads: padding along the height and width of the output tensor. The value is an int number, a tuple, or a list of two integers. It can be a single integer to specify the same value for all spatial dimensions. The amount of output padding along a dimension must be less than the stride along this dimension. - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: output: computing result after convolution and transposition.
OH_NN_OPS_DEPTHWISE_CONV2D_NATIVE	Sets 2D depthwise separable convolution. If the input contains the padMode parameter: Input: - input: input tensor. - weight: convolution weight in the format of [outChannel, kernelHeight, kernelWidth, 1]. outChannel is equal to channelMultiplier multiplied by inChannel. - bias: bias of the convolution. It is an array with a length of [outChannel]. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. Parameters: - stride: moving stride of the convolution kernel in height and weight. It is an int array in the format of [strideHeight, strideWidth]. - dilation: dilation size of the convolution kernel in height and weight. It is an int array in the format of [dilationHeight, dilationWidth]. The value must be greater than or equal to 1 and cannot exceed the height and width of input. - padMode: padding mode of input. The value is of the int type and can be 0 (same) or 1 (valid). The value 0 (same) indicates that the height and width of the output are the same as those of the input. The total padding quantity is calculated horizontally and vertically and evenly distributed to the top, bottom, left, and right if possible. The value 1 (valid) indicates that the possible maximum height and width of the output will be returned in case of no padding. The excessive pixels will be discarded. - activationType: integer constant contained in FuseType. The specified activation function is called before output. If the input contains the padList parameter: Input: - input: input tensor. - weight: convolution weight in the format of [outChannel, kernelHeight, kernelWidth, 1]. outChannel is equal to channelMultiplier multiplied by inChannel. - bias: bias of the convolution. It is an array with a length of [outChannel]. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. Parameters: - stride: moving stride of the convolution kernel in height and weight. It is an int array in the format of [strideHeight, strideWidth]. - dilation: dilation size of the convolution kernel in height and weight. It is an int array in the format of [dilationHeight, dilationWidth]. The value must be greater than or equal to 1 and cannot exceed the height and width of input. - padList: padding around input. It is an int array in the format of [top, bottom, left, right]. - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: convolution computing result.
OH_NN_OPS_DIV	Divides two input scalars or tensors. Input: - input1: first input, which is a number, a bool, or a tensor whose data type is number or Boolean. - input2: second input, which must meet the following requirements: If the first input is a real number or Boolean value, the second input must be a tensor whose data type is real number or Boolean value. Parameters: - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: result of dividing input1 by input2.
OH_NN_OPS_ELTWISE	Sets parameters to perform product (dot product), sum (addition and subtraction), or max (larger value) on the input. Input: - input1: first input tensor. - input2: second input tensor. Parameters: - mode: operation mode. The value is an enumerated value. Output: - output: computing result, which has the same data type and shape of output and input1.
OH_NN_OPS_EXPAND_DIMS	Adds an additional dimension to a tensor in the given dimension. Input: - input: input tensor. - axis: index of the dimension to be added. The value is of the int32_t type and must be a constant in the range [-dim-1, dim]. Output: - output: tensor after dimension expansion.
OH_NN_OPS_FILL	Creates a tensor of the specified dimensions and fills it with a scalar. Input: - value: scalar used to fill the tensor. - shape: dimensions of the tensor to be created. Output: - output: generated tensor, which has the same data type as value. The tensor shape is specified by the shape parameter.
OH_NN_OPS_FULL_CONNECTION	Sets a full connection. The entire input is used as the feature map for feature extraction. Input: - input: full-connection input tensor. - weight: weight tensor for a full connection. - bias: full-connection bias. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. Parameters: - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: computed tensor. If the input contains the axis parameter: Input: - input: full-connection input tensor. - weight: weight tensor for a full connection. - bias: full-connection bias. In quantization scenarios, quantization parameters are not required for bias. You only need to input data of the OH_NN_INT32 type. The actual quantization parameters are determined by input and weight. Parameters: - axis: axis in which the full connection is applied. The specified axis and its following axes are converted into a 1D tensor for applying the full connection. - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: computed tensor.
OH_NN_OPS_GATHER	Returns the slice of the input tensor based on the specified index and axis. Input: - input: tensor to be sliced. - inputIndices: indices of the specified input on the axis. The value is an array of the int type and must be in the range [0,input.shape[axis]). - axis: axis on which input is sliced. The value is an array with one element of the int32_t type. Output: - output: sliced tensor.
OH_NN_OPS_HSWISH	Calculates the activation value of the input Hswish. Input: - An n-dimensional input tensor. Output: n-dimensional Hswish activation value. The data type is the same as that of shape and input.
OH_NN_OPS_LESS_EQUAL	For input1 and input2, calculate the result of input1[i]<=input2[i] for each pair of elements, where i is the index of each element in the input tensor. Input: - input1, which can be a real number, Boolean value, or tensor whose data type is real number or NN_BOOL. - input2, which can be a real number or a Boolean value if input_1 is a tensor and must be a tensor with the data type of real number or NN_BOOL if input_1 is not a tensor. Output: A tensor of the NN_BOOL type. When a quantization model is used, the quantization parameters of the output cannot be omitted. However, values of the quantization parameters do not affect the result.
OH_NN_OPS_MATMUL	Calculates the inner product of input1 and input2. Input: - input1: n-dimensional input tensor. - input2: n-dimensional input tensor. Parameters: - TransposeX: Boolean value indicating whether to transpose input1. - TransposeY: Boolean value indicating whether to transpose input2. Output: - output: inner product obtained after calculation. In case of type!=NN_UNKNOWN, the output data type is determined by type. In case of type==NN_UNKNOWN, the output data type depends on the data type converted during computing of inputX and inputY.
OH_NN_OPS_MAXIMUM	Calculates the maximum of input1 and input2 element-wise. The inputs of input1 and input2 comply with the implicit type conversion rules to make the data types consistent. The input must be two tensors or one tensor and one scalar. If the input contains two tensors, their data types cannot be both NN_BOOL. Their shapes can be broadcast to the same size. If the input contains one tensor and one scalar, the scalar must be a constant. Input: - input1: n-dimensional input tensor of the real number or NN_BOOL type. - input2: n-dimensional input tensor of the real number or NN_BOOL type. Output: - output: n-dimensional output tensor. The shape and data type of output are the same as those of the two inputs with higher precision or bits.
OH_NN_OPS_MAX_POOL	Applies 2D maximum pooling to the input tensor. If the input contains the padMode parameter: Input: - input: a tensor. Parameters: - kernelSize: average kernel size. It is an int array whose length is 2. It is in the format of [kernel_height, kernel_weight], where the first number indicates the kernel height, and the second number indicates the kernel width. - strides: kernel moving stride. It is an int array whose length is 2. It is in the format of [stride_height, stride_weight], where the first number indicates the moving stride in height, and the second number indicates the moving stride in width. - padMode: padding mode, which is optional. It is an int value, which can be 0 (same) or 1 (valid). The nearest neighbor value is used for padding. 0 (same): The height and width of the output are the same as those of the input. The total padding quantity is calculated horizontally and vertically and evenly distributed to the top, bottom, left, and right if possible. Otherwise, the last additional padding will be completed from the bottom and right. 1 (valid): The possible maximum height and width of the output will be returned in case of no padding. The excessive pixels will be discarded. - activationType: integer constant contained in FuseType. The specified activation function is called before output. If the input contains the padList parameter: Input: - input: a tensor. Parameters: - kernelSize: average kernel size. It is an int array whose length is 2. It is in the format of [kernel_height, kernel_weight], where the first number indicates the kernel height, and the second number indicates the kernel width. - strides: kernel moving stride. It is an int array whose length is 2. It is in the format of [stride_height, stride_weight], where the first number indicates the moving stride in height, and the second number indicates the moving stride in width. padList: padding around input. It is an int array in the format of [top, bottom, left, right], and the nearest neighbor values are used for padding. - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: tensor obtained after maximum pooling is applied to the input.
OH_NN_OPS_MUL	Multiplies elements in the same positions of inputX and inputY to obtain the output. If inputX and inputY have different shapes, expand them to the same shape through broadcast and then perform multiplication. Input: - input1: n-dimensional tensor. - input2: n-dimensional tensor. Parameters: - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: Product of each element of input1 and input2.
OH_NN_OPS_ONE_HOT	Generates a one-hot tensor based on the positions specified by indices. The positions specified by indices are determined by on_value, and other positions are determined by off_value. Input: - indices: n-dimensional tensor. Each element in indices determines the position of on_value in each one-hot vector. - depth, an integer scalar that determines the depth of the one-hot vector. The value of depth must be greater than 0. - on_value, a scalar that specifies a valid value in the one-hot vector. - off_value, a scalar that specifies the values of other locations in the one-hot vector except the valid value. Parameters: - axis: integer scalar that specifies the dimension for inserting the one-hot. Assume that the shape of indices is [N, C], and the value of depth is D. When axis is 0, the shape of the output is [D, N, C]. When axis is -1, the shape of the output is [N, C, D]. When axis is 1, the shape of the output is [N, D, C]. Output: - output: (n+1)-dimensional tensor if indices is an n-dimensional tensor. The output shape is determined by indices and axis.
OH_NN_OPS_PAD	Pads inputX in the specified dimensions. Input: - inputX: n-dimensional tensor in [BatchSize, ...] format. - paddings: 2D tensor that specifies the length to pad in each dimension. The shape is [n, 2]. For example, paddings[i][0] indicates the number of paddings to be added preceding inputX in the nth dimension, and paddings[i][1] indicates the number of paddings to be added following inputX in the nth dimension. Parameters: - padValues: value to be added to the pad operation. The value is a constant with the same data type as inputX. Output: - output: n-dimensional tensor after padding, with the same dimensions and data type as inputX. The shape is determined by inputX and paddings; that is, output.shape[i] = input.shape[i] + paddings[i][0]+paddings[i][1].
OH_NN_OPS_POW	Calculates the y power of each element in input. The input must contain two tensors or one tensor and one scalar. If the input contains two tensors, their data types cannot be both NN_BOOL, and their shapes must be the same. If the input contains one tensor and one scalar, the scalar must be a constant. Input: - input: real number, Boolean value, or tensor whose data type is real number or NN_BOOL. - y: real number, Boolean value, or tensor whose data type is real number or NN_BOOL. Output: - output: tensor, whose shape is determined by the shape of input and y after broadcasting.
OH_NN_OPS_SCALE	Scales a tensor. Input: - input: n-dimensional tensor. - scale: scaling tensor. - bias: bias tensor. Parameters: - axis: dimensions to be scaled. - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: scaled n-dimensional tensor, whose data type is the same as that of input and shape is determined by axis.
OH_NN_OPS_SHAPE	Calculates the shape of the input tensor. Input: - input: n-dimensional tensor. Output: - output: integer array representing the dimensions of the input tensor.
OH_NN_OPS_SIGMOID	Applies the sigmoid operation to the input tensor. Input: - input: n-dimensional tensor. Output: - output: result of the sigmoid operation. It is an n-dimensional tensor with the same data type and shape as input.
OH_NN_OPS_SLICE	Slices a tensor of the specified size from the input tensor in each dimension. Input: - input: n-dimensional input tensor. - begin: start of the slice, which is an array of integers greater than or equal to 0. - size: slice length, which is an array of integers greater than or equal to 1. Assume that a dimension is 1<=size[i]<=input.shape[i]-begin[i]. Output: - output: n-dimensional tensor obtained by slicing. The TensorType, shape, and size of the output are the same as those of the input.
OH_NN_OPS_SOFTMAX	Applies the softmax operation to the input tensor. Input: - input: n-dimensional input tensor. Parameters: - axis: dimension in which the softmax operation is performed. The value is of the int64 type. It is an integer in the range [-n, n). Output: - output: result of the softmax operation. It is an n-dimensional tensor with the same data type and shape as input.
OH_NN_OPS_SPACE_TO_BATCH_ND	Divides a 4D tensor into small blocks and combines these blocks in the original batch. The number of blocks is blockShape[0] multiplied by blockShape[1]. Input: - input: 4D tensor. Parameters: - blockShape: a pair of integers. Each of them is greater than or equal to 1. - paddings: a pair of arrays. Each of them consists of two integers. The four integers that form paddings must be greater than or equal to 0. paddings[0][0] and paddings[0][1] specify the number of paddings in the third dimension, and paddings[1][0] and paddings[1][1] specify the number of paddings in the fourth dimension. Output: - output: 4D tensor with the same data type as input. The shape is determined by input, blockShape, and paddings. Assume that the input shape is [n,c,h,w], then: output.shape[0] = n * blockShape[0] * blockShape[1] output.shape[1] = c output.shape[2] = (h + paddings[0][0] + paddings[0][1]) / blockShape[0] output.shape[3] = (w + paddings[1][0] + paddings[1][1]) / blockShape[1]. Note that (h + paddings[0][0] + paddings[0][1]) and (w + paddings[1][0] + paddings[1][1]) must be exactly divisible by (h + paddings[0][0] + paddings[0][1]) and (w + paddings[1][0] + paddings[1][1]).
OH_NN_OPS_SPLIT	Splits the input into multiple tensors along the axis dimension. The number of tensors is specified by outputNum. Input: - input: n-dimensional tensor. Parameters: - outputNum: number of output tensors. The data type is long int. - size_splits: size of each tensor split from the input. The value is a 1D tensor of the int type. If size_splits is empty, the input will be evenly split into tensors of the same size. In this case, input.shape[axis] can be exactly divisible by outputNum. If size_splits is not empty, the sum of all its elements must be equal to input.shape[axis]. - axis: splitting dimension of the int type. Output: - outputs: array of n-dimensional tensors, with the same data type and dimensions. The data type of each tensor is the same as that of input.
OH_NN_OPS_SQRT	Calculates the square root of a tensor. Input: - input: n-dimensional tensor. Output: - output: square root of the input. It is an n-dimensional tensor with the same data type and shape as input.
OH_NN_OPS_SQUARED_DIFFERENCE	Calculates the square of the difference between two tensors. The SquaredDifference operator supports tensor and tensor subtraction. If two tensors have different TensorTypes, the Sub operator converts the low-precision tensor to a high-precision one. If two tensors have different shapes, the two tensors can be extended to tensors with the same shape through broadcast. Input: - input1: minuend, which is a tensor of the NN_FLOAT16, NN_FLOAT32, NN_INT32, or NN_BOOL type. - input2: subtrahend, which is a tensor of the NN_FLOAT16, NN_FLOAT32, NN_INT32, or NN_BOOL type. Output: - output: square of the difference between two inputs. The output shape is determined by input1 and input2. If they have the same shape, the output tensor has the same shape as them. If they have different shapes, perform the broadcast operation on input1 and input2 and perform subtraction. TensorType of the output is the same as that of the input tensor with higher precision.
OH_NN_OPS_SQUEEZE	Removes the dimension with a length of 1 from the specified axis. The int8 quantization input is supported. Assume that the input shape is [2, 1, 1, 2, 2] and axis is [0,1], the output shape is [2, 1, 2, 2], which means the dimension whose length is 0 between dimension 0 and dimension 1 is removed. Input: - input: n-dimensional tensor. Parameters: - axis: dimension to be removed. The value is of int64_t type and can be an integer in the range [-n, n) or an array. Output: - output: output tensor.
OH_NN_OPS_STACK	Stacks multiple tensors along the specified axis. If each tensor has n dimensions before stacking, the output tensor will have n+1 dimensions. Input: - input: input for stacking, which can contain multiple n-dimensional tensors. Each of them must have the same shape and type. Parameters: - axis: dimension for tensor stacking, which is an integer. The value range is [-(n+1),(n+1)), which means a negative number is allowed. Output: - output: stacking result of the input along the axis dimension. The value is an n+1-dimensional tensor and has the same TensorType as the input.
OH_NN_OPS_STRIDED_SLICE	Slices a tensor with the specified stride. Input: - input: n-dimensional input tensor. - begin: start of slicing, which is a 1D tensor. The length of begin is n. begin[i] specifies the start of slicing in the ith dimension. - end: end of slicing, which is a 1D tensor. The length of end is n. end[i] specifies the end of slicing in the ith dimension. - strides: slicing stride, which is a 1D tensor. The length of strides is n. strides[i] specifies the stride at which the tensor is sliced in the ith dimension. Parameters: - beginMask: an integer used to mask begin. beginMask is represented in binary code. In case of binary(beginMask)[i]==1, for the ith dimension, elements are sliced from the first element at strides[i] until the end[i]-1 element. - endMask: an integer used to mask end. endMask is represented in binary code. In case of binary(endMask)[i]==1, elements are sliced from the element at the begin[i] position in the ith dimension until the tensor boundary at strides[i]. - ellipsisMask: integer used to mask begin and end. ellipsisMask is represented in binary code. In case of binary(ellipsisMask)[i]==1, elements are sliced from the first element at strides[i] in the ith dimension until the tensor boundary. Only one bit of binary(ellipsisMask) can be a non-zero value. - newAxisMask: new dimension, which is an integer. newAxisMask is represented in binary code. In case of binary(newAxisMask)[i]==1, a new dimension whose length is 1 is inserted into the ith dimension. - shrinkAxisMask: shrinking dimension, which is an integer. shrinkAxisMask is represented in binary code. In the case of binary(shrinkAxisMask)[i]==1, all elements in the ith dimension will be discarded, and the length of the ith dimension is shrunk to 1. Output: - A tensor, with the same data type as input. The number of dimensions of the output tensor is rank(input[0])+1.
OH_NN_OPS_SUB	Calculates the difference between two tensors. Input: - input1: minuend, which is a tensor. - input2: subtrahend, which is a tensor. Parameters: - activationType: integer constant contained in FuseType. The specified activation function is called before output. Output: - output: difference between the two tensors. The shape of the output is determined by input1 and input2. When the shapes of input1 and input2 are the same, the shape of the output is the same as that of input1 and input2. If the shapes of input1 and input2 are different, the output is obtained after the broadcast operation is performed on input1 or input2. TensorType of the output is the same as that of the input tensor with higher precision.
OH_NN_OPS_TANH	Computes hyperbolic tangent of the input tensor. Input: - input: n-dimensional tensor. Output: - output: hyperbolic tangent of the input. The TensorType and tensor shape are the same as those of the input.
OH_NN_OPS_TILE	Copies a tensor for the specified number of times. Input: - input: n-dimensional tensor. - multiples: number of times that the input tensor is copied in each dimension. The value is a 1D tensor. The length m is not less than the number of dimensions, that is, n. Output: - An m-dimensional tensor whose TensorType is the same as that of the input. If input and multiples have the same length, input and output have the same number of dimensions. If the length of multiples is greater than n, 1 is used to fill the input dimension, and then the input is copied in each dimension the specified times to obtain the m-dimensional tensor.
OH_NN_OPS_TRANSPOSE	Transposes data of input 0 based on permutation. Input: - input: n-dimensional tensor to be transposed. - permutation: The value is a 1D tensor whose length is the same as the number of dimensions of input 0. Output: - output: n-dimensional tensor. TensorType of output 0 is the same as that of input 0, and the output shape is determined by the shape and permutation of input 0.
OH_NN_OPS_REDUCE_MEAN	Calculates the average value in the specified dimension. If keepDims is set to false, the number of dimensions is reduced for the input; if keepDims is set to true, the number of dimensions is retained. Input: - input: n-dimensional input tensor, where n is less than 8. - axis: dimension used to calculate the average value. The value is a 1D tensor. The value range of each element in axis is [–n, n). Parameters: - keepDims: whether to retain the dimension. The value is a Boolean value. Output: - output: m-dimensional output tensor whose data type is the same as that of the input. If keepDims is false, m==n. If keepDims is true, m<n.
OH_NN_OPS_RESIZE_BILINEAR	The Bilinear method is used to deform the input based on the given parameters. Input: - input: 4D input tensor. Each element in the input cannot be less than 0. The input layout must be [batchSize, height, width, channels]. Parameters: - newHeight: resized height of the 4D tensor. - newWidth: resized width of the 4D tensor. - preserveAspectRatio: whether to maintain the height/width ratio of input after resizing. - coordinateTransformMode: coordinate transformation method used by the resize operation. The value is an int32 integer. Currently, the following APIs are supported: - excludeOutside: an int64 floating point number. When its value is 1, the sampling weight of the part that exceeds the boundary of input is set to 0, and other weights are normalized. Output: - output: n-dimensional tensor, with the same shape and data type as input.
OH_NN_OPS_RSQRT	Calculates the reciprocal of the square root of a tensor. Input: - input: n-dimensional tensor, where n is less than 8. Each element of the tensor cannot be less than 0. Output: - output: n-dimensional tensor, with the same shape and data type as input.
OH_NN_OPS_RESHAPE	Reshapes a tensor. Input: - input: n-dimensional input tensor. - InputShape: shape of the output tensor. The value is a 1D constant tensor. Output: - output: tensor whose data type is the same as that of input and shape is determined by InputShape.
OH_NN_OPS_PRELU	Calculates the PReLU activation value of input and weight. Input: - input: n-dimensional tensor. If n is greater than or equal to 2, inputX must be [BatchSize, ..., Channels]. The second dimension is the number of channels. - weight: 1D tensor. The length of weight must be 1 or equal to the number of channels. If the length of weight is 1, all channels share the same weight. If the length of weight is equal to the number of channels, each channel exclusively has a weight. If n is less than 2 for inputX, the weight length must be 1. Output: - output: PReLU activation value of input, with the same shape and data type as inputX.
OH_NN_OPS_RELU	Calculates the Relu activation value of input. Input: - input: n-dimensional input tensor. Output: - output: n-dimensional tensor, with the same data type and shape as the input tensor.
OH_NN_OPS_RELU6	Calculates the Relu6 activation value of the input, that is, calculate min(max(x, 0), 6) for each element x in the input. Input: - input: n-dimensional input tensor. Output: - output: n-dimensional Relu6 tensor, with the same data type and shape as the input tensor.
OH_NN_OPS_LAYER_NORM	Applies layer normalization for a tensor from the specified axis. Input: - input: n-dimensional input tensor. - gamma: m-dimensional tensor. The dimensions of gamma must be the same as the shape of the part of the input tensor to normalize. - beta: m-dimensional tensor with the same shape as gamma. Parameters: - beginAxis is an NN_INT32 scalar that specifies the axis from which normalization starts. The value range is [1, rank(input)). - epsilon is a scalar of NN_FLOAT32. It is a tiny amount in the normalization formula. The common value is 1e-7. Output: - output: n-dimensional tensor, with the same data type and shape as the input tensor.
OH_NN_OPS_REDUCE_PROD	Calculates the accumulated value for a tensor along the specified dimension. Input: - input: n-dimensional input tensor, where n is less than 8. - axis: dimension used to calculate the product. The value is a 1D tensor. The value range of each element in axis is [–n, n). Parameters: - keepDims: whether to retain the dimension. The value is a Boolean value. When its value is true, the number of output dimensions is the same as that of the input. When its value is false, the number of output dimensions is reduced. Output: - output: m-dimensional output tensor whose data type is the same as that of the input. If keepDims is false, m==n. If keepDims is true, m<n.
OH_NN_OPS_REDUCE_ALL	Operates the logical OR in the specified dimension. If keepDims is set to false, the number of dimensions is reduced for the input; if keepDims is set to true, the number of dimensions is retained. Input: - An n-dimensional input tensor, where n is less than 8. - A 1D tensor specifying the dimension used to operate the logical OR. The value range of each element in axis is [–n, n). Parameters: - keepDims: whether to retain the dimension. The value is a Boolean value. Output: - output: m-dimensional output tensor whose data type is the same as that of the input. If keepDims is false, m==n. If keepDims is true, m<n.
OH_NN_OPS_QUANT_DTYPE_CAST	Converts the data type. Input: - input: n-dimensional tensor. Parameters: - src_t: data type of the input. - dst_t: data type of the output. Output: - output: n-dimensional tensor. The data type is determined by input2. The output shape is the same as the input shape.
OH_NN_OPS_TOP_K	Obtains the values and indices of the largest k entries in the last dimension. Input: - input: n-dimensional tensor. - input k: first k records of data and their indices. Parameters: - sorted: order of sorting. The value true means descending and false means ascending. Output: - output0: largest k elements in each slice of the last dimension. - output1: index of the value in the last dimension of the input.
OH_NN_OPS_ARG_MAX	Returns the index of the maximum tensor value across axes. Input: - input: n-dimensional tensor (N, ), where means any number of additional dimensions. Parameters: - axis: dimension for calculating the index of the maximum. - keep_dims: whether to maintain the input tensor dimension. The value is a Boolean value. Output: - output: index of the maximum input tensor on the axis. The value is a tensor.
OH_NN_OPS_UNSQUEEZE	Adds a dimension based on the value of axis. Input: - input: n-dimensional tensor. Parameters: - axis: dimension to add. The value of axis can be an integer or an array of integers. The value range of the integer is [-n, n). Output: - output: output tensor.
OH_NN_OPS_GELU	Activates the Gaussian error linear unit. The int quantization input is not supported. output=0.5input(1+tanh(input/2)) Input: - An n-dimensional input tensor. Output: - output: n-dimensional tensor, with the same data type and shape as the input tensor.

OH_NN_PerformanceMode

enum OH_NN_PerformanceMode

Description

Performance modes of the device.

Since: 9

Value	Description
OH_NN_PERFORMANCE_NONE	No performance mode preference.
OH_NN_PERFORMANCE_LOW	Low power consumption mode.
OH_NN_PERFORMANCE_MEDIUM	Medium performance mode.
OH_NN_PERFORMANCE_HIGH	High performance mode.
OH_NN_PERFORMANCE_EXTREME	Ultimate performance mode.

OH_NN_Priority

enum OH_NN_Priority

Description

Priorities of a model inference task.

Since: 9

Value	Description
OH_NN_PRIORITY_NONE	No priority preference.
OH_NN_PRIORITY_LOW	Low priority.
OH_NN_PRIORITY_MEDIUM	Medium priority.
OH_NN_PRIORITY_HIGH	High priority.

OH_NN_ReturnCode

enum OH_NN_ReturnCode

Description

Error codes for NNRt.

Since: 9

Value	Description
OH_NN_SUCCESS	The operation is successful.
OH_NN_FAILED	Operation failed.
OH_NN_INVALID_PARAMETER	Invalid parameter.
OH_NN_MEMORY_ERROR	Memory-related error, for example, insufficient memory, memory data copy failure, or memory application failure.
OH_NN_OPERATION_FORBIDDEN	Invalid operation.
OH_NN_NULL_PTR	Null pointer.
OH_NN_INVALID_FILE	Invalid file.
OH_NN_UNAVALIDABLE_DEVICE^(deprecated)	Hardware error, for example, HDL service crash. Deprecated: This API is deprecated since API version 11. Substitute: OH_NN_UNAVAILABLE_DEVICE is recommended.
OH_NN_INVALID_PATH	Invalid path.
OH_NN_TIMEOUT¹¹⁺	Execution timed out. This API is supported since API version 11.
OH_NN_UNSUPPORTED¹¹⁺	Not supported. This API is supported since API version 11.
OH_NN_CONNECTION_EXCEPTION¹¹⁺	Connection error. This API is supported since API version 11.
OH_NN_SAVE_CACHE_EXCEPTION¹¹⁺	Failed to save the cache. This API is supported since API version 11.
OH_NN_DYNAMIC_SHAPE¹¹⁺	Dynamic shape. This API is supported since API version 11.
OH_NN_UNAVAILABLE_DEVICE¹¹⁺	Hardware error, for example, HDL service crash. This API is supported since API version 11.

OH_NN_TensorType

enum OH_NN_TensorType

Description

Defines tensor types.

Tensors are usually used to set the input, output, and operator parameters of a model. When a tensor is used as the input or output of a model (or operator), set the tensor type to OH_NN_TENSOR. When the tensor is used as an operator parameter, select an enumerated value other than OH_NN_TENSOR as the tensor type. Assume that pad of the OH_NN_OPS_CONV2D operator is being set. You need to set the type attribute of the OH_NN_Tensor instance to OH_NN_CONV2D_PAD. The settings of other operator parameters are similar. The enumerated values are named in the format OH_NN_{Operator name}_{Attribute name}.

Since: 9

Value	Description
OH_NN_TENSOR	Used when the tensor is used as the input or output of a model (or operator).
OH_NN_ADD_ACTIVATIONTYPE	Used when the tensor is used as the activationType parameter of the Add operator.
OH_NN_AVG_POOL_KERNEL_SIZE	Used when the tensor is used as the kernel_size parameter of the AvgPool operator.
OH_NN_AVG_POOL_STRIDE	Used when the tensor is used as the stride parameter of the AvgPool operator.
OH_NN_AVG_POOL_PAD_MODE	Used when the tensor is used as the pad_mode parameter of the AvgPool operator.
OH_NN_AVG_POOL_PAD	Used when the tensor is used as the pad parameter of the AvgPool operator.
OH_NN_AVG_POOL_ACTIVATION_TYPE	Used when the tensor is used as the activation_type parameter of the AvgPool operator.
OH_NN_BATCH_NORM_EPSILON	Used when the tensor is used as the eosilon parameter of the BatchNorm operator.
OH_NN_BATCH_TO_SPACE_ND_BLOCKSIZE	Used when the tensor is used as the blockSize parameter of the BatchToSpaceND operator.
OH_NN_BATCH_TO_SPACE_ND_CROPS	Used when the tensor is used as the crops parameter of the BatchToSpaceND operator.
OH_NN_CONCAT_AXIS	Used when the tensor is used as the axis parameter of the Concat operator.
OH_NN_CONV2D_STRIDES	Used when the tensor is used as the strides parameter of the Conv2D operator.
OH_NN_CONV2D_PAD	Used when the tensor is used as the pad parameter of the Conv2D operator.
OH_NN_CONV2D_DILATION	Used when the tensor is used as the dilation parameter of the Conv2D operator.
OH_NN_CONV2D_PAD_MODE	Used when the tensor is used as the padMode parameter of the Conv2D operator.
OH_NN_CONV2D_ACTIVATION_TYPE	Used when the tensor is used as the activationType parameter of the Conv2D operator.
OH_NN_CONV2D_GROUP	Used when the tensor is used as the group parameter of the Conv2D operator.
OH_NN_CONV2D_TRANSPOSE_STRIDES	Used when the tensor is used as the strides parameter of the Conv2DTranspose operator.
OH_NN_CONV2D_TRANSPOSE_PAD	Used when the tensor is used as the pad parameter of the Conv2DTranspose operator.
OH_NN_CONV2D_TRANSPOSE_DILATION	Used when the tensor is used as the dilation parameter of the Conv2DTranspose operator.
OH_NN_CONV2D_TRANSPOSE_OUTPUT_PADDINGS	Used when the tensor is used as the outputPaddings parameter of the Conv2DTranspose operator.
OH_NN_CONV2D_TRANSPOSE_PAD_MODE	Used when the tensor is used as the padMode parameter of the Conv2DTranspose operator.
OH_NN_CONV2D_TRANSPOSE_ACTIVATION_TYPE	Used when the tensor is used as the activationType parameter of the Conv2DTranspose operator.
OH_NN_CONV2D_TRANSPOSE_GROUP	Used when the tensor is used as the group parameter of the Conv2DTranspose operator.
OH_NN_DEPTHWISE_CONV2D_NATIVE_STRIDES	Used when the tensor is used as the strides parameter of the DepthwiseConv2dNative operator.
OH_NN_DEPTHWISE_CONV2D_NATIVE_PAD	Used when the tensor is used as the pad parameter of the DepthwiseConv2dNative operator.
OH_NN_DEPTHWISE_CONV2D_NATIVE_DILATION	Used when the tensor is used as the dilation parameter of the DepthwiseConv2dNative operator.
OH_NN_DEPTHWISE_CONV2D_NATIVE_PAD_MODE	Used when the tensor is used as the padMode parameter of the DepthwiseConv2dNative operator.
OH_NN_DEPTHWISE_CONV2D_NATIVE_ACTIVATION_TYPE	Used when the tensor is used as the activationType parameter of the DepthwiseConv2dNative operator.
OH_NN_DIV_ACTIVATIONTYPE	Used when the tensor is used as the activationType parameter of the Div operator.
OH_NN_ELTWISE_MODE	Used when the tensor is used as the mode parameter of the Eltwise operator.
OH_NN_FULL_CONNECTION_AXIS	Used when the tensor is used as the axis parameter of the FullConnection operator.
OH_NN_FULL_CONNECTION_ACTIVATIONTYPE	Used when the tensor is used as the activationType parameter of the FullConnection operator.
OH_NN_MATMUL_TRANSPOSE_A	Used when the tensor is used as the transposeA parameter of the Matmul operator.
OH_NN_MATMUL_TRANSPOSE_B	Used when the tensor is used as the transposeB parameter of the Matmul operator.
OH_NN_MATMUL_ACTIVATION_TYPE	Used when the tensor is used as the activationType parameter of the Matmul operator.
OH_NN_MAX_POOL_KERNEL_SIZE	Used when the tensor is used as the kernel_size parameter of the MaxPool operator.
OH_NN_MAX_POOL_STRIDE	Used when the tensor is used as the stride parameter of the MaxPool operator.
OH_NN_MAX_POOL_PAD_MODE	Used when the tensor is used as the pad_mode parameter of the MaxPool operator.
OH_NN_MAX_POOL_PAD	Used when the tensor is used as the pad parameter of the MaxPool operator.
OH_NN_MAX_POOL_ACTIVATION_TYPE	Used when the tensor is used as the activation_type parameter of the MaxPool operator.
OH_NN_MUL_ACTIVATION_TYPE	Used when the tensor is used as the activationType parameter of the Mul operator.
OH_NN_ONE_HOT_AXIS	Used when the tensor is used as the axis parameter of the OneHot operator.
OH_NN_PAD_CONSTANT_VALUE	Used when the tensor is used as the constant_value parameter of the Pad operator.
OH_NN_SCALE_ACTIVATIONTYPE	Used when the tensor is used as the activationType parameter of the Scale operator.
OH_NN_SCALE_AXIS	Used when the tensor is used as the axis parameter of the Scale operator.
OH_NN_SOFTMAX_AXIS	Used when the tensor is used as the axis parameter of the Softmax operator.
OH_NN_SPACE_TO_BATCH_ND_BLOCK_SHAPE	Used when the tensor is used as the BlockShape parameter of the SpaceToBatchND operator.
OH_NN_SPACE_TO_BATCH_ND_PADDINGS	Used when the tensor is used as the Paddings parameter of the SpaceToBatchND operator.
OH_NN_SPLIT_AXIS	Used when the tensor is used as the Axis parameter of the Split operator.
OH_NN_SPLIT_OUTPUT_NUM	Used when the tensor is used as the OutputNum parameter of the Split operator.
OH_NN_SPLIT_SIZE_SPLITS	Used when the tensor is used as the SizeSplits parameter of the Split operator.
OH_NN_SQUEEZE_AXIS	Used when the tensor is used as the Axis parameter of the Squeeze operator.
OH_NN_STACK_AXIS	Used when the tensor is used as the Axis parameter of the Stack operator.
OH_NN_STRIDED_SLICE_BEGIN_MASK	Used when the tensor is used as the BeginMask parameter of the StridedSlice operator.
OH_NN_STRIDED_SLICE_END_MASK	Used when the tensor is used as the EndMask parameter of the StridedSlice operator.
OH_NN_STRIDED_SLICE_ELLIPSIS_MASK	Used when the tensor is used as the EllipsisMask parameter of the StridedSlice operator.
OH_NN_STRIDED_SLICE_NEW_AXIS_MASK	Used when the tensor is used as the NewAxisMask parameter of the StridedSlice operator.
OH_NN_STRIDED_SLICE_SHRINK_AXIS_MASK	Used when the tensor is used as the ShrinkAxisMask parameter of the StridedSlice operator.
OH_NN_SUB_ACTIVATIONTYPE	Used when the tensor is used as the ActivationType parameter of the Sub operator.
OH_NN_REDUCE_MEAN_KEEP_DIMS	Used when the tensor is used as the keep_dims parameter of the ReduceMean operator.
OH_NN_RESIZE_BILINEAR_NEW_HEIGHT	Used when the tensor is used as the new_height parameter of the ResizeBilinear operator.
OH_NN_RESIZE_BILINEAR_NEW_WIDTH	Used when the tensor is used as the new_width parameter of the ResizeBilinear operator.
OH_NN_RESIZE_BILINEAR_PRESERVE_ASPECT_RATIO	Used when the tensor is used as the preserve_aspect_ratio parameter of the ResizeBilinear operator.
OH_NN_RESIZE_BILINEAR_COORDINATE_TRANSFORM_MODE	Used when the tensor is used as the coordinate_transform_mode parameter of the ResizeBilinear operator.
OH_NN_RESIZE_BILINEAR_EXCLUDE_OUTSIDE	Used when the tensor is used as the exclude_outside parameter of the ResizeBilinear operator.
OH_NN_LAYER_NORM_BEGIN_NORM_AXIS	Used when the tensor is used as the beginNormAxis parameter of the LayerNorm operator.
OH_NN_LAYER_NORM_EPSILON	Used when the tensor is used as the epsilon parameter of the LayerNorm operator.
OH_NN_LAYER_NORM_BEGIN_PARAM_AXIS	Used when the tensor is used as the beginParamsAxis parameter of the LayerNorm operator.
OH_NN_LAYER_NORM_ELEMENTWISE_AFFINE	Used when the tensor is used as the elementwiseAffine parameter of the LayerNorm operator.
OH_NN_REDUCE_PROD_KEEP_DIMS	Used when the tensor is used as the keep_dims parameter of the ReduceProd operator.
OH_NN_REDUCE_ALL_KEEP_DIMS	Used when the tensor is used as the keep_dims parameter of the ReduceAll operator.
OH_NN_QUANT_DTYPE_CAST_SRC_T	Used when the tensor is used as the src_t parameter of the QuantDTypeCast operator.
OH_NN_QUANT_DTYPE_CAST_DST_T	Used when the tensor is used as the dst_t parameter of the QuantDTypeCast operator.
OH_NN_TOP_K_SORTED	Used when the tensor is used as the Sorted parameter of the Topk operator.
OH_NN_ARG_MAX_AXIS	Used when the tensor is used as the axis parameter of the ArgMax operator.
OH_NN_ARG_MAX_KEEPDIMS	Used when the tensor is used as the keepDims parameter of the ArgMax operator.
OH_NN_UNSQUEEZE_AXIS	Used when the tensor is used as the Axis parameter of the Unsqueeze operator.

Function Description

OH_NNCompilation_AddExtensionConfig()

OH_NN_ReturnCode OH_NNCompilation_AddExtensionConfig (OH_NNCompilation *compilation, const char *configName, const void *configValue, const size_t configValueSize )

Description

Adds extended configurations for custom device attributes.

Some devices have their own attributes, which have not been enabled in NNRt. This API helps you to set custom attributes for these devices. You need to obtain their names and values from the device vendor's documentation and add them to the model building instance. These attributes are passed directly to the device driver. If the device driver cannot parse the attributes, this API returns an error code.

After OH_NNCompilation_Build is called, configName and configValue can be released.

Since: 11

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
configName	Configuration name.
configValue	Configured value.
configValueSize	Size of the configured value, in bytes.

Returns

Execution result of the function. If the operation is successful, OH_NN_SUCCESS is returned. If the operation fails, an error code is returned. For details about the error codes, see OH_NN_ReturnCode.

OH_NNCompilation_Build()

OH_NN_ReturnCode OH_NNCompilation_Build (OH_NNCompilation *compilation)

Description

Performs model building.

After the build configuration is complete, call this API to start model building. The model building instance pushes the model and build options to the device for building. After this API is called, additional build operations cannot be performed. If OH_NNCompilation_SetDevice, OH_NNCompilation_SetCache, OH_NNCompilation_SetPerformanceMode, OH_NNCompilation_SetPriority, or OH_NNCompilation_EnableFloat16 is called, OH_NN_OPERATION_FORBIDDEN is returned.

Since: 9

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.

Returns

OH_NNCompilation_Construct()

OH_NNCompilation *OH_NNCompilation_Construct (const OH_NNModel *model)

Description

Creates a model building instance of the OH_NNCompilation type.

After the OH_NNModel module completes model construction, APIs provided by the OH_NNCompilation module pass the model to underlying device for building. This API creates an OH_NNCompilation instance based on the passed OH_NNModel instance. The OH_NNCompilation_SetDevice API is called to specify the device for model building, and the OH_NNCompilation_Build API is then called to complete model building.

In addition to computing device selection, the OH_NNCompilation module supports features such as model cache, performance preference, priority setting, and float16 computing, which can be implemented by the following APIs:

OH_NNCompilation_SetCache

OH_NNCompilation_SetPerformanceMode

OH_NNCompilation_SetPriority

OH_NNCompilation_EnableFloat16

After this API is called to create OH_NNCompilation, the OH_NNModel instance can be released.

Since: 9

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.

Returns

Pointer to the OH_NNCompilation instance. If the operation fails, NULL is returned.

OH_NNCompilation_ConstructForCache()

OH_NNCompilation *OH_NNCompilation_ConstructForCache ()

Description

Creates an empty model building instance for later recovery from the model cache.

For details about the model cache, see OH_NNCompilation_SetCache.

The time required for model recovery from the model cache is less than the time required for building using OH_NNModel.

Call OH_NNCompilation_SetCache or OH_NNCompilation_ImportCacheFromBuffer, and then call OH_NNCompilation_Build to complete model recovery.

Since: 11

Returns

Pointer to the OH_NNCompilation instance. If the operation fails, NULL is returned.

OH_NNCompilation_ConstructWithOfflineModelBuffer()

OH_NNCompilation *OH_NNCompilation_ConstructWithOfflineModelBuffer (const void *modelBuffer, size_t modelSize )

Description

Creates a model building instance based on the offline model buffer.

This API conflicts with the one for transferring the path of the online or offline model building file. You can select only one of the three build APIs.

Note: The returned OH_NNCompilation instance only saves the modelBuffer pointer in it, but does not copy its data. The modelBuffer should not be released before the OH_NNCompilation instance is destroyed.

Since: 11

Parameters

Name	Description
modelBuffer	Memory for storing offline model files.
modelSize	Memory size of the offline model.

Returns

Pointer to the OH_NNCompilation instance. If the operation fails, NULL is returned.

OH_NNCompilation_ConstructWithOfflineModelFile()

OH_NNCompilation *OH_NNCompilation_ConstructWithOfflineModelFile (const char *modelPath)

Description

Creates a model building instance based on an offline model file.

This API conflicts with the one for transferring the memory of the online or offline model building file. You can select only one of the three build APIs.

An offline model is a model type built offline by the model converter provided by the device vendor. Therefore, an offline model can be used only on a specified device. However, the build time of an offline model is usually far shorter than that of the image composition instance OH_NNModel.

During development, offline build needs to be performed and offline models need to be deployed in application packages.

Since: 11

Parameters

Name	Description
modelPath	Path of the offline model file.

Returns

Pointer to the OH_NNCompilation instance. If the operation fails, NULL is returned.

OH_NNCompilation_Destroy()

void OH_NNCompilation_Destroy (OH_NNCompilation **compilation)

Description

Destroys a model building instance of the OH_NNCompilation type.

This API needs to be called to destroy the model building instances created by calling OH_NNCompilation_Construct, OH_NNCompilation_ConstructWithOfflineModelFile, OH_NNCompilation_ConstructWithOfflineModelBuffer and OH_NNCompilation_ConstructForCache.

If compilation or *compilation is a null pointer, this API only prints warning logs but does not perform the destruction operation.

Since: 9

Parameters

Name	Description
compilation	Level-2 pointer to the OH_NNCompilation instance. After the model building instance is destroyed, this API sets *compilation to a null pointer.

OH_NNCompilation_EnableFloat16()

OH_NN_ReturnCode OH_NNCompilation_EnableFloat16 (OH_NNCompilation *compilation, bool enableFloat16 )

Description

Enables float16 for computing.

By default, the floating-point model uses float32 for computing. If this API is called on a device that supports float16, floating-point model that supports float32 will use float16 for computing, so to reduce memory usage and execution time.

This option is invalid for fixed-point models, for example, fixed-point models of the int8 type.

If this API is called on the device that does not support float16, the error code OH_NN_UNAVAILABLE_DEVICE is returned.

Since: 9

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
enableFloat16	Whether to enable float16. If this parameter is set to true, float16 inference is performed. If this parameter is set to false, float32 inference is performed.

Returns

OH_NNCompilation_ExportCacheToBuffer()

OH_NN_ReturnCode OH_NNCompilation_ExportCacheToBuffer (OH_NNCompilation *compilation, const void *buffer, size_t length, size_t *modelSize )

Description

Writes the model cache to the specified buffer.

For details about the model cache, see OH_NNCompilation_SetCache.

Note: The model cache is the build result OH_NNCompilation_Build. Therefore, this API must be called after OH_NNCompilation_Build.

Since: 11

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
buffer	Pointer to the given memory.
length	Memory length.
modelSize	Size of the model cache, in bytes.

Returns

OH_NNCompilation_ImportCacheFromBuffer()

OH_NN_ReturnCode OH_NNCompilation_ImportCacheFromBuffer (OH_NNCompilation *compilation, const void *buffer, size_t modelSize )

Description

Reads the model cache from the specified buffer.

For details about the model cache, see OH_NNCompilation_SetCache.

After calling OH_NNCompilation_ImportCacheFromBuffer, call OH_NNCompilation_Build to complete model recovery.

Note: The compilation instance stores the buffer pointer in the buffer, but does not copy its data. You cannot release the memory buffer before the compilation instance is destroyed.

Since: 11

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
buffer	Pointer to the given memory.
modelSize	Size of the model cache, in bytes.

Returns

OH_NNCompilation_SetCache()

OH_NN_ReturnCode OH_NNCompilation_SetCache (OH_NNCompilation *compilation, const char *cachePath, uint32_t version )

Description

Sets the cache directory and version for model building.

On the device that supports model caching, a model can be saved as a cache file after being built at the device driver layer. The model can be directly read from the cache file in the next build, saving the rebuild time. This API performs different operations based on the model cache directory and version:

If no file exists in the specified model cache directory, cache the built model to the directory and set the cache version to the value of version.
If a complete cached file exists in the specified model cache directory, and its version number is equal to version, read the cached file in the directory and pass it to the underlying device to convert it into an executable model instance.
If a complete cached file exists in the specified model cache directory, but its version is earlier than version, update the cached file. After the model is built in the underlying device, the cached file in the cache directory is overwritten and the version is updated to version.
If a complete cached file exists in the specified model cache directory, but its version is later than version, the cached file is not read and the error code OH_NN_INVALID_PARAMETER is returned.
If the cached file in the specified model cache directory is incomplete or you do not have the file access permission, the error code OH_NN_INVALID_FILE is returned.
If the model cache directory does not exist or you do not have the file access permission, the error code OH_NN_INVALID_PATH is returned.

Since: 9

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
cachePath	Directory for storing model cache files. This API creates model cache directories for different devices in the cachePath directory. You are advised to use a separate cache directory for each model.
version	Cached model version.

Returns

OH_NNCompilation_SetDevice()

OH_NN_ReturnCode OH_NNCompilation_SetDevice (OH_NNCompilation *compilation, size_t deviceID )

Description

Sets the device for model building and computing.

In the build phase, you need to specify the device for model building and computing. Call OH_NNDevice_GetAllDevicesID to obtain available device IDs. Then, call OH_NNDevice_GetType and OH_NNDevice_GetType to obtain device information and pass target device IDs to this API for setting.

Since: 9

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
deviceID	Device ID. If the value is 0, the first device in the current device list is used by default.

Returns

OH_NNCompilation_SetPerformanceMode()

OH_NN_ReturnCode OH_NNCompilation_SetPerformanceMode (OH_NNCompilation *compilation, OH_NN_PerformanceMode performanceMode )

Description

Sets the performance mode for model computing.

NNRt allows you to set the performance mode for model computing to meet the requirements of low power consumption and ultimate performance. If this API is not called to set the performance mode in the build phase, the model building instance assigns the OH_NN_PERFORMANCE_NONE mode for the model by default. In this case, the device performs computing in the default performance mode.

If this API is called on a device that does not support setting of the performance mode, the error code OH_NN_UNAVAILABLE_DEVICE is returned.

Since: 9

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
performanceMode	Performance mode for model computing. For details, see OH_NN_PerformanceMode.

Returns

OH_NNCompilation_SetPriority()

OH_NN_ReturnCode OH_NNCompilation_SetPriority (OH_NNCompilation *compilation, OH_NN_Priority priority )

Description

Sets the priority for model computing.

NNRt allows you to set computing priorities for models. The priorities apply only to models created by the process with the same UID. The settings will not affect models created by processes with different UIDs on different devices.

If this API is called on a device that does not support priority setting, the error code OH_NN_UNAVAILABLE_DEVICE is returned.

Since: 9

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.
priority	Priority for model computing. For details, see OH_NN_Priority.

Returns

OH_NNDevice_GetAllDevicesID()

OH_NN_ReturnCode OH_NNDevice_GetAllDevicesID (const size_t **allDevicesID, uint32_t *deviceCount )

Description

Obtains the ID of the device connected to NNRt.

Each device has a unique and fixed ID, which is returned through a uin32_t array.

When device IDs are returned through the size_t array, each element of the array is the ID of a single device. Internal managment is used for array memory. The data pointer remains valid before this API is called next time.

Since: 9

Parameters

Name	Description
allDevicesID	Pointer to the size_t array. The input allDevicesID must be a null pointer. Otherwise, the error code OH_NN_INVALID_PARAMETER* is returned.
deviceCount	Pointer of the uint32_t type, which is used to return the length of *allDevicesID.

Returns

OH_NNDevice_GetName()

OH_NN_ReturnCode OH_NNDevice_GetName (size_t deviceID, const char **name )

Description

Obtains the name of the specified device.

deviceID specifies the device ID used to obtain the device name. You can obtain the device ID by calling OH_NNDevice_GetAllDevicesID. If the value of deviceID is 0, the first device in the device list is used by default.

*name is a C-style string ended with '\0'.

*name must be a null pointer. Otherwise, the error code OH_NN_INVALID_PARAMETER is returned. For example, you should define char* deviceName = NULL, and then pass &deviceName as an input parameter.

Since: 9

Parameters

Name	Description
deviceID	Device ID. If the value of deviceID is 0, the first device in the device list is used by default.
name	Pointer to the char array, which saves the returned device name.

Returns

OH_NNDevice_GetType()

OH_NN_ReturnCode OH_NNDevice_GetType (size_t deviceID, OH_NN_DeviceType *deviceType )

Description

Obtains the type of the specified device.

deviceID specifies the device ID used to obtain the device type. If the value of deviceID is 0, the first device in the device list is used by default. Currently, the following device types are supported:

- - OH_NN_CPU: CPU device.
- - OH_NN_GPU: GPU device.
- - OH_NN_ACCELERATOR: machine learning dedicated accelerator.
- - OH_NN_OTHERS: other device types.

Since: 9

Parameters

Name	Description
deviceID	Device ID. If the value of deviceID is 0, the first device in the device list is used by default.
deviceType	Pointer to the OH_NN_DeviceType instance. The device type information is returned.

Returns

OH_NNExecutor_AllocateInputMemory()^(deprecated)

OH_NN_Memory *OH_NNExecutor_AllocateInputMemory (OH_NNExecutor *executor, uint32_t inputIndex, size_t length )

Description

Applies for shared memory for a single model input on the device.

NNRt provides an API for proactively applying for shared memory on a device. Based on the specified executor and input index value, this API applies for the shared memory whose size is length on the device associated with a single input. Then, it returns the shared memory through the OH_NN_Memory instance.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNTensor_CreateWithSize() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
inputIndex	Input index value, which complies with the data input sequence for calling OH_NNModel_SpecifyInputsAndOutputs. Assume that inputIndices is {1, 5, 9} when OH_NNModel_SpecifyInputsAndOutputs is called. When you apply for the input memory, set the value of this parameter to {0, 1, 2}.
length	Memory size to be applied for, in bytes.

Returns

Pointer to the OH_NN_Memory instance. If the operation fails, NULL is returned.

OH_NNExecutor_AllocateOutputMemory()^(deprecated)

OH_NN_Memory *OH_NNExecutor_AllocateOutputMemory (OH_NNExecutor *executor, uint32_t outputIndex, size_t length )

Description

Applies for shared memory for a single model output on the device.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNTensor_CreateWithSize() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
outputIndex	Output index value, index value, which complies with the data input sequence for calling OH_NNModel_SpecifyInputsAndOutputs. Assume that outputIndices is {4, 6, 8} when OH_NNModel_SpecifyInputsAndOutputs is called. When you apply for the output memory, set the value of this parameter to {0, 1, 2}.
length	Memory size to be applied for, in bytes.

Returns

Pointer to the OH_NN_Memory instance. If the operation fails, NULL is returned.

OH_NNExecutor_Construct()

OH_NNExecutor *OH_NNExecutor_Construct (OH_NNCompilation *compilation)

Description

Creates an OH_NNExecutor instance.

This API constructs a model inference executor for a device based on the specified OH_NNCompilation instance. Use OH_NNExecutor_SetInput to set the model input data. After the input data is set, call OH_NNExecutor_Run to perform inference and then call OH_NNExecutor_SetOutput to obtain the computing result.

After an OH_NNExecutor instance is created through the OH_NNCompilation instance, destroy the OH_NNCompilation instance if it is is no longer needed.

Since: 9

Parameters

Name	Description
compilation	Pointer to the OH_NNCompilation instance.

Returns

Pointer to the OH_NNExecutor instance. If the operation fails, NULL is returned.

OH_NNExecutor_CreateInputTensorDesc()

NN_TensorDesc *OH_NNExecutor_CreateInputTensorDesc (const OH_NNExecutor *executor, size_t index )

Description

Creates the description of an input tensor based on the specified index value.

The description contains all types of attribute values of the tensor. If the value of index reaches or exceeds the number of input tensors, this API returns an error code. You can obtain the number of input tensors by calling OH_NNExecutor_GetInputCount.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
index	Index value of the input tensor.

Returns

Pointer to the NN_TensorDesc instance. If the operation fails, NULL is returned.

OH_NNExecutor_CreateOutputTensorDesc()

NN_TensorDesc *OH_NNExecutor_CreateOutputTensorDesc (const OH_NNExecutor *executor, size_t index )

Description

Creates the description of an output tensor based on the specified index value.

The description contains all types of attribute values of the tensor. If the value of index reaches or exceeds the number of output tensors, this API returns an error code. You can obtain the number of output tensors by calling OH_NNExecutor_GetOutputCount.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
index	Index value of the output tensor.

Returns

Pointer to the NN_TensorDesc instance. If the operation fails, NULL is returned.

OH_NNExecutor_Destroy()

void OH_NNExecutor_Destroy (OH_NNExecutor **executor)

Description

Destroys an executor instance to release the memory occupied by it.

This API needs to be called to release the executor instance created by calling OH_NNExecutor_Construct. Otherwise, memory leak will occur.

If executor or *executor is a null pointer, this API only prints the warning log and does not execute the release logic.

Since: 9

Parameters

Name	Description
executor	Level-2 pointer to the OH_NNExecutor instance.

OH_NNExecutor_DestroyInputMemory()^(deprecated)

void OH_NNExecutor_DestroyInputMemory (OH_NNExecutor *executor, uint32_t inputIndex, OH_NN_Memory **memory )

Description

Releases the input memory pointed by the OH_NN_Memory instance.

This API needs to be called to release the memory instance created by calling OH_NNExecutor_AllocateInputMemory. Otherwise, memory leak will occur. The mapping between inputIndex and memory must be the same as that in memory instance creation.

If memory or *memory is a null pointer, this API only prints the warning log and does not execute the release logic.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNTensor_Destroy() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
inputIndex	Input index value, which complies with the data input sequence for calling OH_NNModel_SpecifyInputsAndOutputs. Assume that inputIndices is {1, 5, 9} when OH_NNModel_SpecifyInputsAndOutputs is called. When you release the input memory, set the value of this parameter to {0, 1, 2}.
memory	Level-2 pointer to the OH_NN_Memory instance. After the shared memory is released, this API sets *memory to a null pointer.

OH_NNExecutor_DestroyOutputMemory()^(deprecated)

void OH_NNExecutor_DestroyOutputMemory (OH_NNExecutor *executor, uint32_t outputIndex, OH_NN_Memory **memory )

Description

Releases the output memory pointed by the OH_NN_Memory instance.

This API needs to be called to release the memory instance created by calling OH_NNExecutor_AllocateOutputMemory. Otherwise, memory leak will occur. The mapping between outputIndex and memory must be the same as that in memory instance creation.

If memory or *memory is a null pointer, this API only prints the warning log and does not execute the release logic.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNTensor_Destroy() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
outputIndex	Output index value, index value, which complies with the data input sequence for calling OH_NNModel_SpecifyInputsAndOutputs. Assume that outputIndices is {4, 6, 8} when OH_NNModel_SpecifyInputsAndOutputs is called. When you release the output memory, set the value of this parameter to {0, 1, 2}.
memory	Level-2 pointer to the OH_NN_Memory instance. After the shared memory is released, this API sets *memory to a null pointer.

OH_NNExecutor_GetInputCount()

OH_NN_ReturnCode OH_NNExecutor_GetInputCount (const OH_NNExecutor *executor, size_t *inputCount )

Description

Obtains the number of input tensors.

You can obtain the number of input tensors from executor, and then use OH_NNExecutor_CreateInputTensorDesc to create a tensor description based on the specified tensor index.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
inputCount	Number of returned input tensors.

Returns

OH_NNExecutor_GetInputDimRange()

OH_NN_ReturnCode OH_NNExecutor_GetInputDimRange (const OH_NNExecutor *executor, size_t index, size_t **minInputDims, size_t **maxInputDims, size_t *shapeLength )

Description

Obtains the dimension range of all input tensors.

If the input tensor has a dynamic shape, the dimension range supported by the tensor may vary according to device. You can call this API to obtain the dimension range supported by the current device. *minInputDims saves the minimum dimension of the specified input tensor (the number of dimensions matches the shape), while *maxInputDims saves the maximum dimension. For example, if an input tensor has a dynamic shape of [-1, -1, -1, 3], *minInputDims may be [1, 10, 10, 3], and *maxInputDims may be [100, 1024, 1024, 3].

Note: If the value of index reaches or exceeds the number of output tensors, this API returns an error code. You can obtain the number of input tensors by calling OH_NNExecutor_GetInputCount.

As output parameters, *minInputDims and *maxInputDims cannot be null pointers. Otherwise, an error is returned. For example, you should define int32_t* minInDims = NULL, and then pass &minInDims as a parameter.

You do not need to release the memory of *minInputDims and *maxInputDims. It is released with executor.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
index	Index value of the input tensor.
minInputDims	Pointer to the returned array, which saves the minimum dimension of the specified input tensor (the number of dimensions matches the shape).
maxInputDims	Pointer to the returned array, which saves the maximum dimension of the specified input tensor (the number of dimensions matches the shape).
shapeLength	Number of dimensions of the returned input tensor, which is the same as the shape.

Returns

OH_NNExecutor_GetOutputCount()

OH_NN_ReturnCode OH_NNExecutor_GetOutputCount (const OH_NNExecutor *executor, size_t *outputCount )

Description

Obtains the number of output tensors.

You can obtain the number of output tensors from executor, and then use OH_NNExecutor_CreateOutputTensorDesc to create a tensor description based on the specified tensor index.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
outputCount	Number of returned output tensors.

Returns

OH_NNExecutor_GetOutputShape()

OH_NN_ReturnCode OH_NNExecutor_GetOutputShape (OH_NNExecutor *executor, uint32_t outputIndex, int32_t **shape, uint32_t *shapeLength )

Description

Obtains the dimension information about the output tensor.

You can use this API to obtain information about the specified output dimension and number of dimensions after a single inference is performed by calling OH_NNExecutor_Run. It is commonly used in dynamic shape input and output scenarios.

Note: If the value of outputIndex reaches or exceeds the number of output tensors, an error code is returned. You can obtain the number of output tensors by calling OH_NNExecutor_GetOutputCount.

As an output parameter, *shape cannot be a null pointer. Otherwise, an error is returned. For example, you should define int32_t* tensorShape = NULL and pass &tensorShape as a parameter.

You do not need to release the memory of shape. It is released with executor.

Since: 9

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
outputIndex	Output index value, which is the same as the sequence of the output data when OH_NNModel_SpecifyInputsAndOutputs is called. Assume that outputIndices is {4, 6, 8} when OH_NNModel_SpecifyInputsAndOutputs is called. When you obtain information about the output dimension, set the value of this parameter to {0, 1, 2}.
shape	Pointer to the int32_t array. The value of each element in the array is the length of the output tensor in each dimension.
shapeLength	Pointer to the uint32_t type. The number of output dimensions is returned.

Returns

OH_NNExecutor_Run()^(deprecated)

OH_NN_ReturnCode OH_NNExecutor_Run (OH_NNExecutor *executor)

Description

Executes model inference.

This API performs end-to-end model inference and computing on the device associated with the executor.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNExecutor_RunSync() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.

Returns

OH_NNExecutor_RunAsync()

OH_NN_ReturnCode OH_NNExecutor_RunAsync (OH_NNExecutor *executor, NN_Tensor *inputTensor[], size_t inputCount, NN_Tensor *outputTensor[], size_t outputCount, int32_t timeout, void *userData )

Description

Performs asynchronous inference.

You need to create the input and output tensors by calling OH_NNTensor_Create, OH_NNTensor_CreateWithSize, or OH_NNTensor_CreateWithFd.

Then, call OH_NNTensor_GetDataBuffer o obtain the pointer to the tensor data and copies the input data to it. The executor performs model inference, generates the inference result, and writes the result to the output tensor.

If the output tensor has a dynamic shape, you can obtain the actual shape of the output tensor by calling OH_NNExecutor_GetOutputShape. Alternatively, obtain the tensor description from the input tensor by calling OH_NNTensor_GetTensorDesc, and then obtain the actual shape by calling OH_NNTensorDesc_GetShape.

This API works in non-blocking mode and returns the result immediately after being called. You can obtain the inference result and execution return status through the NN_OnRunDone callback. If the device driver service stops abnormally during execution, you can use the NN_OnServiceDied callback for exception processing.

You can set the NN_OnRunDone and NN_OnServiceDied callbacks by calling OH_NNExecutor_SetOnRunDone and OH_NNExecutor_SetOnServiceDied.

If the inference times out, it is terminated immediately and the error code OH_NN_TIMEOUT is returned through the NN_OnRunDone callback.

userData is the identifier used to distinguish different asynchronous inferences and is returned as the first parameter in the callback. You can use any data that can distinguish different inferences as the identifier.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
inputTensor	Array of input tensors.
inputCount	Number of input tensors.
outputTensor	Array of output tensors.
outputCount	Number of output tensors.
timeout	Timeout interval of asynchronous inference, in ms.
userData	Identifier of asynchronous inference.

Returns

OH_NNExecutor_RunSync()

OH_NN_ReturnCode OH_NNExecutor_RunSync (OH_NNExecutor *executor, NN_Tensor *inputTensor[], size_t inputCount, NN_Tensor *outputTensor[], size_t outputCount )

Description

Performs synchronous inference.

You need to create the input and output tensors by calling OH_NNTensor_Create, OH_NNTensor_CreateWithSize, or OH_NNTensor_CreateWithFd. Then, call OH_NNTensor_GetDataBuffer o obtain the pointer to the tensor data and copies the input data to it. The executor performs model inference, generates the inference result, and writes the result to the output tensor.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
inputTensor	Array of input tensors.
inputCount	Number of input tensors.
outputTensor	Array of output tensors.
outputCount	Number of output tensors.

Returns

OH_NNExecutor_SetInput()^(deprecated)

OH_NN_ReturnCode OH_NNExecutor_SetInput (OH_NNExecutor *executor, uint32_t inputIndex, const OH_NN_Tensor *tensor, const void *dataBuffer, size_t length )

Description

Sets the data for a single model input.

This API copies the data whose length is specified by length (in bytes) in dataBuffer to the shared memory of the underlying device. inputIndex specifies the input to be set and tensor sets tensor information such as the shape, type, and quantization parameters.

NNRt supports models with dynamical shape input. For fixed shape input and dynamic shape input scenarios, this API uses different processing policies.

- Fixed shape input: The attributes of tensor must be the same as those of the tensor added by calling OH_NNModel_AddTensor in the build phase.
- Dynamic shape input: In the composition phase, because the shape is not fixed, each value in tensor.dimensions must be greater than 0 in the API calls to determine the shape input in the computing phase. When setting the shape, you can modify only the dimension whose value is -1. Assume that [-1, 224, 224, 3] is input as the the dimension of A in the composition phase. When this API is called, you can only change the size of the first dimension, for example, to [3, 224, 224, 3]. If other dimensions are adjusted, OH_NN_INVALID_PARAMETER is returned.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNExecutor_RunSync() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
inputIndex	Input index value, which complies with the data input sequence for calling OH_NNModel_SpecifyInputsAndOutputs. Assume that inputIndices is {1, 5, 9} when OH_NNModel_SpecifyInputsAndOutputs is called. When you set the input data, set the value of this parameter to {0, 1, 2}.
tensor	Tensor corresponding to the input data.
dataBuffer	Pointer to the input data.
length	Length of the data memory, in bytes.

Returns

OH_NNExecutor_SetInputWithMemory()^(deprecated)

OH_NN_ReturnCode OH_NNExecutor_SetInputWithMemory (OH_NNExecutor *executor, uint32_t inputIndex, const OH_NN_Tensor *tensor, const OH_NN_Memory *memory )

Description

Shared memory pointed by the OH_NN_Memory instance for a single model input.

In the scenario where the memory needs to be managed by yourself, this API binds the execution input to the OH_NN_Memory memory instance. During computing, the underlying device reads the input data from the shared memory pointed by the memory instance. By using this API, you can implement concurrent execution of input setting, computing, and read to improve the data flow inference efficiency.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNExecutor_RunSync() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
inputIndex	Input index value, which complies with the data input sequence for calling OH_NNModel_SpecifyInputsAndOutputs. Assume that inputIndices is {1, 5, 9} when OH_NNModel_SpecifyInputsAndOutputs is called. When you set the input shared memory, set the value of this parameter to {0, 1, 2}.
tensor	Pointer to OH_NN_Tensor, which is used to set the tensor corresponding to a single input.
memory	Pointer that points to OH_NN_Memory.

Returns

OH_NNExecutor_SetOnRunDone()

OH_NN_ReturnCode OH_NNExecutor_SetOnRunDone (OH_NNExecutor *executor, NN_OnRunDone onRunDone )

Description

Sets the callback processing function invoked when the asynchronous inference ends.

For the definition of the callback function, see NN_OnRunDone.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
onRunDone	Handle of the callback function NN_OnRunDone

Returns

OH_NNExecutor_SetOnServiceDied()

OH_NN_ReturnCode OH_NNExecutor_SetOnServiceDied (OH_NNExecutor *executor, NN_OnServiceDied onServiceDied )

Description

Sets the callback processing function invoked when the device driver service terminates unexpectedly during asynchronous inference.

For the definition of the callback function, see NN_OnServiceDied.

Since: 11

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
onServiceDied	Callback function handle NN_OnServiceDied.

Returns

OH_NNExecutor_SetOutput()^(deprecated)

OH_NN_ReturnCode OH_NNExecutor_SetOutput (OH_NNExecutor *executor, uint32_t outputIndex, void *dataBuffer, size_t length )

Description

Sets the memory for a single model output.

This method binds the buffer pointed by dataBuffer to the output specified by outputIndex. The length of the buffer is specified by length.

After OH_NNExecutor_Run is called to complete a single model inference, NNRt compares the length of the buffer pointed by dataBuffer with the length of the output data and returns different results based on the actual situation.

If the memory size is greater than or equal to the data length, the API copies the inference result to the memory and returns OH_NN_SUCCESS. You can read the inference result from dataBuffer.
If the memory size is less than the data length, this API returns the error code OH_NN_INVALID_PARAMETER and generates a log indicating that the memory size is too small.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNExecutor_RunSync() is recommended.

Parameters

Name	Description
executor	Pointer to the OH_NNExecutor instance.
outputIndex	Output index value, which is the same as the sequence of the output data when OH_NNModel_SpecifyInputsAndOutputs is called. Assume that outputIndices is {4, 6, 8} when OH_NNModel_SpecifyInputsAndOutputs is called. When you set the output memory, set the value of this parameter to {0, 1, 2}.
dataBuffer	Pointer to the output data.
length	Length of the data memory, in bytes.

Returns

OH_NNExecutor_SetOutputWithMemory()^(deprecated)

OH_NN_ReturnCode OH_NNExecutor_SetOutputWithMemory (OH_NNExecutor *executor, uint32_t outputIndex, const OH_NN_Memory *memory )

Description

Shared memory pointed by the OH_NN_Memory instance for a single model output.

In the scenario where the memory needs to be managed by yourself, this API binds the execution output to the OH_NN_Memory memory instance. During computing, the underlying device writes the computing result to the shared memory pointed by the memory instance. By using this API, you can implement concurrent execution of input setting, computing, and read to improve the data flow inference efficiency.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNExecutor_RunSync() is recommended.

Parameters

Name	Description
executor	Pointer to the executor.
outputIndex	Output index value, which is the same as the sequence of the output data when OH_NNModel_SpecifyInputsAndOutputs is called. Assume that outputIndices is {4, 6, 8} when OH_NNModel_SpecifyInputsAndOutputs is called. When you set the output shared memory, set the value of this parameter to {0, 1, 2}.
memory	Pointer that points to OH_NN_Memory.

Returns

OH_NNModel_AddOperation()

OH_NN_ReturnCode OH_NNModel_AddOperation (OH_NNModel *model, OH_NN_OperationType op, const OH_NN_UInt32Array *paramIndices, const OH_NN_UInt32Array *inputIndices, const OH_NN_UInt32Array *outputIndices )

Description

Adds an operator to a model instance.

You can use this API to add an operator to a model instance. The operator type is specified by op, and the operator parameters, inputs, and outputs are specified by paramIndices, inputIndices, and outputIndices respectively. This API verifies the attributes of operator parameters and the number of input and output parameters. These attributes must be correctly set when OH_NNModel_AddTensorToModel is called to add tensors. For details about the expected parameters, input attributes, and output attributes of each operator, see OH_NN_OperationType.

paramIndices, inputIndices, and outputIndices store index values of tensors. Index values are determined by the sequence in which tensors are added to the model. For details about how to add a tensor, see OH_NNModel_AddTensorToModel.

If unnecessary parameters are added for adding an operator, this API returns the error code OH_NN_INVALID_PARAMETER. If no operator parameter is set, the operator uses the default parameter value. For details about the default values, see OH_NN_OperationType.

Since: 9

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
op	Type of the operator to be added. For details, see OH_NN_OperationType.
paramIndices	Pointer to the OH_NN_UInt32Array instance, which is used to set the parameter tensor index of the operator.
inputIndices	Pointer to the OH_NN_UInt32Array instance, specifying the input tensor index of the operator.
outputIndices	Pointer to the OH_NN_UInt32Array instance, which is used to set the output tensor index of the operator.

Returns

OH_NNModel_AddTensor()^(deprecated)

OH_NN_ReturnCode OH_NNModel_AddTensor (OH_NNModel *model, const OH_NN_Tensor *tensor )

Description

Adds a tensor to a model instance.

The data node and operator parameters in the NNRt model are composed of tensors of the model. You can use this API to add tensors to a model instance based on the tensor parameter. The sequence of adding tensors is specified by the index value recorded in the model. The OH_NNModel_SetTensorData, OH_NNModel_AddOperation, and OH_NNModel_SpecifyInputsAndOutputs APIs specify tensors based on the index value.

NNRt supports input and output of dynamic shapes. When adding a data node with a dynamic shape, you need to set the dimensions that support dynamic changes in tensor.dimensions to -1. For example, if tensor.dimensions of a four-dimensional tensor is set to [1, -1, 2, 2], the second dimension supports dynamic changes.

Since: 9

Deprecated: This API is deprecated since API version 11.

Substitute: OH_NNModel_AddTensorToModel() is recommended.

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
tensor	Pointer to the OH_NN_Tensor tensor. The tensor specifies the attributes of the tensor added to the model instance.

Returns

OH_NNModel_AddTensorToModel()

OH_NN_ReturnCode OH_NNModel_AddTensorToModel (OH_NNModel *model, const NN_TensorDesc *tensorDesc )

Description

Adds a tensor to a model instance.

The data node and operator parameters in the NNRt model are composed of tensors of the model. This API adds tensors to a model instance based on NN_TensorDesc. The sequence of adding tensors is specified by the index value recorded in the model. The OH_NNModel_SetTensorData, OH_NNModel_AddOperation, and OH_NNModel_SpecifyInputsAndOutputs APIs specify tensors based on the index value.

Since: 9

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
tensorDesc	Pointer to NN_TensorDesc tensor, which specifies the attributes of the tensor added to the model instance.

Returns

OH_NNModel_Construct()

OH_NNModel *OH_NNModel_Construct (void )

Description

Creates a model instance of the OH_NNModel type and constructs a model instance by using the APIs provided by OH_NNModel.

Before composition, call OH_NNModel_Construct to create a model instance. Based on the model topology, call the OH_NNModel_AddTensorToModel, OH_NNModel_AddOperation, and OH_NNModel_SetTensorData APIs to fill in the data and operator nodes of the model, and then call OH_NNModel_SpecifyInputsAndOutputs to specify the input and output of the model. After the model topology is constructed, call OH_NNModel_Finish to build the model.

After a model instance is used, destroy it by calling OH_NNModel_Destroy to avoid memory leak.

Since: 9

Returns

Pointer to the OH_NNModel instance. If the operation fails, NULL is returned.

OH_NNModel_Destroy()

void OH_NNModel_Destroy (OH_NNModel **model)

Description

Destroys a model instance.

This API needs to be called to destroy the model instance created by calling OH_NNModel_Construct. Otherwise, memory leak will occur.

If model or *model is a null pointer, this API only prints warning logs but does not perform the destruction operation.

Since: 9

Parameters

Name	Description
model	Level-2 pointer to the OH_NNModel instance. After a model instance is destroyed, this API sets *model to a null pointer.

OH_NNModel_Finish()

OH_NN_ReturnCode OH_NNModel_Finish (OH_NNModel *model)

Description

Completes model composition.

After the model topology is set up, call this API to indicate that the composition is complete. After this API is called, additional image composition cannot be performed. If OH_NNModel_AddTensorToModel, OH_NNModel_AddOperation, OH_NNModel_SetTensorData, or OH_NNModel_SpecifyInputsAndOutputs is called, OH_NN_OPERATION_FORBIDDEN is returned.

Before calling OH_NNModel_GetAvailableOperations and OH_NNCompilation_Construct, you must call this API to complete composition.

Since: 9

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.

Returns

OH_NNModel_GetAvailableOperations()

OH_NN_ReturnCode OH_NNModel_GetAvailableOperations (OH_NNModel *model, size_t deviceID, const bool **isSupported, uint32_t *opCount )

Description

Checks whether all operators in a model are supported by the device. The result is indicated by a Boolean value.

Checks whether the underlying device supports operators in a model instance. The device is specified by deviceID, and the result is represented by the array pointed by isSupported. If the ith operator is supported, the value of (*isSupported)[i] is true. Otherwise, the value is false.

After this API is successfully executed, *(isSupported) points to the bool array that records the operator support status. The operator quantity for the array length is the same as that for the model instance. The memory corresponding to this array is managed by NNRt and is automatically destroyed after the model instance is destroyed or this API is called again.

Since: 9

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
deviceID	Device ID to be queried, which can be obtained by using OH_NNDevice_GetAllDevicesID.
isSupported	Pointer to the OHHDRMetaData array. The input *(isSupported) must be a null pointer. Otherwise, OH_NN_INVALID_PARAMETER** is returned.
opCount	Number of operators in a model instance, corresponding to the length of the *(isSupported)** array.

Returns

OH_NNModel_SetTensorData()

OH_NN_ReturnCode OH_NNModel_SetTensorData (OH_NNModel *model, uint32_t index, const void *dataBuffer, size_t length )

Description

Sets the tensor value.

For tensors with constant values (such as model weights), you need to use this API in the composition phase. The index value of a tensor is determined by the sequence in which the tensor is added to the model. For details about how to add a tensor, see OH_NNModel_AddTensorToModel.

Since: 9

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
index	Index value of a tensor.
dataBuffer	Pointer to the real data memory.
length	Length of the data memory.

Returns

OH_NNModel_SetTensorQuantParams()

OH_NN_ReturnCode OH_NNModel_SetTensorQuantParams (OH_NNModel *model, uint32_t index, NN_QuantParam *quantParam )

Description

Sets the quantization parameters of a tensor. For details, see NN_QuantParam.

Since: 11

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
index	Index value of a tensor.
quantParam	Pointer to NN_QuantParam.

Returns

OH_NNModel_SetTensorType()

OH_NN_ReturnCode OH_NNModel_SetTensorType (OH_NNModel *model, uint32_t index, OH_NN_TensorType tensorType )

Description

Sets the tensor type. For details, see OH_NN_TensorType.

Since: 11

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
index	Index value of a tensor.
tensorType	Tensor type.

Returns

OH_NNModel_SpecifyInputsAndOutputs()

OH_NN_ReturnCode OH_NNModel_SpecifyInputsAndOutputs (OH_NNModel *model, const OH_NN_UInt32Array *inputIndices, const OH_NN_UInt32Array *outputIndices )

Description

Sets an index value for the input and output tensors of a model.

A tensor must be specified as the end-to-end input and output of a model instance. After a tensor is set as the input or output tensor, you are not allowed to set the tensor data by calling OH_NNModel_SetTensorData. Instead, call OH_NNExecutor in the execution phase to set the input or output tensor data.

The index value of a tensor is determined by the sequence in which the tensor is added to the model. For details about how to add a tensor, see OH_NNModel_AddTensorToModel.

Currently, the model input and output cannot be set asynchronously.

Since: 9

Parameters

Name	Description
model	Pointer to the OH_NNModel instance.
inputIndices	Pointer to the OH_NN_UInt32Array instance, which is used to set the input tensor of the operator.
outputIndices	Pointer to the OH_NN_UInt32Array instance, which is used to set the output tensor of the operator.

Returns

OH_NNQuantParam_Create()

NN_QuantParam *OH_NNQuantParam_Create ()

Description

Creates an NN_QuantParam instance.

After creating an NN_QuantParam quantization parameter instance, set its attribute values by calling OH_NNQuantParam_SetScales, OH_NNQuantParam_SetZeroPoints, or OH_NNQuantParam_SetNumBits, and pass it to NN_Tensor by calling OH_NNModel_SetTensorQuantParams. Finally, call OH_NNQuantParam_Destroy to destroy it to avoid memory leakage.

Since: 11

Returns

Pointer to the NN_QuantParam instance. If the operation fails, NULL is returned.

OH_NNQuantParam_Destroy()

OH_NN_ReturnCode OH_NNQuantParam_Destroy (NN_QuantParam **quantParams)

Description

Destroys an NN_QuantParam instance.

After an NN_QuantParam instance is no longer needed after being passed to NN_Tensor, you need to destroy it to avoid memory leak.

If quantParams or *quantParams is a null pointer, this API only prints warning logs but does not perform the destruction operation.

Since: 11

Parameters

Name	Description
quantParams	Level-2 pointer to the NN_QuantParam instance.

Returns

OH_NNQuantParam_SetNumBits()

OH_NN_ReturnCode OH_NNQuantParam_SetNumBits (NN_QuantParam *quantParams, const uint32_t *numBits, size_t quantCount )

Description

Sets the number of quantization bits for an NN_QuantParam instance.

quantCount is the number of quantization parameters in the tensor. For example, for per-channel quantization, quantCount is the number of channels.

Since: 11

Parameters

Name	Description
quantParams	Pointer to the NN_QuantParam instance.
numBits	Array consisting of quantization bits of all quantization parameters in a tensor.
quantCount	Number of quantization parameters in a tensor.

Returns

OH_NNQuantParam_SetScales()

OH_NN_ReturnCode OH_NNQuantParam_SetScales (NN_QuantParam *quantParams, const double *scales, size_t quantCount )

Description

Sets the scaling coefficient for an NN_QuantParam instance.

quantCount is the number of quantization parameters in the tensor. For example, for per-channel quantization, quantCount is the number of channels.

Since: 11

Parameters

Name	Description
quantParams	Pointer to the NN_QuantParam instance.
scales	Array consisting of scaling coefficients of all quantization parameters in the tensor.
quantCount	Number of quantization parameters in a tensor.

Returns

OH_NNQuantParam_SetZeroPoints()

OH_NN_ReturnCode OH_NNQuantParam_SetZeroPoints (NN_QuantParam *quantParams, const int32_t *zeroPoints, size_t quantCount )

Description

Sets the zero point for an NN_QuantParam instance.

quantCount is the number of quantization parameters in the tensor. For example, for per-channel quantization, quantCount is the number of channels.

Since: 11

Parameters

Name	Description
quantParams	Pointer to the NN_QuantParam instance.
zeroPoints	Array consisting of zero points of all quantization parameters in a tensor.
quantCount	Number of quantization parameters in a tensor.

Returns

OH_NNTensor_Create()

NN_Tensor *OH_NNTensor_Create (size_t deviceID, NN_TensorDesc *tensorDesc )

Description

Creates an NN_Tensor instance from NN_TensorDesc.

This API uses OH_NNTensorDesc_GetByteSize to calculate the number of bytes of tensor data and allocate device memory for it. The device driver directly obtains tensor data in zero-copy mode.

Note: This API copies tensorDesc to NN_Tensor. Therefore, if tensorDesc is no longer needed, destroy it by calling OH_NNTensorDesc_Destroy.

If the tensor shape is dynamic, an error code is returned.

deviceID indicates the selected device. If the value is 0, the first device in the current device list is used by default.

tensorDesc is mandatory. If it is a null pointer, an error code is returned.

If the NN_Tensor instance is no longer needed, destroy it by calling OH_NNTensor_Destroy.

Since: 11

Parameters

Name	Description
deviceID	Device ID. If the value is 0, the first device in the current device list is used by default.
tensorDesc	Pointer to the NN_TensorDesc instance.

Returns

Pointer to the NN_Tensor instance. If the operation fails, NULL is returned.

OH_NNTensor_CreateWithFd()

NN_Tensor *OH_NNTensor_CreateWithFd (size_t deviceID, NN_TensorDesc *tensorDesc, int fd, size_t size, size_t offset )

Description

Creates an {@Link NN_Tensor} instance based on the specified file descriptor of the shared memory and NN_TensorDesc instance.

This API reuses the shared memory corresponding to fd, which may source from another NN_Tensor instance. When the tensor created by calling OH_NNTensor_Destroy is destroyed, the memory for storing the tensor data is not released.

Note: This API copies tensorDesc to NN_Tensor. Therefore, if tensorDesc is no longer needed, destroy it by calling OH_NNTensorDesc_Destroy.

deviceID indicates the selected device. If the value is 0, the first device in the current device list is used by default.

tensorDesc is mandatory. If the pointer is null, an error code is returned.

If the NN_Tensor instance is no longer needed, destroy it by calling OH_NNTensor_Destroy.

Since: 11

Parameters

Name	Description
deviceID	Device ID. If the value is 0, the first device in the current device list is used by default.
tensorDesc	Pointer to the NN_TensorDesc instance.
fd	fd of the shared memory to be used.
size	Size of the shared memory to be used.
offset	Offset of the shared memory to be used.

Returns

Pointer to the NN_Tensor instance. If the operation fails, NULL is returned.

OH_NNTensor_CreateWithSize()

NN_Tensor *OH_NNTensor_CreateWithSize (size_t deviceID, NN_TensorDesc *tensorDesc, size_t size )

Description

Creates an NN_Tensor instance based on the specified memory size and NN_TensorDesc instance.

This API uses size as the number of bytes of tensor data and allocates device memory to it. The device driver directly obtains tensor data in zero-copy mode.

Note that this API copies tensorDesc to NN_Tensor. Therefore, if tensorDesc is no longer needed, destroy it by calling OH_NNTensorDesc_Destroy.

deviceID indicates the ID of the selected device. If the value is 0, the first device is used.

tensorDesc is mandatory. If it is a null pointer, an error code is returned. The value of size must be greater than or equal to the number of bytes occupied by tensorDesc, which can be obtained by calling OH_NNTensorDesc_GetByteSize. Otherwise, an error code is returned. If the tensor shape is dynamic, size is not checked.

If the NN_Tensor instance is no longer needed, destroy it by calling OH_NNTensor_Destroy.

Since: 11

Parameters

Name	Description
deviceID	Device ID. If the value is 0, the first device in the current device list is used by default.
tensorDesc	Pointer to the NN_TensorDesc instance.
size	Size of the tensor data to be allocated.

Returns

Pointer to the NN_Tensor instance. If the operation fails, NULL is returned.

OH_NNTensor_Destroy()

OH_NN_ReturnCode OH_NNTensor_Destroy (NN_Tensor **tensor)

Description

Destroys an NN_Tensor instance.

If the NN_Tensor instance is no longer needed, call this API to destroy it. Otherwise, memory leak will occur.

If the tensor or *tensor pointer is null, this API returns an error code but does not perform the destruction operation.

Since: 11

Parameters

Name	Description
tensor	Level-2 pointer to the NN_Tensor instance.

Returns

OH_NNTensor_GetDataBuffer()

void *OH_NNTensor_GetDataBuffer (const NN_Tensor *tensor)

Description

Obtains the memory address of NN_Tensor data.

You can read/write data from/to tensor data memory. The data memory is mapped from the shared memory on the device. Therefore, the device driver can directly obtain tensor data in zero-copy mode.

Note: Only tensor data in the [offset, size) segment in the corresponding shared memory can be used. offset indicates the offset in the shared memory and can be obtained by calling OH_NNTensor_GetOffset. size indicates the total size of the shared memory, which can be obtained by calling OH_NNTensor_GetSize.

If the tensor pointer is null, a null pointer is returned.

Since: 11

Parameters

Name	Description
tensor	Pointer to the NN_Tensor instance.

Returns

Pointer to the tensor data memory. If the operation fails, a null pointer is returned.

OH_NNTensor_GetFd()

OH_NN_ReturnCode OH_NNTensor_GetFd (const NN_Tensor *tensor, int *fd )

Description

Obtains the file descriptor of the shared memory where NN_Tensor data is stored.

fd corresponds to a device shared memory and can be used by another NN_Tensor through OH_NNTensor_CreateWithFd.

If tensor or fd pointer is null, an error code is returned.

Since: 11

Parameters

Name	Description
tensor	Pointer to the NN_Tensor instance.
fd	fd of the shared memory.

Returns

OH_NNTensor_GetOffset()

OH_NN_ReturnCode OH_NNTensor_GetOffset (const NN_Tensor *tensor, size_t *offset )

Description

Obtains the offset of NN_Tensor data in the shared memory.

offset indicates the offset of tensor data in the corresponding shared memory. It can be used by another NN_Tensor together with fd and size of the shared memory through OH_NNTensor_CreateWithFd.

Note: Only tensor data in the [offset, size) segment in the shared memory corresponding to the fd can be used. offset indicates the offset in the shared memory and can be obtained by calling OH_NNTensor_GetOffset. size indicates the total size of the shared memory, which can be obtained by calling OH_NNTensor_GetSize.

If tensor or offset pointer is null, an error code is returned.

Since: 11

Parameters

Name	Description
tensor	Pointer to the NN_Tensor instance.
offset	Offset for the fd of the tensor.

Returns

OH_NNTensor_GetSize()

OH_NN_ReturnCode OH_NNTensor_GetSize (const NN_Tensor *tensor, size_t *size )

Description

Obtains the size of the shared memory where the NN_Tensor data is stored.

The value of size is the same as that of OH_NNTensor_CreateWithSize and OH_NNTensor_CreateWithFd. However, for a tensor created by using OH_NNTensor_Create, the value of size is equal to the number of bytes actually occupied by the tensor data, which can be obtained by calling OH_NNTensorDesc_GetByteSize.

If the tensor or size pointer is null, an error code is returned.

Since: 11

Parameters

Name	Description
tensor	Pointer to the NN_Tensor instance.
size	Size of the shared memory where the returned data is located.

Returns

OH_NNTensor_GetTensorDesc()

NN_TensorDesc *OH_NNTensor_GetTensorDesc (const NN_Tensor *tensor)

Description

Obtains an NN_TensorDesc instance of NN_Tensor.

You can use this API to obtain the pointer of the NN_TensorDesc instance of the specified NN_Tensor instance. You can obtain tensor attributes of various types from the returned NN_TensorDesc instance, such as the name, data format, data type, and shape.

You should not destroy the returned NN_TensorDesc instance because it points to an internal instance of NN_Tensor. Otherwise, once OH_NNTensor_Destroy is called, a crash may occur due to double memory release.

If the tensor pointer is null, a null pointer is returned.

Since: 11

Parameters

Name	Description
tensor	Pointer to the NN_Tensor instance.

Returns

Pointer to the NN_TensorDesc instance. If the operation fails, NULL is returned.

OH_NNTensorDesc_Create()

NN_TensorDesc *OH_NNTensorDesc_Create ()

Description

Creates an NN_TensorDesc instance.

NN_TensorDesc describes various tensor attributes, such as the name, data type, shape, and format.

You can create an NN_Tensor instance based on the input NNN_TensorDesc instance by calling the following APIs:

OH_NNTensor_Create

OH_NNTensor_CreateWithSize

OH_NNTensor_CreateWithFd

Note: This API copies the NN_TensorDesc instance to NN_Tensor. This way, you can create multiple NN_Tensor instances with the same NN_TensorDesc instance. If the NN_TensorDesc instance is no longer needed, destroy it by calling OH_NNTensorDesc_Destroy.

Since: 11

Returns

Pointer to the NN_TensorDesc instance. If the operation fails, NULL is returned.

OH_NNTensorDesc_Destroy()

OH_NN_ReturnCode OH_NNTensorDesc_Destroy (NN_TensorDesc **tensorDesc)

Description

Releases an NN_TensorDesc instance.

If the NN_TensorDesc instance is no longer needed, call this API to destroy it. Otherwise, memory leakage occurs.

If tensorDesc or *tensorDesc is a null pointer, an error code is returned but the object will not be destroyed.

Since: 11

Parameters

Name	Description
tensorDesc	Level-2 pointer to the NN_TensorDesc instance.

Returns

OH_NNTensorDesc_GetByteSize()

OH_NN_ReturnCode OH_NNTensorDesc_GetByteSize (const NN_TensorDesc *tensorDesc, size_t *byteSize )

Description

Obtains the number of bytes occupied by the tensor data obtained through calculation based on the shape and data type of an NN_TensorDesc instance.

The number of bytes occupied by data can be calculated based on the shape and data type of NN_TensorDesc by calling this API.

If the tensor shape is dynamically variable, this API returns an error code and byteSize is 0.

To obtain the number of elements in the tensor data, call OH_NNTensorDesc_GetElementCount.

If tensorDesc or byteSize is a null pointer, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
byteSize	Size of the returned data, in bytes.

Returns

OH_NNTensorDesc_GetDataType()

OH_NN_ReturnCode OH_NNTensorDesc_GetDataType (const NN_TensorDesc *tensorDesc, OH_NN_DataType *dataType )

Description

Obtains the data type of an NN_TensorDesc instance.

You can use this API to obtain the data type of a specified NN_TensorDesc instance.

If tensorDesc or dataType is a null pointer, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
dataType	Tensor data type.

Returns

OH_NNTensorDesc_GetElementCount()

OH_NN_ReturnCode OH_NNTensorDesc_GetElementCount (const NN_TensorDesc *tensorDesc, size_t *elementCount )

Description

Obtains the number of elements in an NN_TensorDesc instance.

You can use this API to obtain the number of elements in the specified NN_TensorDesc instance. To obtain the size of tensor data, call OH_NNTensorDesc_GetByteSize.

If the tensor shape is dynamically variable, this API returns an error code and elementCount is 0.

If tensorDesc or elementCount is a null pointer, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
elementCount	Number of elements.

Returns

OH_NNTensorDesc_GetFormat()

OH_NN_ReturnCode OH_NNTensorDesc_GetFormat (const NN_TensorDesc *tensorDesc, OH_NN_Format *format )

Description

Obtains the data format of an NN_TensorDesc instance.

You can use this API to obtain the data format (that is, NN_TensorDesc) of the specified NN_TensorDesc instance.

If tensorDesc or format is a null pointer, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
format	Tensor data format.

Returns

OH_NNTensorDesc_GetName()

OH_NN_ReturnCode OH_NNTensorDesc_GetName (const NN_TensorDesc *tensorDesc, const char **name )

Description

Obtains the name of an NN_TensorDesc instance.

You can use this API to obtain the name of the specified NN_TensorDesc instance. The value of *name is a C-style string ending with \0.

If tensorDesc or name is a null pointer, an error code is returned. As an output parameter, *name must be a null pointer. Otherwise, an error code is returned. For example, you should define char* tensorName = NULL and pass &tensorName as a parameter of name.

You do not need to release the memory of name. When tensorDesc is destroyed, it is automatically released.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
name	Tensor name.

Returns

OH_NNTensorDesc_GetShape()

OH_NN_ReturnCode OH_NNTensorDesc_GetShape (const NN_TensorDesc *tensorDesc, int32_t **shape, size_t *shapeLength )

Description

Obtains the shape of an NN_TensorDesc instance.

You can use this API to obtain the shape of the specified NN_TensorDesc instance.

If tensorDesc, shape, or shapeLength is a null pointer, an error code is returned. As an output parameter, *shape must be a null pointer. Otherwise, an error code is returned. For example, you should define int32_t* tensorShape = NULL and pass &tensorShape as a parameter of shape.

You do not need to release the memory of shape. When tensorDesc is destroyed, it is automatically released.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
shape	List of tensor shapes.
shapeLength	Length of the list of tensor shapes.

Returns

OH_NNTensorDesc_SetDataType()

OH_NN_ReturnCode OH_NNTensorDesc_SetDataType (NN_TensorDesc *tensorDesc, OH_NN_DataType dataType )

Description

Sets the data type of an NN_TensorDesc instance.

After an NN_TensorDesc instance is created, call this API to set the tensor data type.

If tensorDesc is a null pointer, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
dataType	Tensor data type to be set.

Returns

OH_NNTensorDesc_SetFormat()

OH_NN_ReturnCode OH_NNTensorDesc_SetFormat (NN_TensorDesc *tensorDesc, OH_NN_Format format )

Description

Sets the data format of an NN_TensorDesc instance.

After an NN_TensorDesc instance is created, call this API to set the data format of the tensor.

If tensorDesc is a null pointer, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
format	Tensor data format to be set.

Returns

OH_NNTensorDesc_SetName()

OH_NN_ReturnCode OH_NNTensorDesc_SetName (NN_TensorDesc *tensorDesc, const char *name )

Description

Sets the name of an NN_TensorDesc instance.

After an NN_TensorDesc instance is created, call this API to set the tensor name. The value of *name is a C-style string ending with \0.

If tensorDesc or name is a null pointer, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
name	Name of the tensor to be set.

Returns

OH_NNTensorDesc_SetShape()

OH_NN_ReturnCode OH_NNTensorDesc_SetShape (NN_TensorDesc *tensorDesc, const int32_t *shape, size_t shapeLength )

Description

Sets the data shape of an NN_TensorDesc instance.

After an NN_TensorDesc instance is created, call this API to set the tensor shape.

If tensorDesc or shape is a null pointer or shapeLength is 0, an error code is returned.

Since: 11

Parameters

Name	Description
tensorDesc	Pointer to the NN_TensorDesc instance.
shape	List of tensor shapes to be set.
shapeLength	Length of the list of tensor shapes.

Returns

NeuralNeworkRuntime

Overview

Summary

File

Structs

Types

Enums

Functions

Type Description

NN_OnRunDone

NN_OnServiceDied

NN_QuantParam

NN_Tensor

NN_TensorDesc

OH_NN_Memory(deprecated)

OH_NN_QuantParam(deprecated)

OH_NN_Tensor(deprecated)

OH_NN_UInt32Array

OH_NNCompilation

OH_NNExecutor

OH_NNModel

Enum Description

OH_NN_DataType

OH_NN_DeviceType

OH_NN_Format

OH_NN_FuseType

OH_NN_OperationType

OH_NN_PerformanceMode

OH_NN_Priority

OH_NN_ReturnCode

OH_NN_TensorType

Function Description

OH_NNCompilation_AddExtensionConfig()

OH_NNCompilation_Build()

OH_NNCompilation_Construct()

OH_NNCompilation_ConstructForCache()

OH_NNCompilation_ConstructWithOfflineModelBuffer()

OH_NNCompilation_ConstructWithOfflineModelFile()

OH_NNCompilation_Destroy()

OH_NNCompilation_EnableFloat16()

OH_NNCompilation_ExportCacheToBuffer()

OH_NNCompilation_ImportCacheFromBuffer()

OH_NNCompilation_SetCache()

OH_NNCompilation_SetDevice()

OH_NNCompilation_SetPerformanceMode()

OH_NNCompilation_SetPriority()

OH_NNDevice_GetAllDevicesID()

OH_NNDevice_GetName()

OH_NNDevice_GetType()

OH_NNExecutor_AllocateInputMemory()(deprecated)

OH_NNExecutor_AllocateOutputMemory()(deprecated)

OH_NNExecutor_Construct()

OH_NNExecutor_CreateInputTensorDesc()

OH_NNExecutor_CreateOutputTensorDesc()

OH_NNExecutor_Destroy()

OH_NNExecutor_DestroyInputMemory()(deprecated)

OH_NNExecutor_DestroyOutputMemory()(deprecated)

OH_NNExecutor_GetInputCount()

OH_NNExecutor_GetInputDimRange()

OH_NNExecutor_GetOutputCount()

OH_NNExecutor_GetOutputShape()

OH_NNExecutor_Run()(deprecated)

OH_NNExecutor_RunAsync()

OH_NNExecutor_RunSync()

OH_NNExecutor_SetInput()(deprecated)

OH_NNExecutor_SetInputWithMemory()(deprecated)

OH_NNExecutor_SetOnRunDone()

OH_NNExecutor_SetOnServiceDied()

OH_NNExecutor_SetOutput()(deprecated)

OH_NNExecutor_SetOutputWithMemory()(deprecated)

OH_NNModel_AddOperation()

OH_NNModel_AddTensor()(deprecated)

OH_NNModel_AddTensorToModel()

OH_NNModel_Construct()

OH_NNModel_Destroy()

OH_NNModel_Finish()

OH_NNModel_GetAvailableOperations()

OH_NNModel_SetTensorData()

OH_NNModel_SetTensorQuantParams()

OH_NNModel_SetTensorType()

OH_NN_Memory^(deprecated)

OH_NN_QuantParam^(deprecated)

OH_NN_Tensor^(deprecated)

OH_NNExecutor_AllocateInputMemory()^(deprecated)

OH_NNExecutor_AllocateOutputMemory()^(deprecated)

OH_NNExecutor_DestroyInputMemory()^(deprecated)

OH_NNExecutor_DestroyOutputMemory()^(deprecated)

OH_NNExecutor_Run()^(deprecated)

OH_NNExecutor_SetInput()^(deprecated)

OH_NNExecutor_SetInputWithMemory()^(deprecated)

OH_NNExecutor_SetOutput()^(deprecated)

OH_NNExecutor_SetOutputWithMemory()^(deprecated)

OH_NNModel_AddTensor()^(deprecated)