File SNPE.h

↰ Parent directory (SNPE)

Definition (SNPE/SNPE.h)

Includes

Functions

Full File Listing

Typedefs

typedef void *Snpe_SNPE_Handle_t

A typedef to indicate a SNPE handle

Functions

Snpe_ErrorCode_t Snpe_SNPE_Delete(Snpe_SNPE_Handle_t snpeHandle)

Destroys/frees a SNPE object

Parameters

snpeHandle – [in] Handle to access the SNPE object

Returns

SNPE_SUCCESS if Delete operation successful.

Snpe_StringList_Handle_t Snpe_SNPE_GetInputTensorNames(Snpe_SNPE_Handle_t snpeHandle)

Gets the names of input tensors to the network.

To support multiple input scenarios, where multiple tensors are passed through execute() in a TensorMap, each tensor needs to be uniquely named. The names of tensors can be retrieved through this function.

In the case of a single input, one name will be returned.

Parameters

snpeHandle – [in] Handle to access the SNPE object

Returns

A StringList of input tensor names.

Snpe_StringList_Handle_t Snpe_SNPE_GetOutputTensorNames(Snpe_SNPE_Handle_t snpeHandle)

Gets the names of output tensors to the network.

Parameters

snpeHandle – [in] Handle to access the SNPE object

Returns

List of output tensor names.

Snpe_StringList_Handle_t Snpe_SNPE_GetOutputTensorNamesByLayerName(Snpe_SNPE_Handle_t snpeHandle, const char *name)

Gets the names of output tensor from the input layer name.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • name – [in] Layer name

Returns

Output tensor names.

Snpe_ErrorCode_t Snpe_SNPE_ExecuteITensors(Snpe_SNPE_Handle_t snpeHandle, Snpe_TensorMap_Handle_t inputHandle, Snpe_TensorMap_Handle_t outputHandle)

Processes the input data and returns the output.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • inputHandle – [in] A map of tensors that contains the input data for each input. The names of tensors needs to be matched with names retrieved through getInputTensorNames()

  • outputHandle – [inout] An empty map of tensors that will contain the output data of potentially multiple layers (the key in the map is the layer name) upon return

Returns

SNPE_SUCCESS upon successful execution

Note

output TensorMap has to be empty. To forward propagate and get results in user-supplied tensors, use Snpe_SNPE_ExecuteUserBuffers().

Snpe_ErrorCode_t Snpe_SNPE_ExecuteITensor(Snpe_SNPE_Handle_t snpeHandle, Snpe_ITensor_Handle_t inputHandle, Snpe_TensorMap_Handle_t outputHandle)

Processes the input data and returns the output.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • inputHandle – [in] A single tensor contains the input data.

  • outputHandle – [inout] An empty map of tensors that will contain the output data of potentially multiple layers (the key in the map is the layer name) upon return

Returns

SNPE_SUCCESS upon successful execution

Note

output TensorMap has to be empty. To forward propagate and get results in user-supplied tensors, use Snpe_SNPE_ExecuteUserBuffers.

Snpe_ErrorCode_t Snpe_SNPE_ExecuteUserBuffers(Snpe_SNPE_Handle_t snpeHandle, Snpe_UserBufferMap_Handle_t inputHandle, Snpe_UserBufferMap_Handle_t outputHandle)

Processes the input data and returns the output, using user-supplied buffers.

Caller must guarantee that for the duration of execute(), the buffer stored in UserBuffer would remain valid. For more detail on buffer ownership and lifetime requirements, please refer to zdl::DlSystem::UserBuffer documentation.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • inputHandle – [in] A map of UserBuffers that contains the input data for each input. The names of UserBuffers needs to be matched with names retrieved through getInputTensorNames()

  • outputHandle – [inout] A map of UserBuffers that will hold the output data of potentially multiple layers (the key in the map is the UserBuffer name)

Returns

SNPE_SUCCESS upon successful execution

Note

input and output UserBuffer maps must be fully pre-populated. with dimensions matching what the network expects. For example, if there are 5 output UserBuffers they all have to be present in map.

Snpe_ErrorCode_t Snpe_SNPE_RegisterIonBuffers(Snpe_SNPE_Handle_t snpeHandle, Snpe_UserMemoryMap_Handle_t ionBufferMapHandle)

Register Client ION Buffers.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • ionBufferMapHandle – [in] A UserMemoryMap of virtual addresses

Returns

SNPE_SUCCESS upon successful ION Buffer registration

Note

To be deprecated, please use new api Snpe_SNPE_RegisterUserMemoryMappedBuffers

Snpe_ErrorCode_t Snpe_SNPE_DeregisterIonBuffers(Snpe_SNPE_Handle_t snpeHandle, Snpe_StringList_Handle_t ionBufferNamesHandle)

Deregister Client ION Buffers.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • ionBufferNamesHandle – [in] A StringList of ION Buffer names

Returns

SNPE_SUCCESS upon successful ION Buffer deregistration

Note

To be deprecated, please use new api Snpe_SNPE_DeregisterUserMemoryMappedBuffers

Snpe_ErrorCode_t Snpe_SNPE_RegisterUserMemoryMappedBuffers(Snpe_SNPE_Handle_t snpeHandle, Snpe_UserMemoryMap_Handle_t bufferMapHandle)

Register Client Memory-Mapped Buffers (Example ION buffers in Android)

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • bufferMapHandle – [in] A UserMemoryMap of virtual addresses

Returns

SNPE_SUCCESS upon successful memory mapped buffer registration

Note

UserBuffer type passed for registration must match the data type of the tensor in the dlc For regular UserBuffers SNPE performs an online data conversion (quantization or dequantization etc). This is not possible for memory mapped buffers hence can lead to issues during execution or accuracy degradation

Snpe_ErrorCode_t Snpe_SNPE_DeregisterUserMemoryMappedBuffers(Snpe_SNPE_Handle_t snpeHandle, Snpe_StringList_Handle_t bufferNamesHandle)

Deregister Client Memory-Mapped Buffers (Example ION buffers in Android)

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • bufferNamesHandle – [in] A StringList of memory mapped buffer names

Returns

SNPE_SUCCESS upon successful memory mapped buffer deregistration

const char *Snpe_SNPE_GetModelVersion(Snpe_SNPE_Handle_t snpeHandle)

Returns the version string embedded at model conversion time.

Parameters

snpeHandle – [in] Handle to access the SNPE object

Returns

Model version string, which is a free-form string supplied at the time of the conversion

Snpe_TensorShape_Handle_t Snpe_SNPE_GetInputDimensions(Snpe_SNPE_Handle_t snpeHandle, const char *name)

Returns the dimensions of the input data to the model in the form of TensorShape. The dimensions in TensorShape corresponds to what the tensor dimensions would need to be for an input tensor to the model.

See also

Snpe_ITensor_Handle_t

See also

Snpe_TensorShape_Handle_t

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • name – [in] input name.

Returns

a TensorShape that maintains dimensions, matching the tensor dimensions for input to the model, where the last entry is the fastest varying dimension, etc.

Note

Note that this function only makes sense for networks that have a fixed input size. For networks in which the input size varies with each call of Execute(), this function should not be used.

Snpe_TensorShape_Handle_t Snpe_SNPE_GetInputDimensionsOfFirstTensor(Snpe_SNPE_Handle_t snpeHandle)

Returns the dimensions of the first input’s data to the model in the form of TensorShape. The dimensions in TensorShape corresponds to what the tensor dimensions would need to be for an input tensor to the model.

See also

Snpe_ITensor_Handle_t

See also

Snpe_TensorShape_Handle_t

Parameters

snpeHandle – [in] Handle to access the SNPE object

Returns

a TensorShape that maintains dimensions, matching the tensor dimensions for first input to the model, where the last entry is the fastest varying dimension, etc.

Note

Note that this function only makes sense for networks that have a fixed input size. For networks in which the input size varies with each call of Execute(), this function should not be used.

Snpe_StringList_Handle_t Snpe_SNPE_GetOutputLayerNames(Snpe_SNPE_Handle_t snpeHandle)

Gets the output layer(s) for the network.

Parameters

snpeHandle – [in] Handle to access the SNPE object

Returns

A StringList of output layer names.

Note

The output layers returned by this function may be different than those specified when the network was created via the SNPEBuilder. For example, if the network was created in debug mode with no explicit output layers specified, this will contain all layers.

Snpe_IBufferAttributes_Handle_t Snpe_SNPE_GetInputOutputBufferAttributes(Snpe_SNPE_Handle_t snpeHandle, const char *name)

Returns attributes of buffers used to feed input tensors and receive result from output tensors.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • name – [in] Tensor name.

Returns

BufferAttributes of input/output tensor named

Snpe_StringList_Handle_t Snpe_SNPE_GetInputTensorNamesForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName)

Gets the names of input tensors to the network.

To support multiple input scenarios, where multiple tensors are passed through execute() in a TensorMap, each tensor needs to be uniquely named. The names of tensors can be retrieved through this function.

In the case of a single input, one name will be returned.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

Returns

A StringList of input tensor names.

Note

The networkName is specified in snpe-dlc-info and defaults to the name of the first graph in the DLC.

Snpe_StringList_Handle_t Snpe_SNPE_GetOutputTensorNamesForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName)

Gets the names of output tensors to the network.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

Returns

List of output tensor names.

Snpe_StringList_Handle_t Snpe_SNPE_GetOutputTensorNamesByLayerNameForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, const char *name)

Gets the names of output tensor from the input layer name.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • name – [in] Layer name

Returns

Output tensor names.

Snpe_ErrorCode_t Snpe_SNPE_ExecuteITensorsForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, Snpe_TensorMap_Handle_t inputHandle, Snpe_TensorMap_Handle_t outputHandle)

Processes the input data and returns the output.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • inputHandle – [in] A map of tensors that contains the input data for each input. The names of tensors needs to be matched with names retrieved through getInputTensorNames()

  • outputHandle – [inout] An empty map of tensors that will contain the output data of potentially multiple layers (the key in the map is the layer name) upon return

Returns

SNPE_SUCCESS upon successful execution

Note

output TensorMap has to be empty. To forward propagate and get results in user-supplied tensors, use Snpe_SNPE_ExecuteUserBuffers().

Snpe_ErrorCode_t Snpe_SNPE_ExecuteITensorForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, Snpe_ITensor_Handle_t inputHandle, Snpe_TensorMap_Handle_t outputHandle)

Processes the input data and returns the output.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • inputHandle – [in] A single tensor contains the input data.

  • outputHandle – [inout] An empty map of tensors that will contain the output data of potentially multiple layers (the key in the map is the layer name) upon return

Returns

SNPE_SUCCESS upon successful execution

Note

output TensorMap has to be empty. To forward propagate and get results in user-supplied tensors, use Snpe_SNPE_ExecuteUserBuffers.

Snpe_ErrorCode_t Snpe_SNPE_ExecuteUserBuffersForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, Snpe_UserBufferMap_Handle_t inputHandle, Snpe_UserBufferMap_Handle_t outputHandle)

Processes the input data and returns the output, using user-supplied buffers.

Caller must guarantee that for the duration of execute(), the buffer stored in UserBuffer would remain valid. For more detail on buffer ownership and lifetime requirements, please refer to zdl::DlSystem::UserBuffer documentation.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • inputHandle – [in] A map of UserBuffers that contains the input data for each input. The names of UserBuffers needs to be matched with names retrieved through getInputTensorNames()

  • outputHandle – [inout] A map of UserBuffers that will hold the output data of potentially multiple layers (the key in the map is the UserBuffer name)

Returns

SNPE_SUCCESS upon successful execution

Note

input and output UserBuffer maps must be fully pre-populated. with dimensions matching what the network expects. For example, if there are 5 output UserBuffers they all have to be present in map.

Snpe_ErrorCode_t Snpe_SNPE_RegisterUserMemoryMappedBuffersForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, Snpe_UserMemoryMap_Handle_t bufferMapHandle)

Register Client Memory-Mapped Buffers (Example ION buffers in Android)

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • bufferMapHandle – [in] A UserMemoryMap of virtual addresses

Returns

SNPE_SUCCESS upon successful memory mapped buffer registration

Note

UserBuffer type passed for registration must match the data type of the tensor in the dlc For regular UserBuffers SNPE performs an online data conversion (quantization or dequantization etc). This is not possible for memory mapped buffers hence can lead to issues during execution or accuracy degradation

Snpe_ErrorCode_t Snpe_SNPE_DeregisterUserMemoryMappedBuffersForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, Snpe_StringList_Handle_t bufferNamesHandle)

Deregister Client Memory-Mapped Buffers (Example ION buffers in Android)

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • bufferNamesHandle – [in] A StringList of memory mapped buffer names

Returns

SNPE_SUCCESS upon successful memory mapped buffer deregistration

Snpe_TensorShape_Handle_t Snpe_SNPE_GetInputDimensionsForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, const char *name)

Returns the dimensions of the input data to the model in the form of TensorShape. The dimensions in TensorShape corresponds to what the tensor dimensions would need to be for an input tensor to the model.

See also

Snpe_ITensor_Handle_t

See also

Snpe_TensorShape_Handle_t

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • name – [in] input name.

Returns

a TensorShape that maintains dimensions, matching the tensor dimensions for input to the model, where the last entry is the fastest varying dimension, etc.

Note

Note that this function only makes sense for networks that have a fixed input size. For networks in which the input size varies with each call of Execute(), this function should not be used.

Snpe_TensorShape_Handle_t Snpe_SNPE_GetInputDimensionsOfFirstTensorForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName)

Returns the dimensions of the first input’s data to the model in the form of TensorShape. The dimensions in TensorShape corresponds to what the tensor dimensions would need to be for an input tensor to the model.

See also

Snpe_ITensor_Handle_t

See also

Snpe_TensorShape_Handle_t

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

Returns

a TensorShape that maintains dimensions, matching the tensor dimensions for first input to the model, where the last entry is the fastest varying dimension, etc.

Note

Note that this function only makes sense for networks that have a fixed input size. For networks in which the input size varies with each call of Execute(), this function should not be used.

Snpe_StringList_Handle_t Snpe_SNPE_GetOutputLayerNamesForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName)

Gets the output layer(s) for the network.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

Returns

A StringList of output layer names.

Note

The output layers returned by this function may be different than those specified when the network was created via the SNPEBuilder. For example, if the network was created in debug mode with no explicit output layers specified, this will contain all layers.

Snpe_IBufferAttributes_Handle_t Snpe_SNPE_GetInputOutputBufferAttributesForNetwork(Snpe_SNPE_Handle_t snpeHandle, const char *networkName, const char *name)

Returns attributes of buffers used to feed input tensors and receive result from output tensors.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • networkName – [in] Network name.

  • name – [in] Tensor name.

Returns

BufferAttributes of input/output tensor named

Snpe_IDiagLog_Handle_t Snpe_SNPE_GetDiagLogInterface_Ref(Snpe_SNPE_Handle_t snpeHandle)

Get the diagnostic logging interface

Parameters

snpeHandle – [in] Handle to access the SNPE object

Snpe_StringList_Handle_t Snpe_SNPE_GetNetworkNames(Snpe_SNPE_Handle_t snpeHandle)

Returns a stringList of network names managed by the snpeHandle.

Parameters

snpeHandle – [in] Handle to access the SNPE object

Returns

StringListHandle of networkNames

Snpe_ErrorCode_t Snpe_SNPE_SetPerformanceProfile(Snpe_SNPE_Handle_t snpeHandle, Snpe_PerformanceProfile_t performanceProfile)

Set Performance profile.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • performanceProfile – [in] Value of performance profile to be set

Returns

true if successfully applied

Snpe_ErrorCode_t Snpe_SNPE_SetCustomPerfProfile(Snpe_SNPE_Handle_t snpeHandle, Snpe_SNPEPerfProfile_Handle_t perfProfileHandle)

Set custom performance profile.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • perfProfileHandle – [in] Handle of perfProfile object

Returns

true if successfully applied

Snpe_ErrorCode_t Snpe_SNPE_SetExecutionPriorityHint(Snpe_SNPE_Handle_t snpeHandle, Snpe_ExecutionPriorityHint_t priority)

Sets a preference for execution priority.

Sets a preference for execution priority. Allows users to set the priority of the graph. Setting this option overwrites the previous priority. SNPE runtime is free to use this information to co-ordinate between different workloads that may or may not extend beyond SNPE.

Parameters

  • snpeHandle – [in] Handle to access the SNPE object

  • priority – [in] The target performance profile

Returns

true if successfully applied

Note

On the Android platform, performance is determined by the priority level. In contrast, on Windows, the operating system can adjust the priority level, which means that performance cannot be guaranteed.