Target library

Deep learning library to use during code generation for target environment

Model Configuration Pane: Code Generation / Interface

Description

The Target library parameter specifies the deep learning library the code generator uses to produce code for the target environment.

To enable this parameter, set the System target file parameter togrt.tlc or ert.tlc and the Language parameter to C++.

None (default) | MKL-DNN | ARM Compute | cuDNN | TensorRT

MKL-DNN

Generates code that uses the Intel^® Math Kernel Library for Deep Neural Networks (Intel MKL-DNN).

To enable this setting (or ARM Compute), clear the GPU acceleration parameter.

When select this setting, the Language standard parameter is set to C++11 (ISO).

ARM Compute

Generates code that uses the ARM^® Compute Library.

To enable this setting (or MKL-DNN), clear the GPU acceleration parameter.

cuDNN

Generates code that uses the CUDA^® Deep Neural Network library (cuDNN).

This setting requires a GPU Coder™ license.

To enable this setting (or TensorRT), select the GPU acceleration parameter.

TensorRT

Generates code that takes advantage of the NVIDIA^® TensorRT – high performance deep learning inference optimizer and run-time library.

This setting requires a GPU Coder license.

To enable this setting (or cuDNN, select the GPU acceleration parameter.

Parameter: DLTargetLibrary

Type: character vector

Value: 'None' | 'MKL-DNN' | 'arm-compute' | 'cuDNN' | 'TensorRT'

Default: 'None'

Introduced in R2020b