Target library

Description

Specify the target deep learning library used during simulation. To enable this parameter, select C++ for the Language.

Category: Simulation Target

Settings

Default: MKL-DNN if GPU acceleration is off. cuDNN if GPU acceleration is on.

MKL-DNN

Use this option for simulation that uses the Intel^® Math Kernel Library for Deep Neural Networks (Intel MKL-DNN).

cuDNN

Use this option for simulation that uses the CUDA^® Deep Neural Network library (cuDNN).

TensorRT

Use this option for simulation that takes advantage of the NVIDIA^® TensorRT – high performance deep learning inference optimizer and run-time library.

Dependencies

• MKL-DNN is available when GPU acceleration on the Simulation Target pane is disabled and Language is set to C++.

• cuDNN or TensorRT requires a GPU Coder™ license.

• cuDNN or TensorRT is available when GPU acceleration on the Simulation Target pane is enabled.

Command-Line Information

Parameter: SimDLTargetLibrary

Type: character vector

Value: 'MKL-DNN' | 'cuDNN' | 'TensorRT'

Default: MKL-DNN

Related Topics

• Model Configuration Parameters: Simulation Target
• GPU Code Generation for Deep Learning Networks Using MATLAB Function Block (GPU Coder)