> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/cudnn/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/cudnn/_mcp/server.

# Cross-Compiling cuDNN Samples

This section describes how to cross-compile cuDNN samples on Ubuntu.

## Linux arm64-SBSA

Follow the steps in this section to cross-compile cuDNN samples on Linux arm64-SBSA. Linux arm64-SBSA incorporates ARM based CPU cores for Server Base System Architecture (SBSA).

### Installing the CUDA Toolkit for Linux arm64-SBSA

Follow the instructions described in the [CUDA Installation Guide for Linux](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html) to install the CUDA Toolkit for the Cross SBSA variant of the toolkit.

### Installing cuDNN for Linux arm64-SBSA

1. Follow the instructions described in [Installing cuDNN on Linux](https://docs.nvidia.com/deeplearning/cudnn/backend/latest/api/overview.html#api-changes-900) to install cuDNN for the Cross SBSA variant of the library.
2. Install the arm64-SBSA host compiler.

   ```
   sudo apt-get -y install g++-aarch64-linux-gnu
   ```

### Cross-Compiling cuDNN Samples for Linux arm64-SBSA

1. Copy the `cudnn_samples_v9` directory to your home directory.

   ```
   cp -r /usr/src/cudnn_samples_v9 $HOME
   ```

2. For each sample, execute the following commands.

   ```
   cd $HOME/cudnn_samples_v9/(each sample)
   sudo make TARGET_ARCH=aarch64 SBSA=1
   ```

## Linux aarch64-Jetson

Follow the steps in this section to cross-compile cuDNN samples on Ubuntu. Linux aarch64-Jetson incorporates ARM based CPU cores for the Tegra platform.

### Installing the CUDA Toolkit for Linux aarch64-Jetson

Follow the instructions described in the [CUDA Installation Guide for Linux](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html) to install the CUDA Toolkit for the Cross Jetson variant of the toolkit.

### Installing cuDNN for Linux aarch64-Jetson

Follow the instructions described in [Installing cuDNN on Linux](https://docs.nvidia.com/deeplearning/cudnn/backend/latest/api/overview.html#api-changes-900) to install cuDNN for the Cross Jetson variant of the library.

### Cross-Compiling cuDNN Samples for Linux aarch64-Jetson

1. Copy the `cudnn_samples_v9` directory to your home directory.

   ```
   cp -r /usr/src/cudnn_samples_v9 $HOME
   ```

2. For each sample, execute the following commands.

   ```
   cd $HOME/cudnn_samples_v9/(each sample)
   sudo make TARGET_ARCH=aarch64
   ```