Jetson Setup

Jetson Setup#

Instructions to setup the Jetson Orin Nano Super Developer Kit 8GB DeveloperKit.

Flash the Board#

Flash the board using the SDK Manager on Ubuntu by following these instructions

Make sure to pay special attention to where the wire goes to launch into recovery mode and to uncheck Jetson Runtime Components.

I also recommend changing OEM Configuration from Runtime to Pre-Config and setting a username and password.

Software Update, Install, Configure#

Once the board is flashed and you’ve disconnected the recovery mode jumper, run the following to:

  1. Upgrade all installed apt packages

  2. Install Docker with the get docker convenience script

  3. Install and configure NVIDIA Container Toolkit

  4. Reboot

# Warning, this will execute a script from the Internet as sudo
sudo apt update && sudo apt upgrade -y && \
wget -qO- https://get.docker.com | sudo sh && \
sudo apt install nvidia-container-toolkit && \
sudo nvidia-ctk runtime configure --runtime=docker && reboot

Get Started with GPU#

To make sure your GPU is online and available on your Jetson, nvidia-smi will tell you about your GPU and CUDA version.

nvidia-smi

A good test of docker is to run a PyTorch container; if the GPU is available to the container’s Python environment it will output True; otherwise, False.

# Version compatible with Jetpack 6.2
sudo docker run -rm --runtime=nvidia --ipc=host nvcr.io/nvidia/pytorch:25.02-py3-igpu python -c "import torch; print(torch.cuda.is_available())"
Contents