Skip to content

Hardware Manager

Flashing Jetson Nano Orin🔗

It is easiest to do this on native Ubuntu.

Follow this guide to install the Nvidia sdkmanager

You will have to create and login with a Nvidia developers account, then follow the instructions in the installer. Make sure to install with CUDA.

It appears that when in recovery mode the carrier board does not spin the fan on the Jetson module, use an external fan to cool the module otherwise it may power off due to overheating. The fan will spin once the OS installation has reached a certain point.

Setting Up Jetson🔗

Sometimes CUDA doesn't get installed?

sudo apt install cuda-toolkit-12-6

Install custom gstreamer to /build or somewhere and point GST_PLUGIN_PATH or else library issues!

Installing PyTorch🔗

Working method for Jetpack 6.2 and CUDA 12.6:

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/arm64/cuda-keyring_1.1-1_all.deb
sudo dpkg -i cuda-keyring_1.1-1_all.deb
sudo apt-get update
sudo apt-get -y install cusparselt-cuda-12
sudo apt-get -y install cudss
sudo apt-get install -y python3-pip libopenblas-dev libcusparselt-dev
sudo apt-get install -y cuda-cupti-12-6
python3 -m pip install torch==2.8.0 torchvision==0.23.0 --index-url=https://pypi.jetson-ai-lab.io/jp6/cu126

Sources🔗

Serial SSH🔗

ssh <user>@192.168.55.1

Flashing from Docker Container🔗

Potentially useful tips:

  • This must be done on a Linux machine (WSL2 might work according to the internet).
  • Add "nfs" to your supported file systems and kernel modules
  • Install udev in the container
  • QEMU with aarch64-linux support must be installed on the host machine. On NixOS install QEMU system wide and add "aarch64-linux" to boot.binfmt.emulatedSystems. https://wiki.nixos.org/wiki/QEMU
  • Run sudo update-binfmts --enable qemu-aarch64 in the container before launching the sdkmanager, you might have to install qemu-user-static before you run this (or after sdkmanger fails?).

Useful forum links: