Jetson Nano EOL Compatibility Solutions

I just learned that the Jetson Nano is now “End-of-Life” on March 2022. So that’s why there are so many compatibility issues when installing TensorFlow or other packages.

Despite this, I still want to use the Nano for my project.

What can I do to help push the compatibility a little further? Like is it okay to upgrade to Ubuntu22.04, like in this article? Is it okay to force upgrade Python? And is it okay to install packages from the Jetson Orin or Jetson Xavier?


The main reason is that TensorFlow depends on many third-party libraries.
These libraries (e.g. NumPy) might update their version and break the compatibility.

To ensure compatibility, you might clone your system image and reflash Nano with that image to make sure all the dependencies are aligned.
If you want to move to a new version, either Python or a package, the compatibility issue might occur.

JetPack 5 for Orin/Xavier doesn’t support Nano, unfortunately, you cannot install them on Jetson Nano.


Thanks for reaching out, @AastaLLL!

Other than the default SD card image, where else could I find images that can support Nano?


You can use JetPack to flash and install the basic image.
If extra libraries are required (ex. PyTorch or TensorFlow), please check our NGC container which has the dependencies pre-installed.

Please download the container with the corresponding flag.
For example, can be used for JetPack 4.6.x.


Thanks, this helped solve most of my problems! I used the Tensorflow container Importing TensorFlow now sees Nano’s GPU as its additional physical devices.

Now all I need is to install the corresponding Python packages to the container. For future references, here are some instructions on configuring NGC containers: Containers For Deep Learning Frameworks User Guide - NVIDIA Docs


Thanks for the update.
Good to know it works now.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.