Hi, everyone!
A few months ago I have purchased a used Jetson Nano 4G (not to be confused with the newer Orin model) for pennies and ever since have tried to the best of my ability to make it useful, i.e. not e-waste.
This machine runs Ubuntu 18.04 on Arm and Tegra. It can be upgraded to newer releases (forcefully), but seems useless outside AI vision stuff as it lacks any meaningful software support.
No HW acceleration in the browser or video playback (outside of barebones gstreamer). Ubuntu seems rather buggy with anything but Gnome and some audio related stuff just keeps breaking every few minutes.
There are some community guides on the official forum and some outside of it, but in general, you have to stick to 18.04 to get the bare minimum working and even then, the outdated browser makes most of the internet unusable.
I’ve tried to contact people who claimed to have made HW acceleration work in general, but no one responded and most links are dead because they are 3-4 years old.
Can anyone here point me towards some if ANY resources to make this little machine useful? I’ve read every single post on the forum, on jetsonhacks (mostly regurgatates other resources) and on github (Ubuntu 20.04 images fail for me).
They have images to flash to it. Just flash whatever they have, or build your own. Plenty of docs and guides out there, but that version of Ubuntu is way too old to make it very useful: https://developer.nvidia.com/embedded/learn/get-started-jetson-nano-devkit
They have one official image for the Nano. All others aren’t supported and cause issues unless you hold some packages (which is the method I tried to get 20.04 working). HW acceleration still won’t work.
I am looking for something beyond the official docs that would help.
Well, like I said, there are plenty of guides out there detailing how to make your own image, like this one
As far as your HW acceleration, I’d check to make sure which driver is actually loaded, and if it’s properly showing it loaded.
Tnx, I’ll give this one a shot.
Idk your background but this seems like a good time for a car metaphor:
“I just got this passenger side rear door to a 2006 Tahoe for super cheap and I want to make use of it so it doesn’t go to waste. The Chevy forums have some information on what models of the same year might mount it and what three letter codes correspond to the options on my door but nothing useful about mounting it to newer models. How can I make use of this thing?”
The Jetson boards are development kits intended for prototyping iot and edge ai devices. You could use your jetson board to prototype or even implement an edge AI device. Machine vision is a phenomenal way to use these smaller thingies like the old coral tpu.
Dev kits are often shipped with limited support and intended to provide the bare minimum to prove to an engineer that a particular chip or platform will work to some application. The rest is left up to the customer.
You don’t have an nvidia branded raspberry pi, you have the thing they send to palantirs hired geeks so they can map out the precise extent to which the panopticon can be weaponized. You could probably use it to sort and catalogue pictures of your pets and family.
I understand fully. When I stumbled upon it, I thought it was just their RPi clone with more powerful hardware (it’s even GPIO compatible). At that point I didn’t have much time to reaseach to what extent it was supported and just assumed at least some general functionallity since it came with Ubuntu and not some random embeded OS. I should have known better, but I do now.
What drew me to it was not the AI stuff, but rather the premise of a decent CPU and GPU package in a small form factor, ideal as an HTPC. And that it was just 7€. You can’t buy anything for that much money, even less a whole ass PC.