If you’re looking for a lightweight alternative to the L4T image that nvidia provides you should check out the Xubuntu image i posted on the forum.
Performance is great and things like youtube playback are much smoother than the L4T image.
It’s based off a minimal Xubuntu-core/xfce installation.
Built from the ground up using debootstrap and has all the same functionality as the L4T image like auto resize as well as the jetson tools like jetson_clocks and nvpmodel.
image size is only 3.5GB.
just search the forum for Xubuntu.
I’m sure this image is good but I did spend a lot of time making sure things were running perfectly.
This was a while ago now but I did test the image, and even managed to get most of the nVidia SDK across to it and working to a degree. Surprisingly the thing I could not install were the Python wheels built for the Nano.
Things like TensorFlow and Torch. For some reason it was refusing to install the python wheels.
I don’t recall if I said but I grabbed all the special nVidia packages from the local repository in /var and forced them to work. Really all that’s needed is the debian-centric directories to be added to the path and the occasional symlink.
The new version of the packages with the kernel. You can try to install them with the command “dpkg” as a replacement for the old kernel. Please note, I have not tested these packages yet. In the near future I plan to release new images with this kernel.
A new designation of the version number for Armbian.
The current version Legacy kernel 4.4 is 19.11.3 (20191126).
All previous images are moved to the OLD directory.
For the starter kit, I expanded the list of available images.
Stretch
Buster
Xenial
Bionic
Disco
Eoan
Add content link for MEGA site. :)
p.s.
Due to the transition to a new format of image names and their significant expansion, I have postponed the addition of packages from the SDK and other changes.