I’ve got a ReComputer J1020 - a Nano on a carrier board with all sorts of ports. It is a real shame that Nvidia aren’t updating the Jetpacks for these older devices.
I’m trying to install and use YOLOv8, and have run into various issues… First of all of course the 16GB SD card was running out of space. The J1020 has an M.2 slot, so I installed a 256GB SSD drive, and moved across to using that (using the instructions at Memory Expansion | Seeed Studio Wiki - genius!).
So now I can try different things and if they don’t work can switch back to the SD card, format the SSD again and start again. It’s a bit like Groundhog Day. Luxury!
I’ve currently got Ubuntu 20.04 and YOLOv8 installed. My next challenge is compiling PyTorch from source, and I think I’m going to have to compile TensorRT from source as well. I have plenty of computer and programming experience but not much with Linux, so it is a fairly steep but enjoyable learning curve.
Any advice would be welcome!
Thanks anyway for this thread - it shows that anything is possible, and has encouraged me to ‘give it a go’.
(Nvidia support haven’t responded yet to my thread YOLOv8 on J1020 (Nano) using TensorRT model )