Nano and TX2 software compatibility?

Since I have both, I’m wondering how much compatibility there is between the two units? For example, now that BOINC runs on ARM’s, I have it running on the Nano because the electricity costs next to nothing. Granted it is 11 times slower than the desktop but still… :)
All the new AI software updates that are coming out is thrilling so I’d like to have both units be capable of the same tasks. My TX2 is still on 16.04 and I’ve not done much with it in a while. The Nano’s job will be to assist a drone in collision avoidance some of the time but when not flying, to play with other things. :) I’m a bit lost about updates on the TX2 versus what’s current on the Nano.

I have so much stuff now on both the TX2 and the Nano that I don’t even want to install new Jetpacks because of all the WORK I’d have to redo with the packages I’ve installed and the software development I’ve put into them. That is, until we can update the Jetpacks WITHOUT losing all the work that is in either unit.

And still looking for 20+ megapixel camera we can connect to the Nano.

The CPUs are the same architecture, and so no changes there. The Nano and TX1 GPUs have the Maxwell architecture (“-gencode compute_53,code=sm_53”), while the TX2 is Pascal (“-gencode compute_62,code=sm_62”). So you’d have to change or add the right GPU architecture to your build, but otherwise they should work if they are using the same CUDA release version.

You should also keep the boards at the same Jetpack level; that will make it easiest to reuse code between them.

Note that the TX2 has more capabilities; linuxdev mentioned the shader mode selector, but there’s also the two additional Carmel cores on the TX2, and the additional output modes for the video encoder.

Thanks, Snarky. I was not aware of those output mode differences. It’s been a while. :)