Some of us have the nVidia Jetson systems, mostly for AI work but my two have LOTS of idle time. The TX2 and Nano units both have GPU's, take VERY little electricity to run (10 and 15 watts), and might be well suited to GPUGrid (not certain about that, though.) I run LHC@home on the Nano at the moment but they don't often have tasks for the ARM-based picocomputers. The TX2 has 6 processors while the Nano has 4 (and considerably faster than the RPi's.) :)
So, would it be convenient for someone to evaluate these systems, please?

Thanks,
David

Might be a challenge with the memory on the Jetson's.

I managed to make a Seti@home gpu app for the Maxwell core on my Nano. But it was the old zi3v app and I could never make the CUDA10 0.98 special app produce valid results. Just not enough memory or not enough CUDA cores.

Considering the acemd3 app needs the boinc wrapper first to be working and then run the acemd3 app inside the wrapper, I wonder if even the TX2 or above is capable.

I put the Nano /root on an external 256 GBUSB 3 SSD. On the SSD, I also set a 9 GB partition for swap. Having that kind of disk space and especially the swap opened up a whole new world of capability for the Nano.