Has anyone got any positive (not it may or should work) feedback with BOINC given the large amount of GPU cores on this Jetson board.

Has anyone tested BOINC to see whether it sees and uses all available cores for processing work units?

Thanks in advance.

It is the science applications that use the cores on any processor - all BOINC does is detect them, along with any required drivers.

I should have been clearer, yes the detection was what i was after. thanks.

The Maxwell iGPU in the Nano is fairly limited. The stock OS does not support OpenCL so the stock OpenCL applications will not run. It however will run CUDA applications. If you use the SDK of the default Nano installation, you can compile your own CUDA science applications from source. I have a Einstein gpu application running along with a Seti gpu application running on my Nano.
https://einsteinathome.org/host/12775352
https://setiathome.berkeley.edu/show_host_detail.php?hostid=8707387
The Einstein application is a cpu application compiled to run on the gpu. Which runs fine with no issues.
The Seti application is an early variant of Petri's special sauce from the zi3v code branch. Which produces a high number of invalids because the app often under reports the pulse count compared to the cpu app.