the above configuration caused both SuSE’s NVIDIA installer and manual efforts to get the system up an running to fail miserably.
Removing one board fixed all issues, we can access both boards after reinserting the second after setup. Manually patching xorg.conf to include the “BusID” in the DEVICE section of the board driving the display did the trick as well.
Not technically CUDA-related… maybe it helps nonetheless when someone sets up a custom “personal supercomputer”…