We made some progress on the Cray at the museum today.
We suspected the reason the Cray I/O subsystem was unresponsive was dead NVRAM chips. We wound up configuring a new NVRAM chip with all the correct values in a separate machine of similar vintage (Sparcstation 1+) and then moving the chip over to the Themis VME board in the Cray.
It worked! We can now talk to the Cray I/O system.
Getting closer to booting the machine for real...
A screenshot of the Cray IO subsystem booting.
A densely packed circuit board from the 90s.
A circuit board being held in front of a Cray supercomputer cabinet.