Okay, there are two cases in which we’ve seen this:
- Following a system update where the Nvidia drivers or libraries were updated but not restarted. Restarting your machine should fix this, try this first.
- Starting in CUDA 10.1, Nvidia changed how some components of CUDA were stored on disk. Sometimes, this prevented cryoSPARC from finding those necessary components.
To fix #2, try my colleague’s instructions here: Patch CTF Job Fails: Child process with PID 31697 has terminated unexpectedly
Let me know how that goes.