I'm running Isabelle2024 on Windows 10. When using sledgehammer, some eprover processes do not exit after sledgehammer is completed. Each of these processes can consume gigabytes of memory (which grows even after sledgehammer exits) and stall my system eventually. This only happens when my theory is "complex" enough somehow.
The process tree of such eprover process looks like
Other ATP processes exit correctly, including the eprover called by zipperposition.
I can provide a process log if needed.
e
for sledgehammer because it tended to eat up my RAM. $\endgroup$