I have cuda code that now locks up the GPU occasionally and when it does windows times out eventually and wands to create a core dump. The setting in the newest version of windows 10 to disable that is well hidden.
(Windows 10 is really a terrible OS for developers. windows 7 was so much better)
Is there a way to debug these kind of issues? If I got another card and used it for the primary display, would it do any good? or would windows just crash one driver crash it all anyway.
The reboot on this machine is long, the time for the core dump to move off 0 to 2% is over 5 min* so I have to power off hard.
- wth heck? why? does it try to make a copy of all ram? 384GB of ram 290GB virtual shared to card and 11GB on card.