So starting about a month ago, my computer started crashing with a dwmcore.dll crash, which in turn turned out to be caused by some unspecified hardware failure upon reading the system logs. Especially Victoria 3 seemed to reliably crash it sooner or later. Usually sooner.
So I try all the obvious stuff and the stuff from the answers on the internet. Disable undervolting the CPU, change the sound card, whatever. It becomes obvious that the GPU must be to blame, in SOME way, especially since throttling down the max temperature in the GForce Experience seems to lengthen the MTBF.
Well, long story short, it turns out the solution was to install the MSI Afterburner overclocking software and run a scan. You know, contrary to the advice to disable all overclocking. Turns out the GPU can go quite a bit higher than standard and still have cooling headroom, BUT past a point it also needs more millivolts. With the new curve that gives it more voltage, it's about as stable as the rock of Gibraltar.
So, yeah, quite unexpected that the solution was to overclock it instead of disabling overclocking.
I wonder if what changed about a month ago was the temperature in my room getting a couple of degrees lower, which allowed the drivers to push the card just that little bit higher...
Feel free to add your own anecdotes of stuff that went contrary to any sane expectation.
So I try all the obvious stuff and the stuff from the answers on the internet. Disable undervolting the CPU, change the sound card, whatever. It becomes obvious that the GPU must be to blame, in SOME way, especially since throttling down the max temperature in the GForce Experience seems to lengthen the MTBF.
Well, long story short, it turns out the solution was to install the MSI Afterburner overclocking software and run a scan. You know, contrary to the advice to disable all overclocking. Turns out the GPU can go quite a bit higher than standard and still have cooling headroom, BUT past a point it also needs more millivolts. With the new curve that gives it more voltage, it's about as stable as the rock of Gibraltar.
So, yeah, quite unexpected that the solution was to overclock it instead of disabling overclocking.
I wonder if what changed about a month ago was the temperature in my room getting a couple of degrees lower, which allowed the drivers to push the card just that little bit higher...
Feel free to add your own anecdotes of stuff that went contrary to any sane expectation.
Aucun commentaire:
Enregistrer un commentaire