Cant find reliable stress tests, and all of them end up revealing that the next program manages to crash my card, making actual overclocking to be mess as I dont know whats crashing
Been ongoing problem since I got my Fury, and I have now Vega 64 with the same eccentric behavior
Latest little GPU crasher is Blender, as soon as I try to alter accretion disk around my black hole, whole machine freezes and Win10 bluescreens about GPU TDP, and I really really really dont want to play this game anymore, and its starting to make me get really fustrated, why cant this piece of shit just work with the overclocks on!
do any of them crash at stock clocks? i’ve seen on CPU and GPU, chips PCed passing 1 stress test and failing another, they all use different instruction sets and resources with in the CPU. to be truly stable on a CPU, i’ll run 50-70 hours on each stress program, i’ve had some fail after 30 hours 1 in program and not the other.
GPUs on the other hand, don’t need THAT stringent of testing, i’ve run GPUs that artifact in stress tests when OCed but, run most or all of my games fine with said OC. I’ve had vise versa, passes stress but, fails in applications.
I had to remodel my Vega 56 overclock/undervolt for Battlefield 5. If the core clocks aren’t perfectly stable it will just exit to desktop, which scared me at first because that can have other causes like bad RAM. Pushing HBM2 a bridge too far shows up as easy to spot artifacts well before it seizes up.