I've been having a host of issues with my PC in the last few days, and I don't know what really started it. The most major problems have been crashes during GPU-intensive games, and occasional extreme system slowdowns accompanied with surges in dwm.exe (Desktop Windows Manager) CPU consumption (often around 100x higher than usual).
I think I may have fixed the dwm.exe issue by removing and playing around with other components, but if it comes back, I'll let you guys know.
However, the crashes continue to happen. My AMD drivers are up to date, I've checked several times, clean-installed them again, and tried the previous version with no avail. The crashes specifically began to occur the night I downloaded a January 13 Battlefield 4 patch, but I doubt that's relevant because the crashes occur in multiple games. Oddly enough, it doesn't occur in Skyrim, which is CPU-heavy.
The crashes themselves appear in the Event Viewer as Event 41 Kernel-Power. The system abruptly restarts several minutes into gameplay with no error messages before or after rebooting, and thus far they happen only in graphically demanding games. The only signal I have before a crash is the noise level of the fans reaching their peak noise levels (at least that I've heard out of them). Haven't tried other stuff like 3D modelling or other applications.
This, and the fact that dwm.exe was slowing down my computer, leads me to believe that it's an issue with my month-old HD 7970 GPU or the amount of power it's drawing from my also-month-old Corsair AX860W 80-Plus Platinum PSU. The cables connecting the two seem fine and correctly inserted, not that I've touched them recently regardless, I never overclocked the GPU past 75 MHz (it's currently running at stock clock speeds), I tried allowing it a higher maximum power ceiling, and all my temperatures are fine. I really can't identify the problem. Help?
First-time system builder here, so sorry if I sound like an idiot or like I don't know what I'm saying, because I really don't. Thanks guys.
7970 stock speeds are 925mhz core clock and 5500mhz (effective).what model do you have?you may need to increase voltage to achieve a stable +75mhz bump.the o/c differs from gpu to gpu .some cards have it locked though. also monitor your cpu and gpu temps on idle and under load. with a psu of that caliber its anlikely for it to be faulty!
Alright, well I've increased the power setting in AMD Catalyst Control Centre, which increases the maximum voltage allowed, by +20%. The GPU is right now running at normal clock speeds. My idle temperatures for the system are usually under 40 degrees Celsius for both the whole system and CPU. Under the loads I usually encounter at the time of a crash, it's possibly 50 (though I've seen it go much higher than that when I didn't have these crashes). I'm not sure about my GPU's idle temp, but after a crash, it's in the mid-60's. So I don't think that should be anything to do with overheating. Hmm, maybe I'll try replacing the cable from the PSU to the GPU. If not, I mean, I can't imagine the PSU being a bottleneck in anyway, so would that be sufficient evidence to believe it's a GPU/AMD driver problem?
ok change the psu to gpu cables.reinstall drivers (13.12 the latest)and disable overclocking through ccc. run a bunch of stress tests like furmark and see what happens while monitoring gpu/cpu temps and load through gpu z of msi afterburner.have you o/ced the cpu too.?if yes apply stock speed to during benckmark. if problem persists then we are dealing with another problem(i supose you tower is properly ventilated, is it?)