failing gpu?

Ok, so I downclocked to 675mhz and have been running msi kombuster (OC stress test full screen full resolution) for about an hour now. All seems good. I turned the fan up to 80% as I was leaving it unattended for a bit. Temps have peaked at 44c. Maybe I'll start working my way up in an attempt to trigger a fault. I did successfully run a preset benchmark on kombuster at 750mhz but didn't have time to fiddle around so I backed it off for the stress test.

Any thoughts?


EDIT:
this stress test is warming up my room quite nicely. I have the radiator turned off so it doesn't get too dry and stuffy which generally leaves it cold. If it wasn't for the extra energy costs i'd do this all winter haha
 
Last edited:
Ok so I decided if I'm gonna figure this out, I need to sacrifice a nights sleep. I took a 2 hour nap and have been playing with afterburner, kombuster, and occt for the past 4 hours.

I ran kombuster while asleep at all stock (factory overclock) no crashes and no artifacts upon my return. So I played with overclocking to see if GPU is really my issue. I had set fan control up so that temps hovered around 65Celsius to ensure that the card was being stressed at higher than real world use as I have never seen my card get hotter than 55 Celsius. I had no artifacts upwards of 850mhz/1900mhz GPU/memClock. So I downloaded occt to run test with error checking enabled. With GPU core at 725mhz I tried to find memClock ceiling. I got it up to 2000mhz (1800mhz stock) when I figured "that's high enough". Then I pushed GPU core up to 875 before needing voltage increase. Bumped voltage up from 950 to 975 and occt didn't throw any errors. Brought gpu core up to 900mhz, occt showed errors. So I bumped voltage up to 1000 and tried again. Within seconds occt threw over 100 errors, I quickly quit the test. At this point I tried something new, I reset gpu to default factory oc settings and upped the voltage to max allowed in afterburner (1087 iirc) ran occt and to my surprise was immediately met with over 1000 errors. Quickly quit the test and while resetting afterburner to stock settings my graphics driver crashed, albeit with successful recovery by Windows.

What does this all mean?

My deduction is that the graphics card is indeed dying or that it's not getting the power it is asking for. If the latter, then my psu must be either inadequate or failing correct? I don't think it would be inadequate as its a corsair 650w single rail. It is powering my AMD x2 555 with a slight voltage bump, running at 4ghz. My WD black 640gb hdd, an optical drive, external 500 gb WD passport, and the gpu. I really don't think all that would tax a quality 650w PSU. So should I look into a failing psu? I will edit this post soon with hwmonitor's voltage readings but they have never raised any red flags that I've noticed, and corsair is one of the most reputable psu makers out there.

Thanks again for all your help.

Edit:
HWMonitor Voltage ranges:
CPU vcore: 1.472 - 1.488
VIN1: 1.616 - 1.616
+3.3V: 3.360 - 3.360
+5V: 5.027 - 5.053
+12V: 12.160 - 12.224
-12V: -12.800 - -9.664
+5V VCCH: 3.629 - 3.629
VBAT: 3.280 - 3.280
CPU: 1.472 - 1.488
Nvidia VIN0: 0.950 - 0.950
 
Last edited:
If you maxed the voltage out without maxing the fan it probably shot the temps up on the VRM causing all of the artifacting. That happens a lot without having the reference cooler directly cooling everything with a huge plate. Your PSU voltages are fine.

Just need to leave your stuff at stock man and replace it soon. And another thing, when you stress test anything for stability ALL parts need to be stock.
 
Just need to leave your stuff at stock man and replace it soon.

That would be lovely If my computer wasn't freezing up regularly. This is what started me on this quest to find out wtf is going on with my computer. I was running the GPU at stock settings and the computer kept freezing. Thats when I tried singling out components with leads such as windows message saying graphics driver crashed and then trying to figure out as much as I can about whats going on. Hence my posting here :)

Update on situation:
Ran OCCT GPU stress test with error checking enabled for 8 hours with GPU at stock settings. No errors. I then closed OCCT, opened up battlefield 3 and played 2 rounds of gun master (approx 30min). Closed battlefield and opened up firefox. I browsed the interwebs for about 10 minutes, I had 3 tabs open and then the pc froze up again. Windows did not recover, I had to do a hard reset.

Anybody think that a clean Win7 install could help this situation? Any other thoughts as to what may be happening here?
 
Hello again :)
Before your suggestion of Prime95 I had reduced my CPU overclock from 4.0ghz to 3.6ghz (stock is 3.2). I left the +.05 voltage bump where it was. So I ran Prime95 for 5 hours, all was good. I then brought my CPU OC back up to 4.0ghz and ran prime95 blend for 2 hours, and then the max heat/max power option for an hour. all stable. I will continue to do more stability testing in the coming days. Such as memtest, and longer runs of prime95. I will be sure to let you all know if something comes up. GPU is currently at stock settings.

I did have another thought though. About the same time that I moved my PC up into the house, I had gotten a webcam, a logitech c510 I think. I often leave it plugged in (I close the lens). Could this possibly have any connection to my issues? I unplugged it earlier today... will see if any issues arise.

I also downloaded the "gaming" nvidia driver that "nvidia experience" had suggested. Now its just a waiting game to see if I crash again. I'd be happy if I don't crash again... even happier if I can find out exactly what was causing it.

Thanks again for all the help. I do love these forums.
 
Well I have been crash free for a few days now. I have run prime95 for 24 hours with cpu at 4.0ghz and I have run occt gpu test for 8 hours. No errors of any sort. I don't know what the issue was but it seems to have gone away for now. Thanks again. If something come up I'll be sure to let y'all know.
 
::pulling hair out::

Ok, so I was crash free for a few days. Then I got a few random crashes and now they are happening frequently again. They seem to occur most often when running firefox with multiple tabs open.

A recap of tests done:
Memtest86+
occt gpu test
occt cpu test
msi kombustor

Any thoughts? I've stopped getting the "graphics driver has crashed but windows has recovered" message. Now my mouse/keyboard will occasionally freeze up temporarily for a few seconds which always precedes a full freeze. I never get a blue screen or a system reboot from a freeze. It always just completely freezes and I need to do a hard reset. What would be a good HDD check?
 
Back
Top Bottom