Do the GPU Stress Test (GPU:OCCT), and then set the Shader Complexity to 3 as shown in the tutorial to be safe, set the resolution to your current screen resolution, and then enable Error Check to allow the program to log any errors. Run the test for 1 hour.
The passes is for the GPU Memory Test if your graphics card supports this option and is a separate test.
I started the test a few minutes ago, but the card tops 80C in a matter of a couple minutes and looks like it would continue rising in temperature if I let it keep running. Either the temperature readings are off or my GPU has an overheating problem by the looks of it. I attached the results of a five minute run and a 3 minute run and it really seems like it shouldn't get that hot so quickly. What do you make of it?
I just checked the GPU fan with the PC off and I noticed the fan's quite dusty, and the frontal fan ent is very dusty as well. I still don't think that quite accounts for the rapid increase in GPU heat. I need to clean this puppy out, though. Stupid dusty-arse Texas.
You need to clean all the fans and the heatsinks, the temperature shouldn't increase so rapidly, seems to be most probably a overheating issue so far, although I'm not sure wherever the heat may have damaged your hardware as of yet.
If it is overheating, I would expect it to rise in heat rapidly whilst in games, which it doesn't seem to do (stays at about 60C-70C). Also, how does overheating explain the crashes when I'm idling on my desktop? Not to mention the USB error and such, as they seem unrelated to a graphics issue (although I read that crashes normally just point to the last major driver running before the crash; so maybe the USB thing is irrelevant?). I'll clean the dust out ASAP anyways, though. It's kind of rainy today so I might have to wait a day or two.
60*C to 70*C is still quite hot, and report back after you have cleaned your computer.
@BlueRobot Will do.
@TwoCables Yeah, I didn't think so, that's why overheating doesn't seem to fit. The only time it's ever (that I know of) gotten up to 80C is with OCCT. But hopefully once I clean out the dust it'll run a bit cooler. Currently and through most of today the card has idled in the mid 50's range, peaking during ArmaIII alpha at 76C. Running the OCCT test for the amount I did couldn't have damaged the card, could it have?
It's hard to ignore the quick rise in heat from OCCT, regardless of the game temperatures.
By the way guys, on GPU-Z I noticed that the maximum GPU power consumption was 101.5% (averaging in the 30%-40% range), is that normal? What does the power consumption tab even indicate, the power being drawn from the PSU (lordy I hope not!), or the maximum amount of power that can be utilized by the GPU? Just thought I'd ask in case it's relevant.
Not a chance. Overclockers run these kind of programs for 12-24 hours (some go longer) just to test the stability of their overclock!
Yeah, but OCCT is designed to work the GPU much harder than anything else.
I'm not sure, personally. I've never really gotten this far into GPU-Z.
@TwoCables Thank you for the very neat, categorical answers! :)
No game I play, at any time, tops 80C (most barely reach and rarely top 70). I've been playing games all day today and the max peak heat is at 77C.
Do you guys think I can rule out an overheating issue, or is that still very likely the problem? If I don't do the stress test should I still do the memory test? I personally no longer think the card is overheating; I've gotten crashes idling on the desktop before and once while playing Ace of Spades (which while playing I barely top 60C) in the past (I've also played more demanding games for hours without crashing, as well as crashed while watching online videos and loading browser tabs). I would think overheating would at least be consistent, crashing only when under high load.
I compared my graphics heat level to a friend's (he has a GTX 570) and his temperatures are equal and sometimes higher than mine in the same games, and his idle temp is 10C higher than mine most of the time (not to mention he lives in Great Britain, whilst I in TX, USA; needless to say, there's a HUGE difference in climate).
I still cling to it being a MOBO or very hard-to-pinpoint software issue. All along I've wondered if it's software, because it seems like if a hardware component was defective/damaged I would see at least some kind of hit in overall performance, but it continues to operate flawlessly...until it doesn't.
You're welcome!
If it were overheating, then your performance would suffer. I've seen it before several times, so that's how I came to find this out.
Forgive me for the following because I was gone for a while and I couldn't follow along, but did you end up testing your memory one stick at a time in one slot at a time?
Quote