f13.net

Was using pc for a long session suddenly gets pop out when running graphic intensive games 'Driver stopped working, save work and reboot'
Didn't happen before. First time I've seen this. I decided to try out other driver. No dice, hard crash and I had to reboot after a short use.

I get no problems when browsing the web etc, but when I start running games the problem would start. Any ideas?

What's the temp on the GPU before it crashes?

Hmm any idea how to monitor that? my game usually hard crashes...will give it a shot again. Currently idle at 56degree
...climbed to 66C while running CIV4

Do you have your computer set to automatically restart on a STOP error, or can you see BSOD's?

Download nTune from nVidia and lower the clock rate on your memory and core. Drop something like 50MHz from memory and see if that helps stability under load.

RivaTuner can do it, though the UI is a bit fiddly. There's a "Customize" button in the Target Adapter section in the Main tab with a Hardware monitoring button popout and if you click it enough times you'll get an option for installing the hardware monitoring plugin and a window will popup with various things being monitored and you can have it save stuff to a file.

http://downloads.guru3d.com/download.php?det=163

You can also run something like this to give you better control of the "stress" on your GPU:

http://www.ozone3d.net/gpu_caps_viewer/index.php#intro

around 69ish degree peak.. I was having a smooth normal game - windowed mode of Civ4 . Then the whole graphic would just flare up and program would crash.
quite worrying. Is it the heat management issue? Cause the Nvidia monitor said it's at constant 69 degree, no higher.

Slightly more stable crash though, just the gave froze. I manage to close it and resume operation as normal.

For the GPU itself 69C isn't unusually high. If it is getting past 80C I would start to worry. However, there may still be a problem with the cooling of the VRAM. How are those cooled on your board?

Edit: Do you have a portable fan you can blow into your case onto your video card?

yeah doing it now, gonna stress test it shortly.

Yes, it's dying.

alright i think i just have to keep it cooled more and have the internal fan checked. It's pretty stable so far. No crashes with constant fan blowing. Hopefully this shit will last.  :why_so_serious: Maybe it's just a minor heat issue.

Dust? I've had cards BSOD and lose performance over dust. I take poor care of my systems.

About 10 years ago I had a card fry, looked at it and the fan wasn't even moving anymore because of the dust.  I invested in a small air compressor and dust my system every 3 months now.  Haven't had an issue since. 

Gotta be VERY careful using a compressor on electronics though... too close and you can blow the smaller circuits right off a board. 

This.  I've been keeping my card fans clean and have had much fewer problems.  They still die due to normal wear, but it usually happens at a slower rate than my normal upgrade pace. 

 This is a major killer, I picked apart a friend's system (http://forums.f13.net/index.php?topic=12553.msg424159#msg424159) a year ago and found the thing was almost cooked.

 I've been running an accelero s2 (http://www.arctic-cooling.com/vga2.php?idx=125) on my 8800gt, thanks for reminding me to clean it out  :awesome_for_real:

I've had horrible dust problems like this too. And I don't even want to talk about how revolting the inside of my PC was back when I used to chainsmoke at the keyboard.

just did a short cleaning, nothing major. I cleaned some dust off the fans. Overall the inner casing is within 'acceptable' clean level. Will it work? We'll find out shortly.

Edit: ok fucked right from the main menu of FFH mod (Civ 4)  :why_so_serious:
I can't pinpoint the cause, temp was ok. Dust was cleaned. Something's broken somewhere but I can't determine whether it's hardware or software.

The whole 8x00 series has issues.  Which issue is biting you?  It could be the "crappy solder grows tin whiskers and shorts out" issue.  Or the "crappy board substrate warps while heating and cooling and breaks the solder joins" issue.  Or the "resin inside the chip casings develops cracks that cause the actual silicon to break" issue.  Or the "lousy QA sent out hundreds (thousands?  nobody knows) of GPU's with faulty texture units" issue.  Or the "drivers designed to keep the users from actually knowing anything went wrong with their cards generated a false positive for one of the preceding issues and are now trying to emulate the broken hardware (that may or may not be actually broken) in software, badly and slowly" issue.  Or maybe you just had too much dust clogging up your box, or a bad ventilation pattern, and your card cooked something critical in a way completely unconnected with NVidia's sloppiness.

Short form: Your card is fucked, and you may never know exactly why.

--Dave

sigh i guess it's time to upgrade anyway.  :heartbreak: then follow up with a whole system format for new year.
Onward to 2009!

I'm sorta with Trippy on this one. Although your main GPU chip was cool enough, those surrounding ram chips get hot as well, and I have seen a few configurations where the heatsinks on the ram chips were essentially an afterthought. Chances are one of them is blown. The reason I think this is because although you're able to operate in the OS, as soon as the GPU's being asked to do some work, it fails, which suggests that the address space on the GPU memory is being accessed, and if you have one bad chip of RAM, you'd BSOD. The same thing happens with defective system ram. You can boot and operate for a while, but then suddenly there's a function call to a higher memory address and boom, BSOD.

The upshot is, as everyone's said, that its time to get a new card.

Reason for my short answer was this just happened to my old 6800GS.

Dust got the fan and it got wildly hot and died.  Once the dust was cleared and the cooling was fantastic again, it still shat bricks.  It had just cooked and no amount of dicking about would get it to behave again.

Sorry for your loss.