GPU or OS?

[WildCat]

I'm having a problem I think is due to a faulty graphics card. The card frequently crashes while playing games, and sometimes even when I'm not playing games I get an error of "graphics driver has stopped working and has recovered" after getting a black screen for a second or so.

I also have occasional difficulties at startup, where the computer freezes right away at startup either just before the motherboard splash screen (when it's just black) or during the splash screen. Sometimes it takes 10-15 tries before it will boot up again.

I've been dealing with EVGA tech support, they wanted me to run a stress test with OC Scanner. 5 minutes into the stress test everything went fubar and froze up, I took this pic of the screen:

It took a good 20 minutes and multiple tries to get the computer to boot up again. The log file of OC Scanner gets almost everything wrong, including the date and time. About the only thing it got right was that I'm running Windows 7 64 bit. This is the log file:

2011/01/28 @ 12:47:12 >> EVGA OC Scanner is starting up.
2011/01/28 @ 12:47:12 >> Build: 1.6.0 [Jan 28 2011 @ 12:39:52]
2011/01/28 @ 12:47:12 >> Operating system: Windows 7 64-bit build 7600 [No Service Pack]
2011/01/28 @ 12:47:12 >> CPU name: Intel(R) Core(TM)2 Extreme CPU X9650 @ 3.00GHz
2011/01/28 @ 12:47:13 >> CPU features: speed: 3007MHz, physical cores: 4, logical cores: 4
2011/01/28 @ 12:47:13 >> OpenGL version: 4.1
2011/01/28 @ 12:47:13 >> Main OpenGL graphics card: NVIDIA GeForce GTX 470
2011/01/28 @ 12:47:13 >> Device ID: 10DE - 06CD
2011/01/28 @ 12:47:13 >> Max viewport size: 16384x16384
2011/01/28 @ 12:47:13 >> Max texture size: 16384x16384
2011/01/28 @ 12:47:13 >> Detected graphics cards and GPUs:
2011/01/28 @ 12:47:13 >> - GPU 1: NVIDIA GeForce GTX 470
2011/01/28 @ 12:47:13 >> - Bus ID: 1
2011/01/28 @ 12:47:13 >> - Cores: 448
2011/01/28 @ 12:47:13 >> - Memory size: 1280MB
2011/01/28 @ 12:47:13 >> - Memory type: GDDR5
2011/01/28 @ 12:47:13 >> - Bios: 70.00.12.00.02
2011/01/28 @ 12:47:13 >> - GPU temperature: 77.000°C
2011/01/28 @ 12:47:13 >> - Graphics driver: 8.17.12.6658
2011/01/28 @ 12:47:13 >> - PState 0 - GPU clock: 607MHz - Memory clock: 1674MHz - Shader clock: 1215MHz - GPU voltage: 0.925V
2011/01/28 @ 12:47:13 >> - PState 3 - GPU clock: 405MHz - Memory clock: 1674MHz - Shader clock: 810MHz - GPU voltage: 0.925V
2011/01/28 @ 12:47:13 >> - PState 8 - GPU clock: 405MHz - Memory clock: 324MHz - Shader clock: 810MHz - GPU voltage: 0.887V
2011/01/28 @ 12:47:13 >> - PState 12 - GPU clock: 50MHz - Memory clock: 135MHz - Shader clock: 101MHz - GPU voltage: 0.875V
2011/01/28 @ 12:47:13 >> - Fan speed: 40%
2011/01/28 @ 12:47:13 >> - Fan speed: 1530 RPM
2011/01/28 @ 12:47:13 >> - GPU 2: NVIDIA GeForce GT 240
2011/01/28 @ 12:47:13 >> - Bus ID: 2
2011/01/28 @ 12:47:13 >> - Cores: 96
2011/01/28 @ 12:47:13 >> - Memory size: 1024MB
2011/01/28 @ 12:47:13 >> - Memory type: DDR3
2011/01/28 @ 12:47:13 >> - Bios: 70.15.24.00.00
2011/01/28 @ 12:47:13 >> - GPU temperature: 56.000°C
2011/01/28 @ 12:47:13 >> - Graphics driver: 8.17.12.6658
2011/01/28 @ 12:47:13 >> - PState 0 - GPU clock: 550MHz - Memory clock: 790MHz - Shader clock: 1340MHz - GPU voltage: 1.000V
2011/01/28 @ 12:47:13 >> - PState 8 - GPU clock: 405MHz - Memory clock: 324MHz - Shader clock: 810MHz - GPU voltage: 0.900V
2011/01/28 @ 12:47:13 >> - PState 12 - GPU clock: 135MHz - Memory clock: 135MHz - Shader clock: 270MHz - GPU voltage: 0.900V
2011/01/28 @ 12:47:13 >> - Fan speed: 10%
2011/01/28 @ 12:47:13 >> - Fan speed: -1 RPM
2011/01/28 @ 12:47:45 >> Exit from EVGA OC Scanner.

This says I have 2 GPUs, a GTX 470 and a GeForce GT 240. The reality is I have only one GPU, a GTX 460 1024 MB. My CPU is not an Intel Extreme quad-core, but a Core 2 Duo E8500. And this result is from yesterday, not Jan. 28.

Now the kicker is I think the GPU is messed up, but EVGA isn't so sure and wants me to reinstall Windows(!).

It would take me weeks to get my system back to where it was by reinstalling Windows, plus I would lose my progress in all my games. I really really don't want to do that. I never had any of these issues prior to installng this GPU in mid-December.

So what I need to know from the experts here is does this sound like a Windows problem, or is it an issue with a faulty GPU? I have already run memtest and my memory showed no errors.

 

[Babbylonian]

I'd lean toward the graphics card based on the timing, but your description of symptoms (particularly the difficulty booting) could also indicate trouble with the motherboard. The OS itself would be low on my list of suspects, and it seems like it could be ruled out completely if you could try booting another OS from CD/DVD/flash drive.

 

[C_Felix]

Did you give the computer a good thumping?

 

[WildCat]

I'd lean toward the graphics card based on the timing, but your description of symptoms (particularly the difficulty booting) could also indicate trouble with the motherboard. The OS itself would be low on my list of suspects, and it seems like it could be ruled out completely if you could try booting another OS from CD/DVD/flash drive.

The booting issues only happen when re-starting after a crash that freezes the system. Normally it boots up fine.

 

[Policenaut]

It looks to me like the gpu vram is bad. Now the graphics card possibly could have also fried the motherboard.

 

[roger]

If you have, or can borrow/buy/steal a hard drive, slap it in your machine and install windows on that. You could probably even try installing to a USB drive, if you can boot from USB. I've only done that with Linux, not windows, but it is theoretically possible.

And, with that said, to expand on what Babblonian says, you can put Ubuntu (a version of Linux) onto a USB stick and boot from it. The Ubuntu site gives full instructions.

But, I would buy a new hard drive. Since you are worried about losing all of your stuff, this tells me you probably don't have a back up solution implemented. Buy a 1TB drive for 60 bucks, install windows on it, and prove that it is your card, not the OS. Then wipe the drive, leave it in your system, and use it as a backup drive.

I know, it's fun to spend other people's money - this may not be a reasonable solution for you.


If you have trouble at boot, is the computer beeping at you? Those are known as post codes - you can look up the post codes on the internet (you'll need to know your motherboard) - and it may narrow down the problem (though the EVGA tech support should know about them and have already asked you to do this). Some motherboards have an LED display on the motherboard to display the post codes.

I will say post codes are not infallible. I had a problem with a new EVGA card I bought - the post codes were complaining about USB or something. I convinced EVGA to do an RMA for me, and the card I received worked perfectly.

edit: if you can cannibalize another machine, you can pull the graphics card from it and put it in your machine. A friend may let you borrow her card for an evening, if you don't have another pc around. It's kind of a pain, but it is what I did to convince EVGA tech support that the problem was with the card, not my PC.


edit2: my card was the exact card as yours, and also in the mid december time frame. I wonder if there was a QC problem with a batch of cards.

 

[Babbylonian]

One other thing...

Now the kicker is I think the GPU is messed up, but EVGA isn't so sure and wants me to reinstall Windows(!).

While I know lying isn't ideal, given that you have good reason to suspect the video card is the problem and assuming that you're not just trying to get EVGA to send you a big upgrade (I obviously don't think you are), simply tell them that you tried reinstalling Windows and it still doesn't work.

Reinstalling Windows on my personal PC wouldn't be as big a deal for me as it sounds like it would be to you, but I still wouldn't do it as part of a diagnostic routine in an attempt to save a video card company a few bucks.

 

[Darat]

Is Windows otherwise working fine? If so then I would say it's unlikely to be your Windows installation - I take it you've done the usual - uninstalling the graphic drivers and reinstalling them and trying Safemode? If that hasn't fixed it I very much doubt that a reinstall of Windows would fix it. If they really want to see if it's your Windows installation you could downland a Ubuntu installation and install that alongside your Windows (it can create its own partition etc) that way you can check out whether it is an OS problem or not as if the problem persists when you boot into Ubuntu it's very, very unlikely to be a Windows problem.

 

[Chris L]

It would make more sense to me to just try to reinstall the video driver instead of the entire OS. If that doesn't work, it probably the card.

Darat, you beat me to it.

 

[ddt]

If they really want to see if it's your Windows installation you could downland a Ubuntu installation and install that alongside your Windows (it can create its own partition etc) that way you can check out whether it is an OS problem or not as if the problem persists when you boot into Ubuntu it's very, very unlikely to be a Windows problem.

I'm not sure if trying it out in Linux is an appropriate diagnostic here. Standard Linux installs typically include only basic graphics drivers that don't take advantage of the acceleration features of modern graphics cards. So you'd have to install the proprietary NVidia driver as well to have a good comparison - but if you do so, it is.

 

[The Norseman]

I'm not sure if trying it out in Linux is an appropriate diagnostic here. Standard Linux installs typically include only basic graphics drivers that don't take advantage of the acceleration features of modern graphics cards. So you'd have to install the proprietary NVidia driver as well to have a good comparison - but if you do so, it is.

I don't see why not. It can help separate OS vs. hardware vs. drivers issues by eliminating the OS portion for one thing. Linux also can probe the hardware in different methods and my personal experience is that I can get much more detailed information from a live Linux boot. I have, however, a live Win7 and live WinXP disk, so I try to cover all my bases.

Besides, I didn't see that the card was failing only during stress, so I wouldn't be too worried to try and install the Linux Nvidia drivers for a 3D test at this point. Also, just as in Windows, there are Linux programs made specifically for stress-testing.

Most mobo's these days have on-board graphics, so it might just be easiest to first remove the video card and see if problems replicate.

Take the card and install it into another box if possible and try that.

I would not have suggested reinstalling the OS. Rolling back the drivers or trying a system restore point would be the maximum I'd go with at first.

 

[WildCat]

But, I would buy a new hard drive. Since you are worried about losing all of your stuff, this tells me you probably don't have a back up solution implemented. Buy a 1TB drive for 60 bucks, install windows on it, and prove that it is your card, not the OS. Then wipe the drive, leave it in your system, and use it as a backup drive.

I actually have 3 hard drves installed on this machine, 2 are 1TB SATA drives in Raid 1, the 3rd is a 250GB IDE drive I use to back up pics and documents for an extra layer of protection.

If you have trouble at boot, is the computer beeping at you?

No idea, my case doesn't have a speaker.

edit: if you can cannibalize another machine, you can pull the graphics card from it and put it in your machine. A friend may let you borrow her card for an evening, if you don't have another pc around. It's kind of a pain, but it is what I did to convince EVGA tech support that the problem was with the card, not my PC.

I have my old GPU still, it's a XFX GeForce 9600 GSO. Problem is it has several obviously blown capacitors on it, which is why I replaced it. But I wasn't having these problems with that one even with the blown capacitors, it just acted flaky when playing games. I could try putting it back in and see what happens.

edit2: my card was the exact card as yours, and also in the mid december time frame. I wonder if there was a QC problem with a batch of cards.

That's interesting!

 

[WildCat]

Is Windows otherwise working fine? If so then I would say it's unlikely to be your Windows installation - I take it you've done the usual - uninstalling the graphic drivers and reinstalling them and trying Safemode? If that hasn't fixed it I very much doubt that a reinstall of Windows would fix it. If they really want to see if it's your Windows installation you could downland a Ubuntu installation and install that alongside your Windows (it can create its own partition etc) that way you can check out whether it is an OS problem or not as if the problem persists when you boot into Ubuntu it's very, very unlikely to be a Windows problem.

Yes, it works fine otherwise. I have already reinstalled all the drivers to no avail.

 

[WildCat]

I think the simplest solution is to disconnect my SATA drives and put Windows on the IDE drive.

I'll have to clean that drive up a bit first, it only has 20GB of free space but there's lots I can delete. Should probably defrag it first also.

Or I could just buy a new 1 TB SATA drive, and replace the IDE drive with that then I can do proper backups.

Hmmm, the latter sounds like a better option. There's a Tiger Direct a short drive from here.

 

[WildCat]

Well last night whle playing CoD WaW multiplayer the computer froze. I was unable to start it normally, only in safe mode. It would get to the Windows welcome screen and freeze.

Today, same thing. So I removed the 460 GTX and put in the old 9600 GSO card with the blown capacitors. The computer booted right up with no problems.

I have contacted EVGA again and requested an RMA. I am not about to jump through any more hoops with this card, I am convinved it is the GPU and not any other hardware or system issues.

 

[Mongrel]

Well last night whle playing CoD WaW multiplayer the computer froze. I was unable to start it normally, only in safe mode. It would get to the Windows welcome screen and freeze.

Today, same thing. So I removed the 460 GTX and put in the old 9600 GSO card with the blown capacitors. The computer booted right up with no problems.

I have contacted EVGA again and requested an RMA. I am not about to jump through any more hoops with this card, I am convinved it is the GPU and not any other hardware or system issues.

Make sure you back everything up then. A wipe and re-install will probably be top of the list of things they do

 

[Chris L]

Mongrel,

If I'm reading Wildcat correctly, he has decided it is the GPU and wants them to take the thing back. I don't think he really wants to reinstall his OS at this point. The whole "reinstall your OS" thing sounds like it was just their way of not having to deal with him further for a while.

 

[WildCat]

Mongrel,

If I'm reading Wildcat correctly, he has decided it is the GPU and wants them to take the thing back. I don't think he really wants to reinstall his OS at this point. The whole "reinstall your OS" thing sounds like it was just their way of not having to deal with him further for a while.

Exactly, I have an RMA now. I'll be sending it back for a replacement tomorrow. Err, later today since it's after midnight.

 

[Mongrel]

Exactly, I have an RMA now. I'll be sending it back for a replacement tomorrow. Err, later today since it's after midnight.

Apologies, looks like I got confused

For some reason I thought it was a full system you where talking about

 

[WildCat]

Update!

I got the replacement card via RMA, and have had no isues at all with the computer with the replacement card installed.

Just one problem... the replacement card (the replacements are the same model, but refurbished) gets very hot very fast during game play. 98C after just a few minutes, when the last one worked it never got above 75C or so. I didn't notice this until yesterday because I hadn't got around to installing the monitoring software until then.

So now I have an RMA for this one as well, at least EVGA is paying for the shipping this time around.

I suspect the heat sink isn't installed properly, the fan seems to be working fine and there's no obstructions in my case and in fact there's a fan on the side of the case directly over the GPU.

I don't think I've ever had to RMA twice for a product. Seeing as Roger has had problems with this exact card I think EVGA is having some quality control issues.