Constant display driver resets under load, occasionally unresponsive after standby

Discussion in 'Computer Help' started by Phris, Nov 4, 2012.

  1. Phris

    Phris Member

    Joined:
    Jul 15, 2007
    Messages:
    74
    Likes Received:
    0
    Hello,

    My desktop, after playing some Borderlands 2, suddenly decided it didn't want to live anymore and now the graphics driver crashes constantly under load (sitting at the desktop / browsing the internet are fine). Additionally, now and again, it'll forget it has a mouse and keyboard when coming out of standby and occasionally forgets it has hard drives. This would've been cause for concern, I suppose, but until the graphics errors, nothing couldn't be fixed by a reboot. All signs point to needing a new motherboard: funny, because the warranty just ran out. I've replaced the graphics card and the RAM and reformatted, which has done nothing to help.

    GPU stress tests alternate between no crashes after half an hour and crashing after two minutes or so; every time, it's a "display driver reset" in event viewer. This would be upsetting enough if this weren't the second motherboard to do so in this computer; every single component save the graphics card and a storage hard drive have been replaced since the original build and it STILL fails near-constantly. No overclocking or anything, either; nor do I even use it all that often.

    Am I cursed? It just seems like this computer hasn't wanted to live since I built it. Sure, I'm an electrical engineer, but maybe I should just buy pre-made from now on...

    Oh, right, specs:
    Gigabyte GA-EP45-UD3P
    Intel E8400
    Corsair 750HX
    EVGA 8800GT
    Some amount of RAM, Crucial Ballistix (terrible stuff, would not buy again)
    Caviar Black HDD, 650GB
    Some 200GB PATA drive

    Anything I'm missing?
     
    #1 Phris, Nov 4, 2012
    Last edited: Nov 4, 2012
  2. Steltek

    Steltek Golden Member

    Joined:
    Mar 29, 2001
    Messages:
    1,626
    Likes Received:
    0
    System specs?
     
  3. Phris

    Phris Member

    Joined:
    Jul 15, 2007
    Messages:
    74
    Likes Received:
    0
    Now with system specs!
     
  4. Steltek

    Steltek Golden Member

    Joined:
    Mar 29, 2001
    Messages:
    1,626
    Likes Received:
    0
    I previously had a socket 775 ASUS Rampage Extreme board that never worked right when coming out of sleep mode. I eventually had to disable sleep altogether because of memory corruption problems (ASUS just stopped BIOS development in lieu of releasing new motherboards, so it couldn't be resolved). So, I'd separate the issues resulting when you resume from sleep from those video driver failures. Push come to shove, you can always disable sleep (a pain, but cheaper than buying a new board).

    First, I'd try booting to safe mode and totally remove the video driver (google how to manually remove the remaining files and edit registry after you run add/remove). Then, reboot and reinstall the driver. It also might not hurt to move the video card to a different motherboard slot (if one is available) after you remove the driver and right before you restart to reinstall. If you have been running the most recent nVidia video drivers (v306.97 dated 10/10/12), try to drop back one or two releases to ensure it isn't a random driver bug. If you don't have any type of monitoring software installed, install SpeedFan so you can check your system temps when the driver dies to make sure your video card (or some other component) isn't overheating next time the driver dies. I'd also update your motherboard chipset drivers if you haven't done it in a while as well. If you aren't running the most recent BIOS, it might also be worth updating.

    It also wouldn't hurt to run a few sessions of MEMTEST86+ on each separate memory stick. If you are running 4x4GB sticks, you might even try running on just two sticks for a while (some older boards just don't do well with 4 modules installed). Are your memory modules on the motherboard's memory compatibility list?

    Also, I'm not clear - at one point, you said you replaced the video card, but at another that the video card and hard drive are the only components you haven't ever replaced. Have you swapped out the video card? If not, it is worth considering as video cards do sometimes have an unfortunate tendency to die (usually related to heat).

    Finally, what specific power supply do you have? How old is it? Have you ever tried to replace the power supply before?
     
  5. Phris

    Phris Member

    Joined:
    Jul 15, 2007
    Messages:
    74
    Likes Received:
    0
    Good to know about the sleep thing; that always seemed sporadic. I don't mind disabling sleep, I only use the desktop for my music library and gaming, especially since it has reliability issues as long as I've owned it.

    I can try removing the driver in safe mode, but I did completely wipe my hard drive and reinstall windows when the problem started, to no avail. I've also dropped back several times, all the way to 301 and less. I would think it would be a driver problem, but it only started appearing after I put in about twenty hours of gaming over a week or so after not using the computer for a while.

    To clarify what I meant on the graphics card, I have a spare older card that I've swapped in for testing purposes (a 8800GTX) and I've tried it in both slots on my motherboard (although it seems that the second slot is only a PCI-E 8x). The 8800GT that I use most of the time is the original card (never been RMA'd, it never met error criteria) and hasn't ever seemed to die, at least not yet. Sometimes it doesn't even error out when I run EVGA's stress tests, although it seems strange that two cards (one of which had never been in the computer before, the 8800GTX) have the exact same error in either slot.

    Right now I'm down to one 1GB memory stick; once upon a time, I ran 4x2GB Crucial Ballistix but those sticks seem to die fairly often. I'll double-check the compatibility listing for the memory; I don't know if I've ever checked it. Last time I ran memtest on the single stick, it came up clean.

    My specific power supply is the 750W Corsair 750HX and it's almost two years old (but, again, doesn't see a terrific amount of use). It replaced an Antec Earthwatts 500 that died and took everything with it (everything but the graphics card got RMA'd after that, although the motherboard came back clean and Gigabyte just returned it). That power supply killed a previous Abit IP35-E board, as best I can tell. Otherwise, this computer may just be possessed. Pretty sure that's it.
     
  6. Steltek

    Steltek Golden Member

    Joined:
    Mar 29, 2001
    Messages:
    1,626
    Likes Received:
    0
    It sounds like you've done just about everything you can do in this case. I don't suppose you have any indications of overheating (say specifically on the chipset North Bridge - it has a big heatsink, but that doesn't preclude overheating) at the time of the graphics failure? If not, I have to agree that you are possibly looking at a motherboard fault.

    The only other thing I could think of off the top of my head would be a grounding issue related to the motherboard mounting. As a last resort, you could pull the motherboard out of the case and place it on a cardboard box or antistatic bag to see if this eliminates the problem.

    The bad thing about it is there aren't that many decent socket 775 motherboards out there now, and most of the ones left are only in micro-ATX form factor (which means replacing a lot more than just the motherboard if you can't live with 2 memory slots and decide to build a new system - I found that out myself while researching a replacement for my aforementioned ASUS board which died 3 weeks ago, less than one month out of warranty....).
     
    #6 Steltek, Nov 4, 2012
    Last edited: Nov 4, 2012
  7. Ketchup

    Ketchup Lifer

    Joined:
    Sep 1, 2002
    Messages:
    12,593
    Likes Received:
    8
    After reading your thread Phris, the first thing that comes to my mind is power delivery. If your pc won't come out of sleep mode, and your video card is randomly dying, that just scream power delivery problem to me. It could be something outside the computer, but I don't see that as affecting every sleep mode but none of your startups. I would try a new power supply. 500 watts should be plenty for your system.

    Second guess would be memory. I have lost count of how many people say memtest comes up clean, only to find out memory was the issue. And you could even try raising the memory voltage in the bios to check this. The only problem with this guess is that I don't see how this would cause your specific restart out of sleep issue.
     
  8. VirtualLarry

    VirtualLarry Lifer

    Joined:
    Aug 25, 2001
    Messages:
    33,700
    Likes Received:
    88
    I agree with Ketchup79. Most likely PSU or RAM.
     
  9. Phris

    Phris Member

    Joined:
    Jul 15, 2007
    Messages:
    74
    Likes Received:
    0
    I guess just to clarify on the sleep issue: it does do it on startup, too. Sometimes I have to restart it a couple of times to get it to remember it has hard drives. And sometimes my PATA drive drops out for no reason at all.

    I've got a spare PSU here from a Dell XPS that died...chances are the 24-pin is dell proprietary, but I suppose I could try running the graphics card off the secondary supply and take a lot of load off the current one?

    I'll run some more memtests shortly.
     
  10. Phris

    Phris Member

    Joined:
    Jul 15, 2007
    Messages:
    74
    Likes Received:
    0
    So...I decided to pursue the overheating route. Only thing that even got warm was my Southbridge, which would actually get hot under load.

    Pointed a desk fan inside my case and, well, now things seem to be okay.

    I may have also turned up every fan inside the computer to max, but that was after the fact, just to give me a little more headroom.

    Thoughts? Odd that the southbridge would get so much hotter, but then again, this motherboard has a GIANT northbridge heatsink and a relatively tiny southbridge one.

    Just thought: what's the chances of the thermal paste under the northbridge degrading? Odd that the northbridge wasn't getting even warm under load, but it is a giant heatsink...
     
    #10 Phris, Nov 4, 2012
    Last edited: Nov 4, 2012
  11. wirednuts

    wirednuts Diamond Member

    Joined:
    Jan 26, 2007
    Messages:
    7,121
    Likes Received:
    1
    southbridge chips usually dont need any heatsinks, so a small one should be more then enough

    your motherboard might be a lemon....
     
  12. Steltek

    Steltek Golden Member

    Joined:
    Mar 29, 2001
    Messages:
    1,626
    Likes Received:
    0
    It may just not have a good application of thermal paste under the heatsink. The only way to tell is to pull the heatsink, clean it up, and re-apply it.

    Another solution would be to get a small fan to mount there, provided it wouldn't be in the way of anything important:

    http://www.frozencpu.com/cat/l2/g40/c16/list/p1/Air_Cooling-Chipset_HeatsinksCoolers.html
     
Loading...