Penny Arcade Forums› Games and Technology› Moe's Stupid Technology Tavern
The new forums will be named Coin Return (based on the most recent vote)! You can check on the status and timeline of the transition to the new forums here.
The Guiding Principles and New Rules document is now in effect.

Overheating video card?

LednehLedneh shinesquawkRegistered User regular
edited November 2009 in Moe's Stupid Technology Tavern
Up to yesterday I had a Geforce GTS 512. Infrequently, it would hardlock my machine--black screen, looping sound, the works. Usually while playing something high stress (Dragon Age), but not always (WoW).

Yesterday, thinking that since it only happens when the video card is under stress it must needs replacing, I got a new video card with my work bonus. An XFX Radeon HD 4890, to be specific.

Well, it does it worse--probably 30 minutes into a game, or 3 minutes into a GPU burn stress test (OCCT v3, specifically), it dies. GPU-Z shows the temperature at between 85 and 90C at the time of the crash, though it sometimes crashes far lower (73C during Dragon Age, for example).

So this led me to think, could running the video card be causing some OTHER component of my system to overheat? Some piece of the motherboard, perhaps? Or did I really get two videocards in a row with cooling issues?

Help :(

Case: Antec P180
PSU : OCZ 700W (forget the model, sorry)
CPU: Q6600 (clocked at default of 266MHz FSB * 9 = 2.4 GHz)
Memory: 2x2GB+2x1GB Corsair DDR2-6400 (800MHz, underclocked at 2x266MHz FSB = 533MHz)
Video: see above
No overclock or voltage settings besides underclocked RAM (I checked, it had no impact)

Ledneh on

Posts

  • AzioAzio Registered User regular
    edited November 2009
    I had a similar problem a while back. Freezing/crashing in certain games, not solved by reinstalling Windows or using a different graphics card. Eventually I got a new motherboard and the problem went away. It's entirely likely that some other component is responsible for the instability.

    Try removing some RAM and see what happens.

    Azio on
  • LednehLedneh shinesquawk Registered User regular
    edited November 2009
    Did some more fiddling with the OCCT GPU test... opened up the side of the PC and ran it. Same crash after about three minutes.

    Aimed a window fan on full blast into the open PC and ran the test...made it six minutes before I turned the fan off whereupon it died some 30 seconds later, though the recorded temperature of the GPU never actually rose at all in that 30 seconds.

    I'm at a loss, I've read that 85C is not all that hot for a GPU under full load, so I don't know what the fuck anymore. :(

    (edit) I'll try taking out all but one 2gb stick, see what it does... though earlier I ran prime95 and memtest86 and stuff, and the RAM was all okay.

    Ledneh on
  • AzioAzio Registered User regular
    edited November 2009
    85-90 is pretty hot but not quite in the danger zone for a graphics card

    the broken motherboard I mentioned passed prime95 and memtest86 without any trouble. I think it was just the combination of stresses on every component that was causing it to crap out. Could have been bad capacitors or voltage regulators or something.

    Azio on
  • LednehLedneh shinesquawk Registered User regular
    edited November 2009
    I reeeeeally can't afford to get a new CPU/mobo and probably new RAM too right now though D:

    Ledneh on
  • LednehLedneh shinesquawk Registered User regular
    edited November 2009
    Welp, update: now I can't boot into Windows, and I have to reset the CMOS every time I want to get into the BIOS options at all because the keyboard is nonresponsive until the Windows boot menu comes up. And it won't let me boot from my install CD either.

    I think the benchmarking/burning to identify the problem broke either the motherboard or the CPU :cry:

    (edit) Oh good! Now the Windows 7 install CD boots to an IRQL_NOT_LESS_OR_EQUAL BSOD. Wonderful!

    Ledneh on
  • AzioAzio Registered User regular
    edited November 2009
    sounds like a dead motherboard, hopefully you can get it replaced under warranty

    Azio on
  • SatsumomoSatsumomo Rated PG! Registered User regular
    edited November 2009
    That videocard is dying actually, IRQ_NOT_LESS_OR_EQUAL errors are directly related to it. Or the PCI-E port :P

    Satsumomo on
  • LednehLedneh shinesquawk Registered User regular
    edited November 2009
    If it's the PCI-E port wouldn't that mean a dead motherboard?

    And I would hope the video card's not dying, seeing as how it's new yesterday :(


    (edit) It booted! To a BSOD with PAGE_FAULT_IN_NONPAGED_AREA.

    uuuuughhhhhh

    (edit again) Took out all but one 2GB ram stick and it booted cleanly, and OCCT's GPU test has run for more than six minutes now.

    (seven minutes) whoops there it goes :(

    Ledneh on
  • AzioAzio Registered User regular
    edited November 2009
    Satsumomo wrote: »
    That videocard is dying actually, IRQ_NOT_LESS_OR_EQUAL errors are directly related to it. Or the PCI-E port :P
    what? Those errors can be indicative of any number of hardware issues

    Azio on
  • SatsumomoSatsumomo Rated PG! Registered User regular
    edited November 2009
    Could be, I should take that statement back since it's been a long time since I've troubleshooted a machine with that error. But it's somehow embedded in my memory that it had to do with hardware in expansion slots.

    Satsumomo on
  • AzioAzio Registered User regular
    edited November 2009
    IRQL_NOT_LESS_OR_EQUAL it means that a driver or system process tried to access memory that it's not allowed to access, which can be caused by all manner of things: buggy drivers, broken parts, overheating chips, bad power supply, etc

    it's more of a symptom than a cause

    Azio on
  • LednehLedneh shinesquawk Registered User regular
    edited November 2009
    Well, I've underclocked the card from its factory 850MHz core/975MHz memory to 750MHz core/850MHz memory (which GPU-Z believes to be the default clock rate for this chip), and it's holding steady at 79 celsius under full load from OCCT's GPU suite.

    So I bought a new video card... just so I could underclock it? Wtf?

    But then if underclocking this thing is making it work, why the hell would my old video card ALSO crash the machine under normal stresses? AAAAAAAAAAAA

    Ledneh on
  • LednehLedneh shinesquawk Registered User regular
    edited November 2009
    Well, put all my ram back in and resumed running OCCT-GPU while underclocked--crashed five minutes in.

    To this I say fuck it, I'm replacing the mobo, CPU, ram and case. Fuck hardware diagnosis in the god damned ear. :x

    Ledneh on
  • AzioAzio Registered User regular
    edited November 2009
    the core i5 is pretty goddamn nice and you can get good deals on mobo+chip+ram combos

    Azio on
  • DedianDedian Registered User regular
    edited November 2009
    Ledneh wrote: »
    Well, put all my ram back in and resumed running OCCT-GPU while underclocked--crashed five minutes in.

    To this I say fuck it, I'm replacing the mobo, CPU, ram and case. Fuck hardware diagnosis in the god damned ear. :x

    Make sure you evaluate all aspects of the issue, not just the GPU. It does sound like the mobo was going, either that or some cooling issue with the CPU or something on the mobo (crash after shutting off the huge fan pointed into the open case). I didn't notice if you mentioned any CPU temps...

    The underclocking may have helped alleviate whatever was going on with the mobo, but hey, good time for an upgrade :D

    Dedian on
Sign In or Register to comment.
Penny Arcade Forums› Games and Technology› Moe's Stupid Technology Tavern