1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

GEForce 8800 Ultra Problems

Discussion in 'Nvidia' started by W, Dec 16, 2012.

  1. W

    W Guest

    I have a GEForce 8800 Ultra PCIe card installed under Windows XP on a Dell
    Precision 390 workstation with a quad core processor. Whenever I attempt
    to use Microsoft Flight Simulator, soon after starting the game I noticed
    that tile areas of the ground surface start to blink and appear in reverse
    video. This problem worsens, and within minutes the game freezes. It
    appears to be some kind of massive video corruption. The computer locks,
    and at some point I get a blue screen of death. NVidia driver version
    under Windows Device Manager shows as The only clue on
    reboot about the error is a System eventviewer message:

    The computer has rebooted from a bugcheck. The bugcheck was: 0x100000ea
    (0x87eb1da0, 0x89362378, 0xb84efcbc, 0x00000001). A ywas saved in:

    Does anyone have any idea what might be causing this problem? Would it be
    bad hardware or just an unstable driver release?

    Is there a preferred version of the nVidia driver for this card?
    W, Dec 16, 2012
    1. Advertisements

  2. W

    Paul Guest

    I don't have an answer for you, but I can offer a couple hints.




    A device driver problem has caused the system to pause indefinitely
    (hang). Typically, this is caused by a display driver waiting for the
    video hardware to enter an idle state. This might indicate a hardware
    problem with the video adapter, or a faulty video driver.



    NVidia had an era, where the GPU chips they were selling,
    had a thermal coefficient of expansion issue. Now, one solution
    for this, is underfill, a curable compound that they place in the
    same area as the BGA solder joints, and it has something to do
    with taking stress off the whole mess.

    Some of the cards fail after a while, because a solder joint cracks.

    This led to the "home solution" of baking the card. Whereas in a
    lab environment, you'd get a hot air rework station (heats a large
    chip from top and bottom, using a custom fitted hood to limit where
    the heat goes). Home users just heat the whole damn card. When
    cards are double-sided, and have components on both sides, this
    can lead to stuff falling off.

    Never the less, there are reports out there, that re-baking a card
    can correct a connection problem. What I don't know, is how long
    such a repair lasts. Whether the repair is permanent or not. I
    would expect the same stresses are still present, and the GPU
    "moves" with respect to the PCB, attempting to tear the solder
    joints again.


    There are patent applications on the web, that address the need
    to use underfill on BGAs with more than 1000 balls. And a
    change from one of those materials, to another, can increase
    the number of balls that can safely be used on a hot chip,
    to decent level (extending lifetime by many years).

    An inevitable question you'd have is "how do I know my card
    has this problem, a broken solder joint ?". Well, your symptoms
    would likely be worse, and not restricted to a problem showing
    up in one game. There'd be problems all over the place. Maybe
    a black screen at startup. Perhaps you'd play a Flash video,
    and see issues. Or depending on the OS, maybe you'd use
    hardware-driven features of Aero, and see anomalies. If only
    the single game has issues, then driver or game patch are just
    as likely to fix it (no baking required).

    For my own purposes, I use 3DMark benchmarks, as a separate
    test. You can download some of those, and they'll test
    fewer features of DirectX than perhaps a recent game would.
    I might use these, to hint at whether I thought it was
    driver or hardware. If all of these have artifacts, I'd
    be prepping my EasyBake oven :) Or doing more focused
    hardware diagnostics. There are also things like Furmark,
    for doing power-limited testing, but the drivers recognize
    some of those apps, and prevent them from pounding on
    the card to 100% level.

    (40MB, pretty old, 3Dmark 2001 SE (Build 330))

    (178 MB, 3DMark03 3.60 1901)

    (209MB, a bit newer, 3DMark05 1.3.0 1901)

    I don't know what value testing like the following has, but
    it's another test case you can run. When apps like this come
    out, the driver designers have to respond and add code to prevent
    the power used, from exceeding safe levels. I don't know
    exactly how they do that. The core power regulator on the
    video card, would have overcurrent protection, but the
    GPU might be pretty warm before that kicks in.


    Paul, Dec 16, 2012
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.