1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Very stange issues... PSU to blame?

Discussion in 'Overclocking & Cooling' started by hat, Dec 5, 2010.

  1. hat

    hat Maximum Overclocker

    Joined:
    Nov 20, 2006
    Messages:
    16,820 (6.21/day)
    Thanks Received:
    2,032
    Location:
    Ohio
    I've been getting some very strange issues with my system that appear to only happen when my GPU is overclocked. Darkplaces, an advanced Quake engine, crashes. L4D2 was acting up on me... it stalls and plays the last sound byte (for about a second), but doesn't actually crash (it did once as soon as I joined a game, though). Sometimes, folding@home crashes on start (generic Windows error message... folding@home has stopped working and needs to close). It occurs to me that none of these issues are present when my GPU is overclocked. I've run 2.5 hours of the OCCT GPU test on it without a single error, so the card itself should be stable.

    This leads me to wonder if my psu is giving out under the added load an overclocked GPU imposes. The GTX260 is a pretty heavy GPU, the first I've ever owned that required two 6-pin connectors. It doesn't fail in the stability test, but messes with my games (which would impose load on the entire system rather than just the GPU like the OCCT test). I've tested the rest of my system with 10+ hours of OCCT Linpack, and I've ran memtest86+ for 4.5 hours (enough time for at least 3 passes), so I know the rest of my system should be good to go.

    Power supply is an Antec NeoHe 550w. I would guess this ought to be enough to handle an overclocked GTX260, however, it's one of those "multiple" +12v rail units (all 3 rated for 18A), which really isn't 3 rails, just a single rail split up into 3.

    And so, I'm thinking the PSU can't handle it. What are your thoughts?
  2. slyfox2151

    slyfox2151

    Joined:
    Jan 14, 2009
    Messages:
    2,605 (1.36/day)
    Thanks Received:
    524
    Location:
    Brisbane, Australia
    i have nearly the exact same system as you with an overclocked GTX260 216 core 65nm @690mhz core.... running 6 HDDs in raid and nearly 4ghz on the cpu....


    unless the PSU is dieing that shouldnt be the issue

    (my psu is 4+ years old and only the 500 not 550 model)



    maybe you should drop the Video card down to 680 core....
    hat says thanks.
  3. {JNT}Raptor

    {JNT}Raptor New Member

    Joined:
    Jul 12, 2005
    Messages:
    733 (0.23/day)
    Thanks Received:
    87
    Location:
    NY
    If like your saying it only happens when the Video card is overclocked....then ya....it's either the vid card or the PSU....time to start swapping out parts and track it down bro.....never been a fan of multi-rail PSU's myself...nice big single 12volt rail will always be better......my $'s on the PSU.

    Hope it helps. :)
    hat says thanks.
  4. hat

    hat Maximum Overclocker

    Joined:
    Nov 20, 2006
    Messages:
    16,820 (6.21/day)
    Thanks Received:
    2,032
    Location:
    Ohio
    How does it pass 2.5 hours in OCCT GPU without a single error yet fail in all kinds of games though?
  5. avatar_raq

    avatar_raq

    Joined:
    May 22, 2010
    Messages:
    229 (0.16/day)
    Thanks Received:
    36
    Location:
    Iraq
    If possible overvolt the gpu and/or vRAM. If not step the clocks down a notch or two.
    hat says thanks.
  6. xBruce88x

    xBruce88x

    Joined:
    Oct 29, 2009
    Messages:
    2,330 (1.43/day)
    Thanks Received:
    532
    it could be that in OCCT GPU its mostly your gpu using power and your cpu not so much, whereas in a game your cpu is working harder and using more power plus the power of the gpu

    as for the problem... how old is the psu? try blowing out some dust in it, could be getting hot. Check and make sure the fans are working good.

    Try this: http://www.thermaltake.outervision.com/

    It'll let you adjust for aging of the psu. doesn't take into account how the rails are set up though
    Last edited: Dec 5, 2010
    hat says thanks.
  7. avatar_raq

    avatar_raq

    Joined:
    May 22, 2010
    Messages:
    229 (0.16/day)
    Thanks Received:
    36
    Location:
    Iraq
    Or it could be that OCCT is not using the whole video ram while games do use them, and the Vram clock is unstable!
    hat says thanks.
  8. hat

    hat Maximum Overclocker

    Joined:
    Nov 20, 2006
    Messages:
    16,820 (6.21/day)
    Thanks Received:
    2,032
    Location:
    Ohio
    I did the OCCT memtest too... and it locked up with my vram at stock clocks, just the GPU and shaders clocked up.

    Now at previously stable speeds of 640 core/1405 shader, but with the voltage bump (now 1.18v as opposed to the default 1.12v when I ran those clocks before). We'll see if it still crashes... if not it's most likely the psu.
  9. avatar_raq

    avatar_raq

    Joined:
    May 22, 2010
    Messages:
    229 (0.16/day)
    Thanks Received:
    36
    Location:
    Iraq
    Well that excludes the unstable ram clock. Nevertheless try bumping up the voltage a little and see what happens.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page