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

2 dead GTX570s, PSU or motherboard?

Discussion in 'NVIDIA' started by steiner666, Dec 8, 2011.

  1. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    I've had 2 GTX570s die in my computer over the past couple weeks. The first card was on that i had running fine for months, i got the new version of MSI Afterburner one day and decided to see if I could bump my core clock up 10mhz and when i tested it in OCCT my computer shut completely off. I powered back on and lowered the clock back down, tested again and it shut off again. I set everything to stock and tested again, this time with Furmark and it shut off again. This time when i turned the computer back on there was no video signal and the GPU fan was running full blast. I put this GPU in another computer and it behaved the same way, so I RMAd the card back to EVGA.

    While awaiting the replacement, my friend brought his rig over and put his GTX570 in my computer and we set it to the clockspeeds that were stable in his computer, ran OCCT for a few mins to make sure it wasnt going to error right away and it seemed fine, temps were low. But we noticed that my PCI soundcard and onboard NIC were both missing. Tried restarts and checked bios but no luck. Shut PC down, removed soundcard, turned back on and the NIC showed up again. Shut down and put the soundcard back in and turned it on and the soundcard showed up again.


    Played BF3 for a while and got a display adapter/driver crash error, so I lowered the clock 50mhz and then played for the rest of the night no problem. The next day I wanted to see what was up and bumped the card back up 50mhz and started a OCCT test and it shut my computer off completely. Turned back on and set everything to stock and tried OCCT again and it shut off again, the instant the test started. Well, this card died too, no video signal in either computer, and every few times we would try turning the PC on with the card in it the fan would run at max fanspeed, but not every time like the other card.

    I was thinking "ok this seems like a textbook PSU issue" but then i got my PC running on the onboard video and noticed that the onboard NIC was missing again but the soundcard is fine. i tried everything i could think of to get it working again but nothing has got the NIC to start working again.

    So i have a Corsair AX750 PSu and a new GTX570 showing up tomorrow, but i'm worried that i'll put them both in and then still have the same problems an the GPU will get fried like the others. The fact that the onboard NIC died and the soundcard was missing for a while makes me worry that its the motherboard somehow killing these cards.
     
  2. DOM

    DOM

    Joined:
    May 30, 2006
    Messages:
    7,552 (2.44/day)
    Thanks Received:
    828
    Location:
    TX, USA
    When a psu goes bad it can mess up other things like gpu mb etc.

    I wouldn't use that mb unless u wanna be on the same boat with having to rma
     
  3. Asylum

    Asylum

    Joined:
    Jul 18, 2008
    Messages:
    2,819 (1.22/day)
    Thanks Received:
    686
    Location:
    South Carolina
    Sounds like a PSU issue to me.

    When a Video card goes bad it doesn't crash your hole system.

    Most of the time you get artifacts are black screens.

    I would change PSU's before you do anything else.
     
  4. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    thanks for the replies guys

    yeah i know its not a GPU problem because 2 different cards died in the exact same way in the computer. I know PSU is most likely, which is why i have one on the way.

    But what i'm wanting to know is if theres a chance it could be my motherboard?
    I mean the I've ran prime95 and theres no errors or heat issues, other than the onboard NIC dying it works fine. After my GPU died i enabled the onboard video and it seems to work fine (granted i had to lower my FSB with the onboard vid enabled). I looked the board over and didnt see or smell any damage, my ram tests fine, my RAID array is fine... IME when mobos die they just die completely, but theres a first time for everything. i'm just trying to find some sort of precedent for this situation so that i can have some sort of assurances that my PSU really is the problem.

    I ran a PSU test just now in OCCT and it seemed fine, only let it run for 5mins or so. But i dont know how well it can really test the PSU with just the onboard video. I did notice that the +12v is reading 9.0-9.2v in OCCT, but when i go in my BIOS it says its 11.x V... not sure what to trust there...
     
  5. DOM

    DOM

    Joined:
    May 30, 2006
    Messages:
    7,552 (2.44/day)
    Thanks Received:
    828
    Location:
    TX, USA
    I would test everything out with the new psu before I add the gpu
     
  6. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    id very much like to but im not sure how really. i mean i can swap the PSU and then boot into windows with the onboard video, but it's already working perfectly fine as it is with the current PSU using the onboard video, problems only seem to arise when a GPU is added to the equation. Unless the new PSU revives my dead onboard NIC or makes OCCT report higher than 9.2v on the +12v (not sure how accurate its readings even are) i dont know what difference i would even be able to see before trying out the GPU.

    i mean thankfully EVGA has good warranty and if i fry this one i'll be able to get another RMA replacement, but i'll go insane if i have to go through another 1-2 week period of having to resort to using the onboard video.

    the thing that sucks is that i have to run my mobo/CPU at stock speeds pretty much to use the onboard video, so even doing prime95 now doesnt mean that it will be fine when i bump it back up to my (modest) overclock. I cant really test things out without a GPU in because thats the only way i can get my clocks back to their normal speed and probably the only way i can put enough of a load on the PSU to tell if its stable or not. so the only way to know whether my new PSU fixes the problem is to take the risk of putting in the GPU.

    :banghead:
     
  7. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    Well i just did a closer examination of the motherboard and found something interesting. There is a small scratch on the bottom left corner of the motherboard by the expansion slots. The scratch is right next to a chip, i was able to make out what the chip said and its Atheros, so that must be the NIC. Now i dont have a good camera so what i did was take a stock photo and tried to approximate where the scratch is in MSPaint:

    [​IMG]

    as you can see, its right by the PCI-E x16 slot, the scratch was no doubt caused by the bracket on the end of a GPU scraping against the motherboard when it was being put in or removed. now whats really hard to make out in the picture is that there are lines running along the edge of the board that the scratch goes over, and there is exposed copper lookin material.

    I'm sure this could definitely account for the dead NIC, but does it explain the dead GPUs? the GPUs shouldnt actually be touching the motherboard where the scratch is, or anywhere, once seated. and i doubt that these lines that the scratch goes across have anything to do with the PCI-E bus because they go down to the bottom of my motherboard, not to or from the slots. They seem to be running between my onboard audio (which i dont use) on the bottom left of the picture to the CD and SPDIF output headers along the bottom of the mobo, along the right side of the picture.

    Is there a chance this caused the problems with my GPUs? If i keep the onboard NIC/Sound disabled will there be anything going through there that could potentially zap my GPU? Would it be safer if i put a little piece of electrical tape over the scratch, or maybe clear nail polish or something? I really dont want to buy another mobo cause i'm just trying to get by with this one until Ivy bridge is out and i build a new rig.
     
  8. phanbuey

    phanbuey

    Joined:
    Nov 13, 2007
    Messages:
    5,207 (2.03/day)
    Thanks Received:
    975
    Location:
    Miami
    Computer shut off is PSU related or a catastrophic fail... when I popped my GTX570 it shut off the same way, and I was having issues with my cards not being able to sustain OC's... right now im undervolted on my card until the new PSU shows up.
     
  9. animal007uk

    Joined:
    Feb 19, 2009
    Messages:
    1,254 (0.60/day)
    Thanks Received:
    292
    Location:
    Warwick, Warwickshire, England
    Maybe its just me but your mobo battery looks melted to me or has something on it.
     
  10. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    like i said, thats a stock pic i took from a website to show where the scratch is, since i dont have anything but a crappy ~1MP cameraphone ATM.

    im just wondering, in worst-case scenario, could a scratch in that location really have caused damage to the nearby GPU? I can understand that if it did break one of the tracers that its the reason why my onboard NIC went bad, but would it be able to damage the cards? And if so, will keeping the onboard NIC/Sound disabled prevent future damage? I got some clear nail polish and electrical tape, not sure which to use to cover it up to ensure that its safe. I mean its a small scratch but its decently deep, the copper is exposed on pretty much all the tracers but i cant tell if theyre broke complete (assuming at least one is since the NIC doesnt work)
     
  11. sneekypeet

    sneekypeet Unpaid Babysitter Staff Member

    Joined:
    Apr 12, 2006
    Messages:
    21,671 (6.88/day)
    Thanks Received:
    6,181
    The exposed copper is wiring in the motherboard. If exposed it can be shorted. Usually exposing the traces isn't a real issue, its more when you cut through one that issues arise.
     
    steiner666 says thanks.
  12. FreedomEclipse

    FreedomEclipse ~Technological Technocrat~

    Joined:
    Apr 20, 2007
    Messages:
    13,927 (5.02/day)
    Thanks Received:
    2,361
    sounds that that boards gotta go on a long journey to RMA heaven along with the fried GPU's
     
  13. Swamp Monster

    Swamp Monster

    Joined:
    Oct 22, 2010
    Messages:
    561 (0.38/day)
    Thanks Received:
    84
    Location:
    Latvia
    If you can't check the traces for continuity and shorts, only option is to RMA or buy a new one. Don't forget that all your trouble started when PC was fully loaded, so test/measure your PSU.
    My opinion is that you have two different issues here and that NIC and traces problem has nothing to do with your dead cards.
     
    steiner666 and phanbuey say thanks.
  14. phanbuey

    phanbuey

    Joined:
    Nov 13, 2007
    Messages:
    5,207 (2.03/day)
    Thanks Received:
    975
    Location:
    Miami
    +1
     
  15. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    well the GPUs have already been RMAd, my replacement gets here tomorrow which is why im trying to figure out if the PC is safe to put it in or not tonight.

    And the motherboard is an Asus with a 3 year warranty, which just ran at the end of last month, of course. So RMA isnt an option there or i probably would have already while ive been waiting for the new GPU and PSU to show up. If i want to replace the motherboard i'm going to have to spend 2x what I paid back in 2008 for this board to get a similar refurb, or buy a G41 board and then have to replace my ram and either give up RAID or have to buy a second controller.

    Yeah i'm really hoping youre right with your last statement. i was thinking that the issues might have been unrelated. maybe it was originally a PSU problem and during all of the swapping out of the GPU the board got scratched and thats why the NIC appeared to die around the same time that these cards were.

    I put a couple coats of clear nail polish on it, and i plan on putting a piece of electrical tape around the edge of the GPU bracket that will be close to this scratch, just in case.

    Well I checked UPS tracking and GPU and PSU will definitely be here around this time tomorrow, so i guess i'll find out then and let you all know.



    Just want to say thanks a lot for all of your input. I've started threads about these issues on multiple tech boards and this one has by far been the most helpful and gotten the most replies. TPU FTW! :rockout:
     
  16. FreedomEclipse

    FreedomEclipse ~Technological Technocrat~

    Joined:
    Apr 20, 2007
    Messages:
    13,927 (5.02/day)
    Thanks Received:
    2,361
    well...

    TBH, if your gonna be running dual 570's then you might aswell overhaul the rest and get a 2500k setup. that CPU will really get the best out of those 570s

    That is if you can afford it. I also recommend getting an 850w PSU just to be on the safe side so you know youre not stressing it if you ever need to hook anything else up.

    that would be what i would do anyway if all the critical components in my rig just went boom.

    Its something to consider anyway.
     
  17. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    heh, nah i'm not running SLi or anything, just one GTx570 in the boards one PCI-E slot. 2 GPUs died because i tried them one at a time and they both died one at a time.

    and i read a lot of reviews for the AX750 and its supposed to be pretty awesome, gold certified and all that, one review say that the overload protection is so good on the PSU that Corsair could have sold it as a 850w easily. it better be pretty awesome for $160 lol
     
  18. phanbuey

    phanbuey

    Joined:
    Nov 13, 2007
    Messages:
    5,207 (2.03/day)
    Thanks Received:
    975
    Location:
    Miami
    no matter how awesome a product is, you will always have defects... and production falls along a bell curve anyways, while that part was not labeled as a "defect" because it passed the ripple tolerance or w/e... maybe the fact that u are using stock 570's with weak VRMs to begin with, that combination is killing cards...

    There could be alot of factors in play - a close to defective/defective PSU part, combined with possible bad wiring from an older house/ shoddy power delivery + a component with weak VRMs and you have dead hardware.

    a 580 or a 6970 might have been fine because the power circuitry could cope... who knows. When things die, 99% of the time it is power delivery. Whether that bad power happened because the mobo failed at regulating power to the PCI-E, or the PSU decided to play Captain Kangaroo with the voltage to your card, you wont know.

    My advice is to replace both parts if possible. But if i had to pick one it would be the PSU.
     
  19. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    Thats one thing i was wondering about, how much is the motherboard in control of powering these PCI-E cards that have their own plug(s) directly from the? How possible is it that the motherboard somehow damaged the cards through the slot, without there being any damage to the entire board. I mean the thing works perfectly fine otherwise...

    like when you raise the voltage to the card in a program like Afterburner, is that extra voltage being delivered through the motherboard/slot directly or is the program just communicating through the slot to tell the card to take more juice straight from the PSU cables? Is there any way that a program like afterburner could damage a card because of some sort of software driver bug?

    i really hope this new PSU fixes it all, but i'll be sweatin when i'm testing things out tomorrow...
     
  20. DragonBorn

    DragonBorn New Member

    Joined:
    Dec 7, 2011
    Messages:
    75 (0.07/day)
    Thanks Received:
    10
    Whats the PIN,s on the 20/24 psu lead connectors look like I.E bent or out mite be just a short
    ON the p.s.u. power connector to the mobo
     
  21. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    looks fine to me, i dont see anything unusual



    If the VRM on a GPU dies, does it normally produce some sort of smell or even noise? i dont remember detecting anything like that with either of these GPUs
     
  22. DragonBorn

    DragonBorn New Member

    Joined:
    Dec 7, 2011
    Messages:
    75 (0.07/day)
    Thanks Received:
    10
    is the fan on the p.s.u spinning ok ? if not blowing hot air out the p.s.u mite be a over heating p.s.u. when running your rig at full load my p.s.u fan spins fast on a full load when gameing
     
  23. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    Well got the gtx570 RMA and the Corsair AX750 put in...

    and my games crash or cause the screen to shut off (not the whole computer this time). Even a light stress test revealed artifacts at stock speeds/volts. Skyrim loads into the game and then the screen goes black after a few seconds, comes back on after 5 seconds, then repeats. Crysis crashed after less than 10 seconds, Shogun 2 goes right to blue screen (my displays no-signal default) and only shutting my computer down (by pushing the button and waiting) fixes it.

    I installed Afterburner and bumped the volts from .975 to 1.025 and it made things more stable but Skyrim still goes to black screen and display driver crash message after a short while. So i bumped the volts up more and things seem to be stable now at 1.075v.

    So obviously I'm not very happy atm, since it seems that EVGA has sent me a crippled card...
     
  24. FreedomEclipse

    FreedomEclipse ~Technological Technocrat~

    Joined:
    Apr 20, 2007
    Messages:
    13,927 (5.02/day)
    Thanks Received:
    2,361
    tried testing the new card in another rig?? It could really be your board thats kicking the bucket
     
  25. steiner666

    steiner666 New Member

    Joined:
    Oct 6, 2005
    Messages:
    110 (0.03/day)
    Thanks Received:
    0
    Location:
    USA
    thats first thing on my agenda, but unfortunately i dont have anyone around with a capable PC for me to try it out in for... well, quite a while :banghead: :banghead: :banghead:

    *edit* is there a setting to enable/disable avatars? I notice that every time i log in the avatars disappear
     

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

Share This Page