PC instability while gaming - GPU or PSU

Discussion in 'Troubleshooting Help' started by Menzies, Apr 17, 2019.

  1. Menzies

    Menzies Member

    Joined:
    Oct 13, 2010
    Messages:
    34
    Hi there,

    Have a bit of a weird one, newely built PC (second hand GPU & PSU) has been crashing constantly since the beginning of Feb (started with BSOD during Anthem closed demo weeked)

    8700k (de-lidded, stability no different at stock clocks or 5.2GHz)
    Strix Z370-F Gaming (latest BIOS, BIOS update didn't solve anything)
    970 Evo 1TB m.2 SSD
    Corsair AX760 (purchased 2013, custom sleeved cables)
    16gb 3200MHz Trident Z (no different at stock speeds or XMP profile)
    Inno 3D GTX 1080Ti (purchased second hand 2018)
    Custom loop, 600mm worth of rad's (CPU Peak mid 60's, GPU peak low/mid 40's)

    The first issue I ever had was when the PC was first built in Dec 2018, I experienced artefacting after the GPU drivers didn't install correctly, this was solved by re-installing the GPU driver and the system was rock solid until early Feb when I experienced constant BSOD's during the Anthem demo's. Removing my OC and XMP profile seemed to fix things.

    On the full release of Anthem the problems got worse, less BSODS's and more lock ups (black screen, audio looping) with no crash dumps or logs. The only thing I could do to get a few hours of gameplay was power limit the GPU with MSI Afterburner.

    PC was fine playing less intensive games like Apex and experienced no crashing whatsoever. Started playing The Division 2 and the frequency of crashes increased, power limiting GPU was again the only solution but it would still experience crashes. Checking Event Viewer I am getting Event 13 and Event 14 constantly (nvlddmkm cannot be found) and now even crashing in Apex every 2-3 games. Cannot find the nvlddmkm file anywhere so not sure if it is an issue with drivers or hardware.

    I've tried multiple versions of drivers (latest and ones that pre-date the crashes), disabled Windows from installing any default display drivers and a fresh install of Windows. Voltages in HWMonitor look good.

    I have the feeling that it is the GPU, I have spare GTX 780's but since my custom loop is hardline it is a pain to swap out GPU's for testing. I am yet to check the PSU cables with a multimeter as I don't think testing without load on the PSU is going to tell me anything.

    Sorry for the wall of text but I am hoping I am missing some simple fix and it isn't having to replace my GPU, don't mind replacing the PSU so much but I really don't think it is the PSU...

    Thanks
     
  2. JSmithDTV

    JSmithDTV Member

    Joined:
    Jun 13, 2018
    Messages:
    2,169
    Location:
    Algol, Perseus
    Have you tried a different PS?

    JSmith
     
  3. im late

    im late Member

    Joined:
    Jan 5, 2012
    Messages:
    1,519
    Location:
    Canning Vale WA 6155
    I have seen this before......have you checked your pcie power connectors are firmly plugged in at both ends (gpu end and power supply end)?

    Such a simple thing sometimes overlooked....
     
  4. straiton

    straiton Member

    Joined:
    Feb 12, 2003
    Messages:
    423
    Location:
    Adelaide
    most of the time, black screen and audio looping is GPU related IMO.
    Is it a full cover block on the ti ?
    As above, if poss. reseat all power cables.
    Is it possible to wiggle the gpu out slightly and reseat?
     
  5. OP
    OP
    Menzies

    Menzies Member

    Joined:
    Oct 13, 2010
    Messages:
    34
    Sorry for the late replies, cracked the shits and ended up not touching the PC since this was posted.

    Don't have a spare and don't really want to buy one if I don't have to

    All checked, replaced with spares just in case. Same issues

    Yep full cover block, as above tried reseating. Might attempt eliminating the PCI-E riser and see if that solves it.
     
  6. Bold Eagle

    Bold Eagle Member

    Joined:
    Jun 28, 2008
    Messages:
    8,655
    Location:
    Brisbane
    OCCT does some serious stress testing and even if the system hard crashes you will have DATA plots (graphs) that you can review on the next reboot to try and analyse the possible cause.
    http://www.ocbase.com/index.php/download

    Give it a little test run on everything so you know where the plots are being saved to. It will do all of the voltages on your PSU, etc, etc.
     
  7. poony

    poony Member

    Joined:
    Dec 16, 2005
    Messages:
    16
    Location:
    Melbourne
    I used to have this problem on a HD4870 specifically with Bad Company 2.
    I did that MSI Afterburner trick too, but it just delayed the inevitable.

    I was 90% sure it was the GPU, but also had an inkling it was the motherboard. But I was much younger back then and didn't have the resources or money to test to confirm it.

    IMO, switch the GPU out.
     
    Last edited: May 3, 2019
  8. Aratahu

    Aratahu Member

    Joined:
    Jul 12, 2004
    Messages:
    559
    Location:
    Melbourne VIC
    Done a full test of the memory? I ran for ages with too little vcore.. Fixing that up sorted me out (for now).
     
  9. OP
    OP
    Menzies

    Menzies Member

    Joined:
    Oct 13, 2010
    Messages:
    34
    Ran the test and got some artefacting before the system locked up, same as what happens in game. Not too sure what to make of the results though. I've added a link to the tests if you would be willing to have a look at it.

    https://imgur.com/a/As4eTfn

    Tedious solution but I suppose I could throw one of my old 780's in, guess this is the downside to rigid tubing :(

    Ran both Memtest and the Windows diagnostic with no issues.
     
  10. OP
    OP
    Menzies

    Menzies Member

    Joined:
    Oct 13, 2010
    Messages:
    34
    Ran the test again tonight as I cannot be sure of the last tests run on OCCT as I came back to the screen in standby mode. As you can see between 18-19 minutes something seems to happen on the graphs. This correlates with the PC first locking up for 10+ seconds followed by my main display dropping back to a default resolution, (not sure what resolution, but definitely not 1920x1080).

    When the display did come back, it was joined by what sounded like coil whine coming from the GPU, wasn't able to pinpoint exactly where it was coming from, but it would momentarily go away when unplugging either monitor (Alienware 120Hz - DVI, BenQ 60Hz - HDMI). At this point I'm guessing that it is a GPU on it's last legs, I guess I will have to start looking for a replacement.

    https://imgur.com/a/VSFm1v4
     
  11. Bold Eagle

    Bold Eagle Member

    Joined:
    Jun 28, 2008
    Messages:
    8,655
    Location:
    Brisbane
    Let's be clear from the onset - I have only used OCCT on a limited number of occasions but find it's capacity to plot a system up to the point of a crash can be useful. I am not an expert in the interpretation of output but can make some observations at best. I take no responsibility nor offer assurances in the advice that is provided.

    Just wanted to clear that up make sure I am not blamed for anything!

    In the first lot of results (https://imgur.com/a/As4eTfn) you present it can be noted that at about the 8min mark their is a system spike affecting the;
    - CPU,
    - GPU,
    - Memory,
    - +12V

    This last element (the +12V rail) could that be starting to fail intermittently? Causing the GPU,CPU and Mem to hang at the 8mintues mark, all temps drop as expected during that time as the system looks like it struggles (hangs) for about 5minutes but then continues as expected for another 50mins.

    On the other hand in the 2nd lot of DATA the +12v seems to be fine but at the 18min mark we can see the GPU FPS crash, CPU usage goes up - it does seem to show a GPU issue.
     
  12. OP
    OP
    Menzies

    Menzies Member

    Joined:
    Oct 13, 2010
    Messages:
    34
    Thanks for the feedback, I felt the same way when setting the first set of data but looking at the variance on the 12v rail it wasn’t like it was a huge spike, afterwards the GPU seemed to recover for a short while before the FPS halved itself. I’m wondering if the increase on the 12v rail was due to the system trying to constantly recover the GPU after it crashed, which I’m guessing is what happened eventually but I wasn’t at the PC to confirm this.

    I suppose for now I will just have to chuck in a 780 which should confirm if it is GPU or PSU as I’m sure it draws more power than then 1080Ti.

    Thanks everyone for your help, it is much appreciated!
     
  13. Bold Eagle

    Bold Eagle Member

    Joined:
    Jun 28, 2008
    Messages:
    8,655
    Location:
    Brisbane

Share This Page

Advertisement: