PC reboots when video card is under stress

Discussion in 'Video Cards' started by Amadeo, Apr 1, 2016.

  1. Amadeo

    Amadeo n00b

    Messages:
    9
    Joined:
    Apr 1, 2016
    Hello, I have the following problem: a while ago I was doing some work that's GPU intensive, and at the exact second I started it and the video card went 100%, the PC reboots itself. To check that it wasn't a problem specific to the task I was doing, I tried some games like Fallout 4 and GTA V (which a couple of months back I had played with no problems) and I got the same results, while other less consuming games like League of Legends run with no problems (most of the time. Sometimes it reboots too). This happens in both Linux and Windows, so I kind of discarded that it's software related.
    I also run 3DMark's benchmarks, with mixed results. The first and most basic benchmark, Ice Storm, runs fine, but the second one, Cloud Gate, makes the PC reboot instantly.
    The Furmark GPU stress test reboots instantly.
    Because I lack hardware knowledge, I find it difficult to diagnose if the problem is in the video card, or if it's some problem with the power supply.

    Relevant hardware:

    Video card: Gigabyte GTX 980 ti.
    PS: Thermaltake Smart 750w 80plus Bronze

    Both have 6 months of use.
     
  2. Scizyr

    Scizyr Limp Gawd

    Messages:
    235
    Joined:
    Jun 24, 2015
    It could be heat related, can you start FO4 or GTA5 in window mode so you can monitor your GPU temperatures? Try going to the Nvidia control panel and force the fans to 100% and test it like that. Physically touch the heatsink of the graphics card while it's stressed, it should be getting warm or hot, if it's cool then the gpu isn't transferring heat to the heatsink. If all these checks still cause the issue, then move on to check your power supply.

    You could also try running folding@home doing a CPU load test to ensure the problem is caused by the GPU itself (whether it's heat related or power related).

    Also try run a memory test to ensure your RAM is ok.
     
  3. Drewis

    Drewis Limp Gawd

    Messages:
    453
    Joined:
    Jul 25, 2006
    usually its a PSU problem in this case...

    it just reboots?? no bluescreen? no error message upon windows restarting??

    is everything connected properly; GPU all power plugs are secure?? motherboard 8 pin secure?? how are your CPU temperatures?
     
    YeuEmMaiMai likes this.
  4. Amadeo

    Amadeo n00b

    Messages:
    9
    Joined:
    Apr 1, 2016
    FO 4 and GTA 5 reboot instantly, so there is no time for the temperature to go up. In the case of League of Legends, I did it and the temp. is around 53º. The few times it makes the PC reboot, is exactly at the beginning of the game, when all 5 players spawn in the Nexus.

    Regarding the fans, I manually set them to 100% and they run just fine.
     
  5. Brian_B

    Brian_B 2[H]4U

    Messages:
    2,729
    Joined:
    Mar 23, 2012
    Usually, not always, but usually

    If the video card is failing, you'll typically get a BSOD with it.

    If the system cold reboots (like hitting the reset switch) - it's a power issue, that's usually the overcurrent protection in your PSU kicking it off. That may be your PSU going bad, it could be the VRMs on the video card, it could be a totally unrelated short circuit (like a stray wire or screw underneath the motherboard)
     
  6. Amadeo

    Amadeo n00b

    Messages:
    9
    Joined:
    Apr 1, 2016
    Absolutely nothing. It's exactly like when you manually push the restart button.

    I think so, yes, because 2 or 3 months ago I run all the games I listed with no problems. I don't know what could have changed. CPU temp is about 43º
     
  7. Armenius

    Armenius I Drive Myself to the [H]ospital

    Messages:
    17,431
    Joined:
    Jan 28, 2014
    Reboots under load with no blue screen is usually indicative of a power issue. Your power supply is considered a tier 4 unit out of 5 tiers. I would replace it ASAP.

    PSU tier list 2.0

    I would at least replace it with something in the Tier 2 list, if not Tier 1. It's a shame to be running a top-end card like a 980 Ti on a bad power supply.

    You can also check the motherboard's temperatures with HWInfo64 to eliminate a PCH or VRM cooling issue.
     
  8. jojo69

    jojo69 [H]ardForum Junkie

    Messages:
    10,371
    Joined:
    Sep 13, 2009
    so guys, let me slightly derail as I have a similar thing going on

    2 6950s in crossfire, original issue Corsair AX1200

    after some time under load the screens would go black, computer appeared to still be running, this happened several times. I tried reseating everything related to the GPUs and blowing out the dust, still happened.

    Pull the #2 card and now everything is fine...except of course my framerates are down


    #2 card or PSU??
     
  9. Armenius

    Armenius I Drive Myself to the [H]ospital

    Messages:
    17,431
    Joined:
    Jan 28, 2014
    In your case I would say it's the video card. Try testing the #2 video card in your system by itself. This happened to me on an old system when the second 8800 GTX started to fail. Tested it by itself and got artifacts on boot.
     
  10. jojo69

    jojo69 [H]ardForum Junkie

    Messages:
    10,371
    Joined:
    Sep 13, 2009
    yeah, that's the next step

    maybe its time for a bake
     
  11. Brian_B

    Brian_B 2[H]4U

    Messages:
    2,729
    Joined:
    Mar 23, 2012
    Could be the PCI slot too... or possibly driver related
     
  12. jojo69

    jojo69 [H]ardForum Junkie

    Messages:
    10,371
    Joined:
    Sep 13, 2009
    well, let's just say that these Caymans (ahem, cough cough) paid for themselves many times over. I am not going to be the least surprised if they are feeling a bit tired
     
  13. skeeder

    skeeder [H]ard|Gawd

    Messages:
    1,316
    Joined:
    Jul 9, 2004
    Sounds to me like a memory error. It could not be graphic card related at all. I had a similar issue back in the day with a stick of memory that went bad (Turns out that slot for memory went bad on the board). Run a Memtest on it and see if you get errors. Shouldn't take that long to figure that out--it probably occurs when gaming since that is when it is getting to that part of the memory.

    Good luck.
     
  14. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    6,790
    Joined:
    Jul 29, 2009
    Unstable CPU over clock?
     
  15. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    I agree that He should run some cpu stress test just to make sure the board/cpu/ram is stable and configured correctly. cpu stress tests dont pull as much power usually and should be able to pass even with a bad power supply. (maybe) See if it can pass intel burn test or prime95 or whatever. Unstable cpu or memory overclocks can reboot just as well......a bad power supply turns off in many cases i think rather than just restart.

    edit: you have furmark installed already. Try just the cpu burner if nothing else
     
    Last edited: Apr 1, 2016
  16. Amadeo

    Amadeo n00b

    Messages:
    9
    Joined:
    Apr 1, 2016
    I tried Intel Burn Test with maximum stress and memtest86. Both did OK.
     
  17. YeuEmMaiMai

    YeuEmMaiMai [H]ardForum Junkie

    Messages:
    14,338
    Joined:
    Jun 11, 2004
    exactly... PSU demand typically causes voltage drops and then boom, reboot.
     
  18. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    that very good news...it really helps to rule out all possibility's. So were down either power supply or the card being faulty. You have a volt meter for testing the power supply rails? Maybe fire up gpu-z and see if your seeing any noticeable voltage being off there. If the level are off real bad we can lean that direction and focus more on checking that.
     
  19. Amadeo

    Amadeo n00b

    Messages:
    9
    Joined:
    Apr 1, 2016
    No I haven't got one, and unfortunately my electronics knowledge is zero, so I would probably do it wrong :p

    I have tried two programs that specifically test the GPU ram: OCCT and memtestG80. Both restarted the PC after a couple of seconds. In the case of OCCT, I managed to take a picture of some voltage values: http://s23.postimg.org/jphthfijv/image.jpg

    Would this mean that the problem is within the GPU, or not necessarily?
     
  20. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    your voltages look terrible.....i mean BAD. See if you can get me some voltages while just stressing the cpu. Cause right off the get go its displaying 6 for the 12, 3 for the 5 and so on. It still might just be a software error since some motherboards dont read voltages correctly. All mine always have but i hear some dont. Really need you to stop by Walmart and pick up a cheap digital voltmeter out of the electronics area like in the automotive or tool section. Should be able to get one under 20 bucks. Dont worry about your electronics knowledge, shit any of us here can give easy instructions on what to do.. Basically your just putting a black lead on black and the red lead on yellow (12 volt), and then red (5volt) with meter on volts setting. Doesn't have to be Walmart as any electronics or even automotive should have meters good enough. A store with decent return policy would be best.
     
  21. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    Now if a meter backs up what were seeing in OCCT then we can call the power supply bad. If it doesn't then well focus more on the video card and run a few more test in that direction. its just a terrible idea to RMA a gpu without doing simple tests on the power supply. You dont want to RMA the wrong part:)
     
  22. Armenius

    Armenius I Drive Myself to the [H]ospital

    Messages:
    17,431
    Joined:
    Jan 28, 2014
    It's a power supply issue. It looks like As soon as the 12V gets loaded by the GPU the power supply is nope'ing the fuck out. Like I said: the Thermaltake Smart series is a poor quality unit and you should replace it. I would recommend anything from Seasonic, anything from XFX, the Corsair HX- or AX-series, or the EVGA GS- or P2-series. Here is the tier list again.

    PSU tier list 2.0
     
  23. silent-circuit

    silent-circuit [H]ardForum Junkie

    Messages:
    16,133
    Joined:
    Sep 18, 2005
    Honestly, I'd replace that PSU on principle. It might not be the problem now (though it likely is) but if not now, it will be given enough time.
     
    Armenius likes this.
  24. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    So you guys would RMA the power supply without even even putting a volt meter on it? I would never do that on mine lol. It probably is the supply yes- but it needs to be confirmed with a meter lol
     
  25. Armenius

    Armenius I Drive Myself to the [H]ospital

    Messages:
    17,431
    Joined:
    Jan 28, 2014
    RMA? I would take it outside and melt it with thermite to put it out of its misery.
     
  26. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    LOL that cracked me up

    Ill be the first to admit, that particular model doesn't have the best track record (69 bucks at new egg doesn't help my impression either)....I know my thermaltake supply has been flawless.(total different model) By the same logic anyone who owns a 980ti should a least have a volt meter to test their supply. Like having 5000 dollar set of rims and tires on your sports car but not even having a tire air gauge. If your going to own a high end pc you should own a meter for troubleshooting.
     
    pendragon1 likes this.
  27. Armenius

    Armenius I Drive Myself to the [H]ospital

    Messages:
    17,431
    Joined:
    Jan 28, 2014
    Not all power supplies are made equal. I can't count how many times I cringe when I see someone on PCMR post their new build and then they have a Corsair CM or RM unit powering it. Thermaltake, just like Corsair, has their gems and their turds.
     
  28. pendragon1

    pendragon1 [H]ardForum Junkie

    Messages:
    12,877
    Joined:
    Oct 7, 2000
    looks like psu. all your voltages are dropping off. if its only 6 month old, rma...
     
  29. admiralperpetual

    admiralperpetual Limp Gawd

    Messages:
    415
    Joined:
    May 7, 2015
    i've been having similar problem for awhile, old games were fine but whenever I ran anything current and stressful I'd get black screens... I swapped out my PSU and added another GPU.. was running SLI and noticed sometimes I'd get flickering grass and textures, things like that. Then occasionally started getting black screens again. Pretty sure it's when games start using all the VRAM - so for now I've disabled SLI and am using the troubled card for physx only.. i'm debating if I can do anything to fix it or if I should just toss it :/
     
  30. Amadeo

    Amadeo n00b

    Messages:
    9
    Joined:
    Apr 1, 2016
    Last edited: Apr 7, 2016
  31. athenian200

    athenian200 Gawd

    Messages:
    727
    Joined:
    Mar 29, 2012
    It's probably the PSU. I once swapped in a 560Ti into an old system upgrading from a 8800GTX, and it started crashing in video games. Turned out that even though the PSU was rated for that card, it couldn't actually handle it. Things got better for a while after swapping out the PSU, but unfortunately I still had a few problems with the system randomly locking up during games and benchmarks... and it turned out that the motherboard's SLI implementation and power circuitry was a little dodgy. To be fair, it might have been my fault as I tried to overclock the Q6600... failed, because it wouldn't even take 100MHz over the stock settings in a stable manner. I put it back on stock settings, but it never worked very well after that. So I ended up having to replace the Motherboard, CPU, and RAM as well (at the time, decent LGA 775 motherboards were inflated to the point it was cheaper to buy all three components new than to buy a replacement off eBay and keep the other two)... and then, finally, the PSU and the video card worked fine in every benchmark I could throw at them. I've never really liked that machine since, though... I ended up building a new machine in a junky old case with an old hard drive, old fans, and a video card I only intended as a stop gap... about a year before I built the system I actually wanted. But I was right in the middle of a new game and I wanted to finish it.

    In other words, hope your PSU is the only problem, because otherwise it's your Motherboard or your Video Card, both of which will be fairly expensive to replace.
     
    Last edited: Apr 7, 2016
  32. pendragon1

    pendragon1 [H]ardForum Junkie

    Messages:
    12,877
    Joined:
    Oct 7, 2000
    you posted a OCCT gpu test, did you ever run the psu test? Do that first before spending money and poking around inside. But I really think you should just rma it, your voltages are fucked!
     
  33. Amadeo

    Amadeo n00b

    Messages:
    9
    Joined:
    Apr 1, 2016
    Oh yes, I forgot to mention it. It was also instant reboot.

    Nevertheless, I want to make the multimeter test, not because I don't believe you about the PSU being the issue, but because I want to have more arguments so that the vendor cannot "turn a blind eye?" (english is not my first tongue, I don't know how to say this exactly, but you get the idea).
     
  34. pendragon1

    pendragon1 [H]ardForum Junkie

    Messages:
    12,877
    Joined:
    Oct 7, 2000
    I get what you mean. youre time youre money don't zap yourself....
     
    Armenius likes this.
  35. jojo69

    jojo69 [H]ardForum Junkie

    Messages:
    10,371
    Joined:
    Sep 13, 2009
    also don't void your warranty opening the PSU case
     
    pendragon1 likes this.
  36. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    Yep thats fine....I think most of the modern meters in that price range are close to the same in quality. I see this particular brand recommend to DIY people in many cases Extech MN35 Digital Mini MultiMeter w/ Large easy to read digital display. In the us we would pick one up on the way home from work just about anywhere...If money is no object than get a Fluke:). For most DIY people any modern brand is fine.

    Also btw all your really doing is checking voltage on the connectors......dont open the power supply lol
    [​IMG]
     
    Last edited: Apr 7, 2016
  37. jackofalltrades

    jackofalltrades 2[H]4U

    Messages:
    3,657
    Joined:
    Feb 7, 2006
    I have had good luck with pawn shops obtaining high quality Fluke meters for next to nothing. Most pawn shops don't know the difference between Fluke and a cheapo.
     
    maro and pendragon1 like this.
  38. Guille_arg

    Guille_arg n00b

    Messages:
    25
    Joined:
    Mar 7, 2013
    PSU or temp problem, do some test or maybe try to put the vga on other computer to discard which are the issue. Good luck!
     
  39. primetime

    primetime [H]ardness Supreme

    Messages:
    5,594
    Joined:
    Aug 17, 2005
    Anyways op what you want to do is to put one lead on black and the red on yellow (12volt) you can see the yellow wires in my picture (same as yours). you cant really check those since they have to be plugged in to run the test. Your just trying to confirm what you seen with occt is actually happening or not. Run the same stress test and watch your meter and see if the 12 volt rail dips as it did before when the stress test starts. Either it does or it doesn't is really all your checking for. Some meters can keep track off a low range but the cheaper ones you'll probably have to watch it closely for the voltage dips. My meter will record the min/max volts but most will not and you just have to watch it closely. (easy enough)
     
  40. vmirjamali

    vmirjamali Limp Gawd

    Messages:
    169
    Joined:
    May 14, 2010
    If it is a heat issue try grabbing some copper heatsinks off of ebay. They work wonders for mine.