PC Crashes Under Load

Sorry for the looooooong post, but I wanted to get as much detail as I could into the first post so that this wouldn't become a question-fest.

Recently my PC has been crashing whenever there is a load put on it (e.g. gaming). Whenever the GPU starts to spin up the game I'm in will start dropping frames and finally my screen will go black, with the sound still playing for a few seconds, which then slowly lapses into a loop of repeated sounds and the normal game audio for several seconds until finally the just goes into a complete dead tone, then a restart. One time this happened, the PC went to a dead tone and then a fan (I can't attribute it to a single one in my PC, since I have a top chassis fan, back chassis fan, GPU fan and CPU fan, but one of them started spinning ~3-4x faster than normal for about a second or so before the PC restarted.

I used SpeedFan to log GPU temperatures while I went for a test run and it appeared to stop logging when the temperature clocked in at 103 C (212 F). I have a EVGA GTX 670 and have read around the place that it happily works in the 90 C range. When I had been gaming previously (before the issue started), it would go to around 94 C and sit there until I finished the game, dropping dramatically back to 40 C at idle and 60 C at medium load.

I explained the issue to a friend and he said that he had the same issue. Upon describing his issue I concluded that it was indeed the same issue, and he said that after countless RMAs they (not sure who) told him it was a motherboard problem. Once he changed his motherboard it appears he hasn't had any issues, but I can't personally attest to that.

When he told me it was a motherboard issue I was thinking of any possibilities for why it would suddenly start dying and could only really think of one thing: A few days ago my grandmother turned on a really old cooking pot on the other side of the house which blew the fuse for the part of the house my room is connected to, along with the safety breaker for the entire house. Would this have caused some form of short circuiting in my motherboard, resulting in some malfunction when it is under heavier loads?

I have sent a Tech Support Ticket to EVGA as well about the graphics card to see what they say, because after taking logs with GPU-Z it seems the computer crashes after clocking in at 979.8 MHz Core Clock Speed, Memory Clock at 1502.3 MHz and 99% GPU load. I have further uploaded the log here for viewing.

Thanks for reading, and hopefully someone here will be able to shed some light on this mystery...

If it is indeed a motherboard problem, would it be worth going and forking out some money on a new motherboard (looking at the Asus Z77 series right now, but I'm up for suggestions), or trying to get some warranty from AsRock for this current motherboard (H77M bought last January).

Oh, I should also point out that I have not overclocked the computer at all, so overheating from incorrect voltages etc isn't possible.

 

EDIT: Forgot to paste my current build...

  • CPU: Intel i5 3570K 3.4GHz
  • CPU Cooler: Thermaltake Contac 16 CPU Cooler Aluminium
  • Motherboard: AsRock H77M.H77
  • Memory: Corsair 16GB (2x8GB) CMZ16GX3M2A1600C9 Vengeance DDR3 1600MHz
  • Storage: Intel 520 series 120 GB SSD | WD Green 1TB WD10EURXAV-GP
  • GPU: EVGA nVidia GeForce GTX 670 2GB
  • PSU: Thermaltake TR-600P TR2 RX 600W 80+ Bronze
  • Case: Antec Three Hundred Gaming Tower No PSU (Black)
  • Network Card: TP-LINK TL-WN881ND
  • DVD: Samsung SATA Black Internal DVD RW Drive (I know there are a lot of dumb decisions in there, but for now that's not an issue)

This sounds like a thermal issue Your Mobo I think has onboard video Pull your GPU use the onboad and then Bench test the hell out of it see if it still crashes. I bet your GPU is over heating

Could be the card...

But what about the power supply?

I had a similar if not identical issue recently and I replaced my now seemingly faulty 660 Ti with my brothers old 560 Ti and my PC works fine now.

I would say it is due to your GPU, try and get your hands on another from a buddy or something and use it for a while to see what happens.

My old 660 Ti had good temps underneath 70 celsius under full load, so I'm gonna send it back under warranty.

Yeah that was going to be my next method of action. I'll try it the next time I'm free.

Nothing happened when I loaded the CPU with stress tests, so I doubt it's an issue with the PSU not providing enough power, and I don't think I'm close to the 600W it can output.

My brother has a 770 Ti with ACX Cooler, but I doubt he'll give it to me to try because it's relatively difficult to get his tower out from where it's stationed.

Your gtx 670 is being stupid , probably overheating ?

As said above , try the onboard video . If you deduce that the gpu is the problem , contact the seller , not EVGA ( unless evga says so ) because going through RMA has always been easier with the seller , but some companies require you to go through them ( Eg : corsair ).

Do you still have the warranty ?

If not under warranty , might be that because you have had the gpu for a while , the cooler has started to seperate a bit from the gpu core ( happened to a 7850 after ~1year and 9 months ) I fixed it by taking it appart and re-doing the thermal paste.

Also , my windforce 670 never got above 85°c under load , and I would not recommend having a gpu go above 90°c ( even though 290's go above that , it does end up being harmful ).

The inside of my box has a sticker saying to go straight to them and not the seller, so I assume that's what you're referring to?

The model I have says I have a 3 year limited warranty, so I'm still covered by that. I'll take a look at the cooler and see if it's still pasted onto the core well, and if not I'll get some thermal paste and reapply it.

I can't remember exactly where I saw it, but I do remember people saying the GTX 670 happily runs in the 80s quietly, with the fan spinning up at low 90s with load. If it does happen to be my card, would it be a better option to hold onto it until it completely dies and continue repasting it, or getting a GTX 770 with an ACX cooler? My brother has one and it stays incredibly cool and quiet.

 

If you're wondering why I'm biased towards NVIDIA cards, it's mainly because from several comparisons I've seen the AMD cards perform better in artificial situations (e.g. stress testing), but the NVIDIA cards get more frames in real life situations (gaming). I also do some video editing using After Effects, so I would like to take advantage of the CUDA cores. Once again, I'm up for critique on this topic because I'm not the most knowledgeable when it comes down to the specifics of each brand.

I would hold off on reapplying thermal paste on the card. It might void the warranty just test with the onboard video and if it seems like the GPU is the problem then send it back

Good point. Someone on another forum said EVGA allow repasting, but I'll stay on the safe side.

Yes, I was recommending re pasting if you had no warranty .

My windforce 3x was pretty loud hitting 85°c under stress tests and ~72 under real gaming with a custom fan curve . I would not run my cards over 90°c , just because I know that pcb's and heat don't mix well , so that's just a personal thing .

But I don't think it's dying , but if it is , you NEED to use your warranty , don't just throw the money away .

btw if you get your 670 RMA'd , and want extra perf , just go for SLI.

But then again , sound is subjective , I find the H100i's pump loud .

I would go for a 290 tri X , and underclock it and undervolt it .

Because the 290 is faster than the 770 , if you underclock it a bit to 770 performance , you get silence .

Mu under clock for my 290 : http://imgur.com/5bMIlYt

Yeah, I ended up pulling the card out and sending the integrated graphics through some stress tests in games and haven't had a crash so far, so it's definitely an issue with the graphics card overheating. When I took it out it was extremely hot to the touch for around 10 minutes, unable to cool itself off properly.

Like I said before I don't know the exact differences between AMD and NVIDIA cards, but for now I'll see how I go with an RMA'd card, and if that doesn't work I'll probably wait for the GTX 800 series to come out and get one of those. Like I said before, more of the programs I use prefer NVIDIA architecture to AMD, so I would prefer to stick with NVIDIA if possible.