Graphics card crashing under load

[ethanwbeardsley]

I’ve just upgraded to a Gainward GTX 970, and i’ve been having some big issues so far. I also installed a new power supply with the graphics card, a Thermaltake TR2 S 600W. After installing the card, I installed the drivers from Nvidia’s site and ran heaven benchmark to check if it worked correctly. I ran a few graphics heavy games like Assassin’s Creed Origins and Middle Earth: Shadow of War for about 30 minutes, and everything seemed to work fine. I did notice that the card was a bit louder than my previous GPU (750 Ti). The next day, I tried to run Dishonored 2. 5 minutes into the game, the screen goes black and says no signal. I restarted my computer and uninstalled my drivers with DDU and reinstalled, then opened Afterburner to monitor temperatures, power, and gpu usage. All values were normal, the GPU at about 30-35°C idle. Under load, temps went to 57° and we’re steadily increasing until 78°. At this point, I shut off Heaven Benchmark and turned the settings from ultra to high. I lowered the core clock by -50 in afterburner, and ran Heaven again. After about 30 seconds (after the temps had returned to normal), the screen snapped to a pink/white color and was unresponsive. Mouse, keyboard, and audio all work, but it’s just the monitor that is unresponsive. I have no idea why this is happening, I checked the condition of the GPU fan and it seems to be working. How can I fix this??

My specs:
Case: Sharkoon VG5-W
Mobo: Gigabyte H110M-DS2
PSU: Thermaltake TR2-S
RAM: HyperX Fury Black 2x8GB
CPU: i5 6600K
GPU: Gainward GTX 970
Monitor: Acer S242HL

 

[chemmajorp53]

Well I mean you described 2 different types of issues. Just going black sounds more like a PSU issue than anything else and tbh the Thermaltake TR2 series is pretty garbage, but a new PSU still shouldn't be doing that. The pink/white thing sounds like either the GPU/VRAM being unstable or the VRAM itself being faulty, could be a lot of things. Did you overclock anything? Both of these issues could be caused by bad ripple regulation, if you can still return the power supply I would, for only a little more money you can get something WAYYYY better.

 

[dorsai]

If I had to guess I'd say the power supply unit is no good. Have you tried swapping the old GPU back in to see if it works ? Honestly I didn't know they were still making 970's...

 

[chemmajorp53]

He didn't say he bought it new

 

[ethanwbeardsley]

Well I mean you described 2 different types of issues. Just going black sounds more like a PSU issue than anything else and tbh the Thermaltake TR2 series is pretty garbage, but a new PSU still shouldn't be doing that. The pink/white thing sounds like either the GPU/VRAM being unstable or the VRAM itself being faulty, could be a lot of things. Did you overclock anything? Both of these issues could be caused by bad ripple regulation, if you can still return the power supply I would, for only a little more money you can get something WAYYYY better.

Accidently downvoted oops. Nothing is overclocked. The graphics card is used, but I was told it works fine. I reapplied thermal paste which lowered the temps by a few degrees but it still went to the pink screen. I just bought this PSU, I really don't want to go through the process of returning it on amazon. Is there any other solutions besides getting another one?

 

[ethanwbeardsley]

If I had to guess I'd say the power supply unit is no good. Have you tried swapping the old GPU back in to see if it works ? Honestly I didn't know they were still making 970's...

Thanks for the answer. Yes, I tried swapping the old card in, ran heaven benchmark, and it was fine. I can try doing some extended gaming with it, but i doubt it'll crash. The 970 is used, got it for 120 euro.

 

[chemmajorp53]

If the PSU is faulty, then no there's no other solution.

 

[ethanwbeardsley]

If the PSU is faulty, then no there's no other solution.

I'll get a new PSU and see what happens. What do you reccomend? I need a foolproof solution here.

 

[chemmajorp53]

Getting a new PSU isn't "foolproof," it could still easily be the GPU itself. That said I'd recommend something like the Corsair CX550M or Seasonic M12ii 520W on the low end and something like the EVGA 550 G3/G2 or Seasonic Focus 550W on the higher end.

 

[pete_101]

The other thing you could try, is find someone who will let you test the card on their computer. Obviously they will need to have a good PSU to rule that problem out.

What worries me is the fact it's a used card, and there is no way of knowing if it's the card or your PSU, and I'd hate for you to spend a lot of money replacing something that's not the problem......although a good PSU is always recommended.