Yet another restart problem (after research)

[Lt Dan]

Of course this would happen right before Total War Warcraft is coming out, haha.

I built my PC about 1.5 years ago and has ran like a champ. But now it restarts during games. Have went through all the other posts with no luck

• 
  Intel Core i5-4670K 3.4GHz Quad-Core Processor
  EVGA GeForce GTX 970 4GB Superclocked ACX Video Card
  Zalman LQ-310 Liquid CPU Cooler
  Asus Z97-A ATX LGA1150 Motherboard
  Kingston HyperX Fury Red 8GB (2 x 4GB) DDR3-1866 Memory
  Crucial MX100 256GB 2.5" Solid State Drive
  Seagate Barracuda 2TB 3.5" 7200RPM Internal Hard Drive
  Thermaltake Smart M 750W 80+ Bronze Certified Semi-Modular ATX Power Supply

I overclocked about a month aftrer I had it, but turned it back down about 5 months ago. I haven't made any hardware changes at all, but now the pc will restart either at the main menu or when gameplay starts. Was working fine just a month ago.

I've been using GPU-Z and temp hits 50C, and seems when the load on the GPU gets to 99% it shuts down, but still not sure its some kind of fault there.

Another fun bit is I ran 3D mark, and the whole benchmark ran through without crashing.

Also checked my SSD and its fine (diagnostic as well as ran game off it)

 

[bignastyid]

Likely failing caps in the PSU. The smart series uses cheap caps, load and heat will degrade them quickly. The most issues I see when secondary caps start to fail is shutdowns/restarts under load.

 

[Lt Dan]

So another fun wrinkle, I just played the Witcher 2 for about 10 minutes with high graphics with no problems. Its a GOG game. Loaded up Rome 2 from Steam and it crashed right when I clicked "Custom Battles". Is there and chance this is a strange Steam thing? Going to try and run some games outside Steam some more

 

[bignastyid]

Try running furmark and a cpuburner at the same time to load up the system, if its what I think it is that should cause it to crash. Have you tried watching the cpu temps aswell?

 

[Lt Dan]

Try running furmark and a cpuburner at the same time to load up the system, if its what I think it is that should cause it to crash. Have you tried watching the cpu temps aswell?

Yep, right when I start GPU stress test it crashes. So you think its a bad PSU? The last log from GPU-z is
GPU Core Clock [MHz] , 135.0
GPU Memory Clock [MHz] , 162
GPU Temperature [°C] , 28
Fan Speed (%) [%] , 35
Fan Speed (RPM) [RPM] , 1412
Memory Used [MB] , 480
GPU Load [%] , 0
Memory Controller Load [%] , 7
Video Engine Load [%] , 0
Bus Interface Load [%] ,
Power Consumption [% TDP] , 10.5
PerfCap Reason [] , 16
VDDC [V] , 0.8750

 

[bignastyid]

That shows the GPU in low power mode. If I was working on the system my next step would be to drop in my spare PSU, but I do realize not everybody has one of those lying around. I do believe the issue is most likely a failing PSU.

 

[Lt Dan]

That shows the GPU in low power mode. If I was working on the system my next step would be to drop in my spare PSU, but I do realize not everybody has one of those lying around. I do believe the issue is most likely a failing PSU.

So what shows it is in low power mode? Also is there any software you would suggest to verify that thats the case? I have a 5 year warranty.

 

[bignastyid]

That shows the GPU in low power mode. If I was working on the system my next step would be to drop in my spare PSU, but I do realize not everybody has one of those lying around. I do believe the issue is most likely a failing PSU.

So what shows it is in low power mode? Also is there any software you would suggest to verify that thats the case? I have a 5 year warranty.

The clock speeds in the log from GPUZ indicates the gpu is is low power mode, This is the mode the gpu usually goes in when idle so that it uses less power. There is no software that can properly test a PSU. To properly load test a PSU requires some rather spendy hardware.

 

[Lt Dan]

That shows the GPU in low power mode. If I was working on the system my next step would be to drop in my spare PSU, but I do realize not everybody has one of those lying around. I do believe the issue is most likely a failing PSU.

So what shows it is in low power mode? Also is there any software you would suggest to verify that thats the case? I have a 5 year warranty.

The clock speeds in the log from GPUZ indicates the gpu is is low power mode, This is the mode the gpu usually goes in when idle so that it uses less power. There is no software that can properly test a PSU. To properly load test a PSU requires some rather spendy hardware.

I see. So that might not have been the best log, but the end result could be the same. In other tests the GPU gets up to its boosted 1442 speed. But there are times it crashes (like in the log I showed) right when I push the button before anything gets up to speed, and others where I can play something for about 5-10 minutes, still heat doesn't seem to be an issue

 

[Themastererr]

Honestly just sounds like a GPU driver issue to me. It's possible the card is on it's way out, but unlikely.

Completely clean drivers > reboot > re-install.

Other possibilities are PSU (check voltage rail), and VRM's over-heating. Put your hand on the VRM's when your under load (be careful please), if they are burning hot to the touch, that could be your issue.

 

[Lt Dan]

Thanks for everyones help. Grabbed a $15 PSU tester from Frys and it showed higher than normal voltage error. Swapped out with a new one I grabbed and things seem to be back to normal! Now time to try and send this one out for warranty.