Computer shuts down (no BSOD) during demanding games

syclick

Distinguished
Jul 25, 2009
14
0
18,510
So I am running the following:

Mobo: Gigabyte GA-Z77X-UD3H mobo
CPU: Intel 3570K (not OCed)
PSU: Seasonic SS-1050XM
RAM: Corsair Vengeance Blue 16 GB DDR3 SDRAM (4x4GB)
GPU: EVGA GTX 970 SC ACX 2.0 (was EVGA GTX 670 FTW)

I've had intermittent issues in the last month or so with my machine shutting down without warning. To be precise, it turns off immediately - no BSOD, no apparent software crashes.

Last week, I purchased both GTA5 and SOMA, and tried playing them both, and I couldn't play for more than several minutes without the computer shutting off again. What's more, the computer would not power on without me unplugging the machine and waiting 5 or 10 minutes.

I figured it might be a GPU issue, so I replaced my old EVGA GTX 670 FTW with a newer EVGA GTX 970 SC ACX 2.0. I booted up SOMA and was able to play for about 90 minutes or so before the computer crashed again.

I don't see my GPU rising above 80°C.

In general it happens when running games, but the following have caused the issue too:

-Ctrl-Tab out of SOMA, then going back into the game. The very second I did this it shut down. (old 670)

-Furmark fullscreen. I was only able to get this to happen once. (old 670)

Any help would be greatly appreciated.
 
If you can't turn it back on for a few minutes, that's a dead giveaway for an overheating problem. 80C is normal for a GPU, but what are your CPU temps?

Odd that the CPU would be overheating without overclocking, but maybe there's an airflow issue in the case, CPU fan is starting to fail, or it's time to remove and reapply the thermal paste.
 

syclick

Distinguished
Jul 25, 2009
14
0
18,510
I ran Prime95 for a couple of hours tonight. No crashes.

My CPU temps were as follows:

Core #0 max: 80°C
Core #1 max: 87°C
Core #2 max: 83°C
Core #3 max: 82°C
Package max: 87°C

Also, mobo Northbridge maxed out at 78°C.
 

Barty1884

Retired Moderator


Weird. Slightly higher than 'ideal' temps, but nothing too concerning there - although a couple of hours may be a bit short. Might try longer & monitor temps then.


 

syclick

Distinguished
Jul 25, 2009
14
0
18,510
Will do. How long would you suggest?

I think during my next gaming session (tonight hopefully) I will log temps/voltages in the background.

Also, I've got a better CPU heatsink on the way (CM 212+) as well as a digital PSU tester.
 

Barty1884

Retired Moderator



It's all a matter of opinion with Prime95 (although there's accepted standards for OC).
As you're testing temps, I'd suggest Blend for maybe 4-6 hours to be 100% sure.

Personally though, with a new CPU cooler and PSU tester on the way - assuming they'll arrive soon, I would just hold off and see how those improve things, and then run Prime.
 
Another, less obvious issue could be that the power supply itself is overheating and shutting itself off. Maybe it wouldn't do that when the CPU or GPU alone is stressed, but only both together. With a quality PSU like that, I wouldn't call it super-likely - but if there's an airflow problem, fan not working, or a bottom-mounted PSU where the intake is covered in dust, those certain situations could cause the same kind of problem.
 

syclick

Distinguished
Jul 25, 2009
14
0
18,510


You may be onto something... I tried running Prime95 and Unigine Valley at the same time. The computer shut down immediately.

EDIT: Just tried Prime95 and Furmark at the same time. Same thing.
 

syclick

Distinguished
Jul 25, 2009
14
0
18,510
Can't get the system to power on at all now. The computer has been off all day, and won't power on at all.

The LED on the motherboard still lights up, but I get nothing when hitting the power button (either on the case or directly on the motherboard).
 

syclick

Distinguished
Jul 25, 2009
14
0
18,510
PSU tester came in, ran a quick test. Everything seems to be in order...

dar09s73mbv2pz26g.jpg


EDIT: I've also run Memtest86 with zero errors. So PSU, RAM, CPU & GPU seem to check out fine. At this point I am left guessing that the motherboard is bad.

2ND EDIT: I am about ready to give up and build a new computer.

For a while, the PC would not make it past the mobo boot screen. After trying this several times, I put the 670 back in, and was able to get to Windows. Put the 970 back in, no luck.

Eventually, I was able to get into Windows with the 970 after waiting about 1 or 2 minutes at the mobo start screen. Everything seemed to be working... I played SOMA for about 20 minutes, then went back to the desktop. I start SOMA up again, but the game froze. I switched back to the desktop, and could do almost nothing, while still being able to move the mouse.
 

syclick

Distinguished
Jul 25, 2009
14
0
18,510
Ok, so I ended up replacing my GPU, CPU, motherboard and memory. While it did make things better, it didn't completely stop the shutdowns. I concluded that it must therefore have been the PSU. I ordered a XFX 550W PSU (a rebranded Seasonic) and I've not had one crash yet.

Luckily, my Seasonic 1050W is still under warranty. The 550W will keep me running until it's time for me to go SLI.

Thanks again to everyone for their input.