Steam games suddenly shut down (Solved)

[Neck]

Hi,
I have a problem where my steam games suddenly shut downs without any error messages after gaming 15min - 1hour.
This happen with any games i have tried, CS:GO, Just cause 3, Lego harry potter, so i think game specs doesn't matter.
I have re-installed whole windows, all applications and games. And its only game that shuts down, not my PC.
I have GTX 1070, 16GB ram, Ryzen 7 1700, 550W PS, 500GB SSD.
I have tried also running 3Dmark Fire Strike Extreme Stress Test, and my CPU temperatures are max 47c and GPU max 65 celcius.
Any help? Is my power supply big enough?

[Sapph]

Yes, your power supply is more than enough wattage. THe system doesn't use more than 300W.

Have you done any overclocking on any part?

[Neck]

I haven't done any overclocking. Everything should be in normal settings.
I just had CPU-Z logging my temperatures etc while i was playing. Game worked 20 minutes and then shutted down and logging doesn't seem to have anything strange. CPU-Z log can be found here [shrib.com] you have to copy that and paste your own PC notepad so you see it normally. Gaming start 17:46 and game shut down at 18:06.

[Sapph]

It seems your gpu driver might be crashing.
Download DDU (DisplayDriverUninstaller) and use it to remove your gpu drivers. Then do a fresh install.

Also, what's the model of your power supply?

[Neck]

I just did fresh windows install, isn't that the same thing? I can also try that one. I am now doing memory check just in case.

My power supply is Corsair model CMPSU-550VX

[Sapph]

Your power supply is REALLY old. I wouldn't be surprised if it can't handle the gpu load for long time.
It's 8 years old and even back then it was fairly low in quality.

[Neck]

Yeah its about that old *eh :D I haven't upgraded it becouse it is still 550W and i know PC parts power consumption is lower than 8 years ago. So i thought it can still handle new parts. Maybe i was wrong. Can i do some kind of stress test to my power supply somehow?

[Sapph]

引用自 Neck：

Yeah its about that old *eh :D I haven't upgraded it becouse it is still 550W and i know PC parts power consumption is lower than 8 years ago. So i thought it can still handle new parts. Maybe i was wrong. Can i do some kind of stress test to my power supply somehow?

Prime95 with Small FFT should do the trick.

[Neck]

Prime95 has now been running 4 hour and not any errors occured

[Neck]

Do you still recommend buying new power supply?

[Sapph]

引用自 Neck：

Do you still recommend buying new power supply?

Could be it's only the PCI-E extra power that can't provide proper power. So you could try running Unigine Heaven at as highest settings and just let it roll. (instead of pressing Benchmark)

[Echoz]

引用自 Neck：

Prime95 has now been running 4 hour and not any errors occured

Prime95 won't use your GPU. I'd say try running a Unigine benchmark (e.g. Unigine Valley) next to Prime95. If nothing crashes it's highly unlikely that your power supply isn't supplying enough power.

[Neck]

引用自 Echoz：

Prime95 won't use your GPU. I'd say try running a Unigine benchmark (e.g. Unigine Valley) next to Prime95. If nothing crashes it's highly unlikely that your power supply isn't supplying enough power.

Okey i had Unique superposition and prime95 on for 51minutes and then prime95 detected failure. Unique still continued normally.

[Nov 6 19:13] FATAL ERROR: Rounding was 0.4938283781, expected less than 0.4
[Nov 6 19:13] Hardware failure detected, consult stress.txt file.
[Nov 6 19:13] Torture Test completed 7 tests in 51 minutes - 1 errors, 0 warnings.

After this i had unique running 3 more hours without any problems. And just to make sure i had 8K settings for unique.

[Echoz]

引用自 Neck：

引用自 Echoz：

Prime95 won't use your GPU. I'd say try running a Unigine benchmark (e.g. Unigine Valley) next to Prime95. If nothing crashes it's highly unlikely that your power supply isn't supplying enough power.

Okey i had Unique superposition and prime95 on for 51minutes and then prime95 detected failure. Unique still continued normally.

[Nov 6 19:13] FATAL ERROR: Rounding was 0.4938283781, expected less than 0.4
[Nov 6 19:13] Hardware failure detected, consult stress.txt file.
[Nov 6 19:13] Torture Test completed 7 tests in 51 minutes - 1 errors, 0 warnings.

After this i had unique running 3 more hours without any problems. And just to make sure i had 8K settings for unique.

Seems highly unlikely that it's a problem with the power supplied/overheating/anything related to stress.
Have you checked event logs at the time when the crashes have occured? (eventvwr.msc)

[Neck]

引用自 Echoz：

Seems highly unlikely that it's a problem with the power supplied/overheating/anything related to stress.
Have you checked event logs at the time when the crashes have occured? (eventvwr.msc)

I checked and in system log i have couple messages at same time when game shutted down

• Process C:\Program Files\Bitdefender\Bitdefender Security\bdservicehost.exe (process ID:2672) reset policy scheme from {381b4222-f694-41f0-9685-ff5bb260df2e} to {381b4222-f694-41f0-9685-ff5bb260df2e}
  
• The start type of the IP Helper service was changed from demand start to auto start.
  
• The start type of the WMI Performance Adapter service was changed from disabled to demand start.
  
• The start type of the Google Update Service (gupdate) service was changed from demand start to auto start.

[Neck]

引用自 Sapph：

It seems your gpu driver might be crashing.
Download DDU (DisplayDriverUninstaller) and use it to remove your gpu drivers. Then do a fresh install.

I did this, but it didnt help.

Edit: I hasn't installed my motherboard basic programs from cd. So i installed Gigabyte app center and system information viewer and put some alarms on for low +12 voltage.
When i was gaming i got first alarm after 1hour that +12v was under 10.2 volts. It didnt save that information anywhere so i am going tomorrow put recordin on for voltages and see if that was right. But hey something positive, game didint crash even playing over 1 hour.

Also i saw my ddr freq vas only 2200MHz. I put it to 3200MHz becouse thats my RAM normal freq.