PC turning off then on, when gaming and GPU stress testing

black_dota

Reputable
Apr 8, 2014
13
0
4,510
I noticed a problem with my PC since yesterday (wasn't spending a lot of time gaming lately, last time I remember gaming was like 2 weeks ago and it worked perfectly.

Specs:

AMD Radeon HD 7750 1GB gpu
i3 2120 3.3ghz cpu
H61m p20 mb
8GB RAM
1TB hdd
LC500H-12 500W psu

Basically, i downloaded a game called Deceit, started up, played for a min, then my PC turned off, then after few seconds turned itself on, no BSOD, just asked me if i want to run in Safe mode or normal.

https://steamcommunity.com/app/466240/discussions/1/1736595227852330728/

Here i explained what i tried, to fix the problem and whole event timeline.

It also happened on a game called Ring of Elysium.

I never played those games before.

I tried to diagnose with various software, Furmark, Prime95, HWMonitor, WhoCrashed, CoreTemp...

Prime95 i ran for a few minutes, didnt notice anything strange, temp was at about 70c when i stopped the test, after like 5min running.

HWMonitor was showing some weird voltages, for 12v it showed 8.096v and it didnt move at all. In BIOS it showed 12.084

CoreTemp showed normal temps.

WhoCrashed didnt generate any crash dumps.

Furmark had some interesting results. When i first ran the gpu stress, with preset 1080p, it didnt crash until it was at 95%, temp was at 67c. Then i tried both fullscreen and windowed, 800*600 and 1920*1080, my PC started crashing on start of the tests.

Tommorow i will get multimeter to test psu, i also opened the case and it looks like my cpu fan is full of dust, i cleaned psu and case like 3 weeks ago, but didnt touch mb gpu and cpu.

Also went to see what device manager had to say, and it had some PCI device warning which i fixed with Intel Engine Management driver and an unkown device which i dont know how to identify, but im pretty sure its not the issue. Ofc i tried newest drivers, stable drivers and optional drivers (for gpu) and it didnt work.

One thing i noticed, before since i bought the PC (2013 i think) i had an issue with gpu and its drivers.
I had an error saying: "Display driver amd driver stopped responding and has recovered succesfully" now i dont see that error any more but my pc instead crashes (might be completely wrong on this one).

Event viewer reports critical kernel power error 41 (63) and an error before that saying windows recovered from an unexpected shutdown with a table of memory codes (in the friendly view).

It only does this when gaming and gpu stress testing, from what ive noticed so far.

Sorry if its too much info, i just want to help you help me.
 

InvalidError

Titan
Moderator
AFAIK, L&C (if that's what it is) is a very low quality PSU brand. I'd start with replacing the PSU with a much better quality one like a 520W Seasonic S12-II which often retails for under $40.

A poor quality or dying PSU will often cause PCs to reboot while under load, rapidly changing load or even for no apparent reason.
 

black_dota

Reputable
Apr 8, 2014
13
0
4,510


It was recommended to me by a guy working in the shop, few years ago when my old PSU died.
It did its job perfectly, never had any problems.

I also checked on the multimeter, when my PC is idle, 3/4 show 12.40V only 1/4 shows 7.7V but i guess that cant be the one i need to check because PC couldnt even turn on at that point.

Im pretty sure its not PSU.
 

InvalidError

Titan
Moderator
Measuring DC voltages isn't enough to determine that a PSU is doing its job properly, you also need to measure transient response, output voltage regulation and noise. If the PSU fails to correctly regulate voltages through changing load conditions, you get a random shutdown or reset even if the nominal DC voltage still looks ok. You only need a momentary power glitch (as short as sub-millisecond, far too fast for most meters to catch) to trip local voltage regulators' under/over-voltage protections.

For example, one of my old PSUs had dead caps on the 5VSB output. The output still measured about 5V as expected on my meter but when I looked at it using an oscilloscope, I could see the output peaking at 7V. After replacing the 5VSB caps to fix the spikes and get the PSU to turn on reliably, I got similar spikes to ~1.5X the nominal voltage on the primary outputs and changed those caps as well.
 

black_dota

Reputable
Apr 8, 2014
13
0
4,510


Sorry, i didnt proof read what i wrote in the last post.

I checked both idle and while stress testing, when idle it showed 12.40 and when running Furmark it showed 12.30.
I still believe its not a PSU issue, because another strange thing happened.

Last night i opened the case, tried to look for any defects, like friend cables, loose plugs, etc.
I disconnected the power cable and turned of the 0/1 power button, and left the PC case opened, changed its position and left it that way over night.

Today, i tried stress testing for like an hour, played the game that caused problems for hours, not even one time the PC turned itself off and on, i even checked the Event Viewer.

My guess is that dust was the problem, its the only logical thing imo...

If i get problems again i will post again.