[SOLVED] My PC turns off when playing games

[r1sh12]

Hi All,

Bit of an annoying one that I just can't pin down.

Specs:
Asus ROG Strix Z370-E Gaming
Intel Core i7-8700K (No OC) - Cooled with Corsair h100i v2
Corsair Vengeance LPX DDR4 3000 C15 4x8GB (NO XMP)
EVGA Nvidia GTX 980 Ti Hybrid
Asus Strix Soar Sound card
Samsung 970 Evo NVMe PCIe M.2 250GB (OS Only)
Crucial MX300 525GB (Game installs)
Samsung 840 Pro 128GB (Spare drive for spill over games)
WD Blue 1TB (2012) - [Storage for media and some Apps - Like office etc.]
2 x Asus PG278Q monitor
Windows 10 pro, 1903, OS build: 18632.267 [Installed 10th June 2019)

This PC had a Bequiet 650W PSU

So on Friday, playing Apex Legends @ 1440p with a roughly 100fps ,

My Pc randomly just cut off completely and then started back up.
Initially I thought it was a power cut in the house (I rarely get them, but it can happen).
Fired the PC backup checked everything was running fine, then got back to gaming with friends.
Then boom it goes off again.

Now Im worried - what is the cause?
I run a quick benchmark - everything seems to be running fine.

So I attempt to game again, while keeping Task manager, and temps on my 2nd screen.

My CPU hovers between 50 - 55 degrees C
GPU about 58 - 60 degrees C

My CPU is roughly about 40% utilised
GPU is 99% utilised (Its always been like this)

In my opinion those are ok, then the PC cuts off again.
Considering the temps are normal, I immediately start looking over my OS - maybe has a virus - run the various scans, its completely clean.

I move on to the GPU drivers, recently updated (In May) so I roll back to April.
No joy, same problem.

Now I think if the power is cutting off, maybe its the PSU - I luckily have an old Bequiet 500W dark power pro, I switch them, and start playing again today.
I manage about 1 hour of gaming, then it cuts out again, this time a little different. Most likely due to the PSU, the system cuts off, but I can see the PC is still running, the HD light is still flashing.
I attempt to RDP from another PC but no joy, after about 90 seconds the PC turns off completely.

I have ruled out PSU in my opinion.
I run a sfc /scannow which fixes two files - namely about having a reference to a location two times.
Nothing serious I think.
**
2019-07-28 20:15:54, Info CSI 000001aa Warning: Overlap: Directory \??\C:\WINDOWS\SysWOW64\drivers\en-US\ is owned twice or has its security set twice
Original owner: Microsoft-Windows-Foundation-Default-Security.Resources, version 10.0.18362.1, arch Host= amd64 Guest= x86, culture [l:5]'en-US', nonSxS, pkt {l:8 b:31bf3856ad364e35}
Both of these get 'fixed' by windows
**



I restart, re-run the scan, no fixes required its fine.
Play again, keeping an eye on task manager and temps - All temps are reporting completely normal 50 - 60 degrees C both for CPU and GPU.

Event viewer would usually be my next place to look, but due to the sudden power cut, its not committing anything to those logs.

My next thought, maybe its the RAM, CPU or even the GPU.
I run some more benchmarks and all seems fine, no issues and temps all look good.

The only thing that has changed, I had a windows update about 4ish weeks ago, but this is completely random so I just cannot think what it might be.
Maybe the CPU temp is not being reported correctly?

Edit: The PC runs COMPLETELY fine doing day to day things - netflix, web browsing, work related apps like slack etc..
Edit 2: It happens when running ALL games
Edit 3: While running 3d Mark time spy, the first test completed fine.
The second test for GPU is when it crashed out and my PC lost power again. I suspect my GPU is faulty


Edit 4: Any help is appreciated, Usually I am able to solve these myself but I am completely stumped


Looks like my GPU is busted somehow, Will continue to test and report back if I find anything else


Thanks
-r1sh12

 

[siaan312]

That sounds like a psu failure
but if you tried 2 psus, from be quiet (its 2 words. Be Quiet, not bequiet.) which is a respectable company,
I doubt its psu.
dead gpu could be it, but my question is why was it on may drivers? update them regularly.
try updating the gpu drivers to most recent. if that doesn't work... its probably those 2 options

(psu or gpu)

 

[r1sh12]

Up to date drivers are not always the best idea, Nvidia in the past couple of years have been releasing absolutely terrible drivers.
I've tried the latest drivers, May, April and even before the RTX changes (despite my GPU not having any RTX capability).
I also initially thought it was a PSU failure, but the power/ voltage going through the system is completely normal - I haven't OC'd the GPU either.
On further testing time spy fragged the PC when the second GPU test starts - It most likely is the VRAM I suspect when it loaded in the textures etc.

When running other benchmarks - they only push the chipset rather than load in data into the VRAM hence why it passed those, but timespy loads something into the VRAM - I can see it using HWMon and once it hits a certain number it craps out.
Unfortunately I could not find out at what point it fails because it happens so fast, but I definitely am not testing it again.

Anyway I have a 1080Ti coming from my office to try and will report back once I make the swap,

Thanks for the comment

-r1sh12

 

[r1sh12]

100% confirmed its my GPU.
Put a new GPU in and its been running fine for about 6 hours.

Not sure what is wrong, I have a suspicion its one of the VRAM modules

 

[King_V]

Glad you managed to pinpoint the issue - also, thanks for listing the detailed steps here. It will definitely help if someone winds up with similar symptoms in the future.

 

[r1sh12]

Not a problem, I used to work in support so anything I can add is great.
I am quite disappointed though, this EVGA Hybrid card has never really been pushed beyond its capability so for it to fail like this after warranty runs out is annoying.
But these things happen I guess