Inconsistent results with benchmarks (Computer Crashes and Restarts/Runs benchmark for hour)

[anfalsiddiqi]

My computer started crashing randomly around 6 months ago. While playing a game, especially when loading into fortnite, the computer would crash with a black screen and a weird noise then restart. I thought it was a corrupt installation so I reinstalled the game but it kept happening so I stopped trying to run fortnite, since it was still in beta.

Around 4 months ago the same type of crash started happening with other games and more frequently. While watching a video, or just surfing the web, my computer would crash with a black screen and restart. It's come to a point where it could crash 10 times in an hour or go 5+ hours without crashing. Since then I've been trying to fix the crashing and find the faulty part. Probably my old GPU.

Things I have noticed/Testing results:
1. When I let the PC sit still for an hour after it crashes it doesn't crash again for a few hours.
2. The temperatures while Idle are around 40 C for both GPU and CPU.
3. Kernal Power 41 63 in the event log, which just says a sudden loss of power etc. Not very helpful.
4. Valley Benchmark does not crash the computer and GPU temperatures are around 70 C.
5. Prime 95 does not crash the computer and CPU temperatures are around 70 C.
6. Windows Memory diagnostic showed nothing wrong.
7. HD Tune showed nothing wrong with SSD.
8. Only thing that crashes my computer is Furmark test, but only sometimes. Around 50% of the time the benchmark will start and once I ran it for an hour, until I closed it. Other times the computer crashes and restarts. These inconsistent results keep me believing that the GPU is not faulty. GPU temps were max at 75 C.

Things I have tried/Results:
1. Reset windows > Continued to crash
2. Installed Nvidia drivers again after resetting windows. Also tried several old drivers, 1 year old, 2 year old etc. > Continued to crash
3. Updated Bios and cleared CMOS > Continued to crash
4. Applied new thermal paste to CPU, took RAM out and cleaned the ports and put it back, took GPU out cleaned the port and put it back, and cleared the CMOS > Continued to crash
5. Used MSI Afterburner to reduce GPU clock and memory to the lowest and set the power limit to the lowest/highest > Computer crashed less often. Again the inconsistency of the crashing makes me believe GPU is fine. It was crashing once a day.
6. Took CPU back to default clocks > Continued to crash.
6. Set Advanced power option for Processor Power Management to Passive since I was getting a Kernal Power Processor error in event log > The error stopped getting produced and computer crashed less often. It was crashing ever few hours.

What do you think might be causing the inconsistent results? Leaving my computer idle reduces the frequency of crashing, and I usually link this with over heating but my temps never exceed 75 C. If the GPU was faulty the furmark test would always crash the computer, and why does the Valley Benchmark not crash the computer? Changing power options for CPU made it crash less, so maybe the CPU is bad too?

Is there any way I can test my PSU, since I think it might not be providing enough power to my GPU and CPU?

Specs:
MSI Z97 PC Mate (MS-7850)
Intel Core i5-4690K @ 3.5 GHz
ASUS Nvidia GeForce GTX 660 Ti (Clock: 915 Mhz, Memory: 2048 MB)
8GB Ram @ 1600 MHz
EVGA SuperNOVA Nex 650
OCZ Vertex Plus

Any help would be appreciated, Thank you in advance.

 

[DRagor]

I would start with replacing motherboard battery. Such inconsistent crashes is something that can be caused by dead battery. Other then that, your only chance would be to swap components one by one and see if it still crashes, as unfortunately your tests are not showing the guilty part.