[SOLVED] Faulty PSU or GPU

[nerevah]

Hello. About 6-9 months ago my PC (build below) started restarting while gaming. It would lose the video signal 1st, then after 1-2 seconds (while sound still working) the PC would restart.

Gaming with medium graphic details would make the PC not restart as often. Gaming with lowest graphic details would solve the problems for good. Stressing the GPU with 3DMark or Furmark would cause the PC to restart.

I immediately thought of a PSU or GPU problem (GPU crashing under heavy load or PSU not being able to provide enough power to a GPU under heavy load). I checked my temps with HardwareMonitor and CPU stays under 70c and GPU stays under 65c.
I tried new Mobo bios, new GPU firmware, new drivers, new windows install, unplugging all the peripherals but for mouse and keyboard, moving the GPU to another PCIe slot. Nothing helped.

I had the GPU tested by a work colleague and he said the GPU works just fine, as he gamed and also benchmarked it (I trust his skills with PC hardware).

Knowing the GPU is not faulty I bought a new PSU (Corsair TX750M), but the restarts persisted. I tried lowering the graphics card’s clocks and power with MSI Afterburner and only bringing it down to 50% power would make it stable. So I concluded it’s the GPU after all, my colleague who tested it failed! (I shamed him so hard next day at work!!!)

With nothing else at had I also bought a new GPU (Gigabyte RTX 2080 Super). With new GPU and PSU, my computer’s problems were solved and I gamed like a maniac for 5 days at QHD, 144Hz, max details.

I then sent back the new PSU to recover my money.

Much to my surprise, with the old PSU and new GPU the PC would restart again while gaming with max details, but in a different way. The restart would be sudden, just like the power was cut off (I had a PC with a PSU not powerful enough to support the GPU under heavy load and it restarted just like this).

Again, gaming with medium graphic details would make the PC not restart as often. Gaming with lowest graphic details would make the PC stable. Bringing the GPU down to 50% power with MSI Afterburner would also make it stable. Stressing the GPU with 3DMark or Furmark would NOT cause the PC to restart anymore! I even compared the 3DMark results online and they are in line with similar hardware.

The only combo that worked was new GPU and also new PSU. At least we know that!
Could it be that the old PSU was faulty and also broke my old GPU? (none of them is under warranty anymore, the PSU is 7 years old and was heavily used, for crypto mining included).

What is there to be done besides considering both the PSU and GPU dead and using new ones?



Original PC build

CPU – I7 8700k

Mobo – Asrock Z370 extreme

RAM – 2x8Gb DDR4 Patriot 3600Mhz

SSD – Team Group Cardea M.2 PCIe

GPU – Asus GTX 1080 Strix

PSU – Corsair AX650M

Monitor – Asus ROG 27 inch, QHD, 144Hz.



Replacement parts

GPU - Gigabyte RTX 2080 Super

PSU - Corsair TX750M

 

[nerevah]

Testing the PC without a dedicated GPU, using only the onboard graphics card and eighter of the PSUs, proved to work just fine.

 

[Vic 40]

Did you test the gpu's with two cables? Instead of one with two connectors. Mean with older psu as well.

Could it be that the old PSU was faulty and also broke my old GPU?

What is there to be done besides considering both the PSU and GPU dead and using new ones?

Might well be on the money with this assumption.

Could take the old gpu apart and give it a good clean like the next person does,maybe it helps.
The Tech YES Lovin' Process in 2019 - (Build a Used PC Faster than ever) Tech YES City Geverifieerd • 14K weergaven Nieuw
It's abit rigorous, but maybe it will help. From what i've seen from hi did he manage to get some gpu's working that way again.

 

[nerevah]

Ok half the problem is solved.
I took all the components out of the case for a close study and noticed a switch (nornal/hybrid) on the side of the PSU where the modular cables stick in. The switch was on hybrid mode and the PSU’s fan would not spin at all, causing it to overheat and reset not to melt. Switching it to normal mode made the fan spin.

I tested the PSU with the new GPU and everything works as intended, i gamed for 10 hours straight with max details.
I also tested the PSU with the old GPU and the PC still restarts, but not as often.

So the question that remains is:
Could an overheating PSU damage my GTX 1080?
I will take the GPU apart tonight to see if i notice something wrong with it. Maybe it can be repaired.

It needs to be mentioned that i clean my PC for dust very often, i changed the thermal paste on both my GPUs and they were tested with 2 cables not one cable split into 2 connectors

 

[Vic 40]

Might be that ripple gets (too) high with a psu that gets too warm. That could cause damage in time. You wouldn't expect that from an AX, but it's possible.

 

[nerevah]

After 2 days of gaming without problems my pc is restarting again.
I have decided to ordered a new PSU to be able to go on with my things and call it a day.
I guess the topic cand be closed, the mystery of what happened remains unsolved.
P.S. I opened the gtx 1080 and i could not notice anything wrong with it.

 

[Vic 40]

Might be alot of dust build up in the old psu after 7 years, can look if you can get the cover off and blow it out, maybe it will help. Don't touch any components in there. If using a can of compressed air might there be some condensed water so let it dry.

 

[nerevah]

There was some dust but not alot. I clean my pc at least once a year with canned air. That’s how i hit the switch of the PSU by mistake and put it on hybrid. Also the case has a filter where the PSU fan is.

 

[Vic 40]

Hope the new psu takes care of the problems.