[SOLVED] GPU or PSU Failing?

imoob

Reputable
Nov 7, 2018
68
7
4,545
A couple months ago i started experiencing random crashes where my display would turn black, my GPU fans would (sometimes) stop spinning but my PC's LED's and fans would continue to spin.

I thought nothing of it and just turned it off and on again and continued to do so whenever i experienced said problem.

As of only a couple weeks, the crashes have become more consistent and still just as random. I could be browsing the internet or gaming and it would crash. I suspected it to be my GPU/PSU so tried it in my brothers computer. However, i cant pinpoint whether its my GPU or my PSU since my brother has the same PSU that i use. I also cant test my PC without the GPU since i dont have integrated graphics and i dont have access to another graphics card but, im almost 99% sure its my GPU or the PSU's since whenever i put the card in his PC, it experinences those random crashes.

When i say the PC crashes randomly, i mean ive managed to run stress testing on the GPU and CPU at the same time for over an hour and the PC didnt crash but sometimes it crashes 5 mins after boot-up

Also, ive tried running a GPU + CPU stress test with HWiNFO64 running and my average +12V is 11.616V with a minimum of 11.520 V (not sure whether these are accurate since i dont have access to a multimeter)

My Graphics Card is a GTX 980 ti (Windforce) and my PSU is a Coolermaster masterwatt 500w.

Im not sure if 500w is enough for a 980ti and a ryzen 7 1700 and if i should look into a better quality PSU (I wass looking at around a 650w more reputable PSU)

Let me know what you guys think. Thanks!
 
Solution
Looks like your gpu is the culprit, try swapping the gpu and see if your system is booting, if it does then it's you gpu might be the culprit

iPeekYou

Distinguished
Jul 7, 2014
392
77
18,790
Also, ive tried running a GPU + CPU stress test with HWiNFO64 running and my average +12V is 11.616V with a minimum of 11.520 V (not sure whether these are accurate since i dont have access to a multimeter)

My Graphics Card is a GTX 980 ti (Windforce) and my PSU is a Coolermaster masterwatt 500w.

Im not sure if 500w is enough for a 980ti and a ryzen 7 1700 and if i should look into a better quality PSU (I wass looking at around a 650w more reputable PSU)

Let me know what you guys think. Thanks!

Seems to me it's the PSU. I know HwInfo isn't 100% accurate for voltage reading, but 11.520V-11.6V is awfully close to the minimum limit at 11.40V. I suspect your random crashes occur when the voltage spikes below or over the 5% threshold and the PSU shuts itself off by means of its protection system.

On top of that, I plugged in your specs to bequiet's PSU calculator and the calculator says you need just under 400W to power both GPU and CPU at the same time. Might seem fine, especially with high tier PSUs, but being near the maximum deviation in voltage on +12V rail could cause the system to lose stability.

Mind that bq PSU calculator is quite conservative, if you use outervision it'll give you higher wattage requirement. I also put no OC on my calculations, and AIB GPUs are often overclocked out of the box.

There's also the chance the 980 Ti is degraded from bad PSU. But the way I see it, you're in need of another PSU anyway.
 

imoob

Reputable
Nov 7, 2018
68
7
4,545
Quick update:

I bought a new PSU (seasonic core GM 650w) and the system booted to windows with no issues. I even managed to leave the pc on overnight to see if it would eventually crash but it didn't.

At this point I thought I had fixed the issue so I played some GTA and it was fine. I then loaded up MSI kombustor as I was hoping to do a GPU overclock. After around 5-10 minutes of running kombustor and tweaking on afterburner, the pc crashed to a white screen (bear in mind GPU temps were around 68 C)

Since then, I haven't been able to boot into windows using the graphics card drivers. Bear in mind I have deleted afterburner in safe mode. I can boot into safe mode and use ddu - then I can boot into normal mode. During the GPU driver installation however, the pc crashes to a black screen.

The card itself shows no visible damage on the pcb and I have cleaned any dust and replaced the thermal paste using mx-4 but I don't think I'm experiencing overheating problems

I'm not sure if using kombustor fried the card. If there's anything else I can try do please let me know otherwise I guess I'll have to replace the card :( . I'll also be able to test my pc using a gtx 970 soon.
 

imoob

Reputable
Nov 7, 2018
68
7
4,545
Looks like your gpu is the culprit, try swapping the gpu and see if your system is booting, if it does then it's you gpu might be the culprit
I think that if you put the GPU in your brother's machine and it crashes....AND....you have tried another PSU.....I'm leaning on that it's the GPU.

Would you have any idea why the GPU will have all of a sudden decided to not work. I've heard furmark and the like are notorious for killing GPUs but I've used it numerous times on different GPUs in the past with no repercussions.

Also would there be any way to know for sure whether it's software (driver conflicts etc.) or hardware.
 
"Would you have any idea why the GPU will have all of a sudden decided to not work. "

There comes a point where pretty much all things must fail.

....and also....things tend to die when they're stressed but not necessarily.

"Also would there be any way to know for sure whether it's software (driver conflicts etc.) or hardware. "
I think if you put it in another machine and it does the same thing.....chances are much better that the problem is hardware.
 

coolraveen

Commendable
May 26, 2020
321
52
1,790
"Would you have any idea why the GPU will have all of a sudden decided to not work. "

There comes a point where pretty much all things must fail.

....and also....things tend to die when they're stressed but not necessarily.

"Also would there be any way to know for sure whether it's software (driver conflicts etc.) or hardware. "
I think if you put it in another machine and it does the same thing.....chances are much better that the problem is hardware.
For the driver or software to be bad, at least you should be getting some sort off display. I had an amd hd7750 years back it was new and I was using it for a couple of years, then the display started to flicker, I tried reloading the drivers and bios update but nothing worked, then I swapped the gpu, which was my old gt 430. The system worked flawless, later I found that the 7750 was faulty at chip level. So when it comes to electronics you can never say. So when I saw the corroded screws of the gpu you showed, screws don't corrode just like that, in an enclosed cabinet. So gpus are sensitive.