[SOLVED] PSU or GPU to blame?

[elpaco]

I recently started to experience shutdowns when I was playing graphically intensive games (most notably Witcher 3). Right when the load on the GPU would be most intense (a lot of particle effects, etc), my PC would shutdown and immediately restart.

I started monitoring closely the sensor data of all components during those outages using HWinfo64, and nothing was out of the ordinary (CPU temp very stable, GPU VRM temp under 70C, no huge drop in current/power or anything) except for "GPU memory errors", of which there always were some.

I ran OCCT memtest for several hours and there was no error found.

I also tried a torture test on prime95 to make sure the issue wasn't the load on the CPU. No problem there.

At this point I'm unsure whether the PSU or GPU is to blame. The reason I am suspecting the PSU is because at approximately the same time, I started to have booting issues every time I would disconnect my PSU to look inside the case. After reconnecting it, the PC would not boot. The fans would start up but there would be no beeping and the BIOS wouldn't even boot. It would usually resolves itself by leaving the PSU unplugged for a day and trying again. Could it be that when the GPU is requiring a lot of power the PSU is unable to comply?

Here is my setup:

CPU: intel i5 3570
(http://www.newegg.com/Product/Product.aspx?Item=N82E16819115233)

MoBo: Gigabyte z77-HD3
(http://www.newegg.com/Product/Product.aspx?Item=N82E16813128582)

GPU: XFX Radeon R9 280X
(https://www.newegg.com/xfx-radeon-r9-280x-r9-280x-tdfd/p/N82E16814150678)

PSU: Seasonic S12G 750W
(https://www.amazon.com/Seasonic-S12...00LPVFOL8/ref=cm_cr_arp_d_product_top?ie=UTF8)

RAM: Kingston 2x4Go 1600MHz DDR3

I have tried replacing the GPU with a older one I had (Radeon HD 6670). I couldn't reproduce the issue with it, but then again it does require much less power than the 280X.

Also, the only other power supply I have on hand is a Corsair CX-430, and I'm afraid that 430W might not be sufficient to power the whole setup, meaning it would crash anyways and not provide much insight.


My question is: How do I figure out which component to replace? What would you suggest I do to diagnose this issue? Could it be something completely different like the motherboard? How do I diagnose that ?

I want to avoid having to spend a few hundred dollars on a new component only to realize it wasn't the issue.

Thank you very much in advance for you help!

 

[clutchc]

Short answer is.. you can't. Not w/o trial and error. I suspect the aging GPU myself. More likely than the Seasonic PSU. But the issue could be other things as well. Motherboard itself for instance, bad HDD/SSD, faulty device plugged into the PC.

I presume you've already checked GPU, CPU temps.

 

[Schlachtwolf]

Reccommended for that card is 550 w PSU so it is unlikely a power issue. I think your VRM chips in the GPU are showing their age and getting hot, hence the artifacting. Normal temps may seem ok but if the VRM is getting hot quick then it will just shut off.

If your GPU is faulty it could corrupt the booting process with error signals, no monitor detected, power issues etc etc
If it was the Mobo I would expect beeps, the SSD/HDD if it won't boot once then that is normally that....