3090 Trinity / mobo/PSU troubles?

[kaloszer]

Hey,

 

For a couple of weeks now I've been having issues with my GPU. It started when the PC shut down randomly during gaming. However after the crash there was no video, the GPU was dead I thought.

I then went ahead and plugged it into another PCI-e slot, breaking both retention clips because I'm dumb. (and had a small sff PC). It started working again after I plugged it back into the first PCI-e slot (because the latter I couldn't fit the GPU in).

 

In the beginning, when the crash happened, I was trying to find anything reminiscent of issues with the GPU in the event viewer. But there was nothing really there. No nvlddmkm error either.

 

Did the usual, DDU reinstalled old drivers, and installed the newest ones. Same issue.

 

Moved my PC into a full-sized case to try the second slot (mind you both of them have the retention clip broken at this point). It works for some time (2D is fine, yt, etc.), 3D works for 1-2h of usual gaming and crashes, and temps are fine. I even repadded the memory and reapplied thermal paste on the die today - was able to play again for an hour and another crash.

 

A couple of things that are quirky at the moment:

 

- Can get through a stress test in Heaven/3dmark, undervolted to 1850/875

 

- System/Application event viewer gets CLEARED completely every crash, no clue why, can't find the reason. Nothing in task scheduler, no additional tools that would cause this. The crash log is so big that it basically causes the log to clear itself?

 

- Can no longer boot with XMP profile enabled - points to an issue with the mobo

 

- When I went ahead and plugged my PSU > GPU connector to another rail where it was in a state of 'black screen on boot, no led, no fan spin' it started to work again - points to the PSU

 

I'm planning on giving the GPU for a couple of weeks to one of my friends to see if they have the same issue on a different build, but I have a mobo and PSU cables on order which I'll try to test again with.

 

What was done:
 

- moved the PC into a larger case to test another PCI-e slot, same issue

- RAM reseated

- Rails used for GPU were changed to the CPU one and vice-versa

- Repad/ Re-apply thermal compound (max junction temp was getting toasty without the undervolt applied, but did not cause the GPU to crash, just throttle - so I'm not sure if this is the GPU)

 

What will be done:

- Change PSU cables

- Test GPU on another machine

- MOBO will be changed

 

Now the question is, which one is the culprit...

- GPU (will test soon on another machine)

- MOBO (on order)

- PSU ( just cables on order ) 

 

Specs:

 

CPU: 13th Gen Intel Core i5-13600K

GPU: 3090 trinity 24gb

RAM: 4x 8gb kit (32gb)

MOBO: Gigabyte Z690M AORUS ELITE AX DDR4 

PSU: Cooler Master V750 SFX Gold Zasilacz z Wtyczką EU - 750W 80 PLUS Gold

 

 

 

061023-9796-01.zip

[emosun]

8 minutes ago, kaloszer said:

RAM: 4x 8gb kit (32gb)

Remove all the ram and run it with a single stick , try each stick alone in the machine.

 

Having 4 dimms basically ensures theres an intermitent issue with one of them or the board just doesnt like them

[emosun]

11 minutes ago, kaloszer said:

750W

Actually yeah the 3090 also is going to want over 1000w for the power spike problems it has

[kaloszer]

1 hour ago, emosun said:

Actually yeah the 3090 also is going to want over 1000w for the power spike problems it has

I mean it's worked flawlessly for the past 6 months, running stuff like Cyberpunk 2077/God of War, everything on ultra and no problems. 

And suddenly now?

[emosun]

4 hours ago, kaloszer said:

I mean it's worked flawlessly for the past 6 months, running stuff like Cyberpunk 2077/God of War, everything on ultra and no problems. 

And suddenly now?

yep thats how most things break , they work , then they don't. be happy that worked for 6 months