Question WHEA UNCORRECTABLE ERROR on RTX 3070 and RTX 3090

Jun 24, 2021
2
0
10
Hey,

So this one might require a little of explaining so i'll dive right in

Over 1 year ago I built this PC with:
Threadripper 3970x
Corsair HX1000 PSU
Samsung MZ-V7S1T0BW SSD 970
64GB Corsair RAM
ASRock TRX40 Creator Gaming ATX Mainboard

I also had a GTX 1080 which I kept using because I was waiting for the rtx 3000 series cards to come out.
About 6 months later I finally got my hands on a Zoltac RTX 3090. And up until this point I have never had any errors on my PC.

My RTX 3090 was a freaking beast! and worked smoothly for about 3 months until I started noticing an error WHEA UNCORRECTABLE ERROR. I kept an eye on temperatures and everything was fine.
I kept doing research and did ALL kinds of tests, RAM tests, HDD and SSD tests, I would even run the GPU through GPU intensive tests to see if it would throw the error.
The error however was entirely random, nothing I did would cause an error, and the error would (at the time) only happen once every few weeks.

This issue eventually got worse and worse and made my PC unusable with restarts every 2 minutes or less. I ended up sending the GPU back to see if they could repair it. Which they couldn't, so they gave me my money back. (about a 6 week process which sucked)
I bought the GPU before the prices were highly inflated so sadly the money they gave me could now only afford a Zoltac RTX 3070. I didn't want another Zoltac but there was litterally no other choices in this GPU shortage.


SO, I inserted fit the new RTX 3070 and instantly the constant errors had stopped. HURRAY!! I could finally have a decent GPU and it was absolutely perfect until... you guessed it, 3 months later the exact same error started happening WHEA UNCORRECTABLE ERROR.
This error started last week and is already getting worse and worse, and before long I know (like the rtx 3090) I won't be able to use my PC again very soon.

So I am wondering, is this a problem with Zoltac GPUs? Or could something in my PC slowly eat away at it over the course of 3 months? I rarely even use this GPU! I haven't gamed on it in ages, and only used it for rendering a few 3D images (15minutes at a time).
The temperatures on the RTX 3070 are always a cool 40/50 degrees, so that can't be the problem.




Sorry for the essay, but if anyone has any suggestion then do let me know! I don't want to send this GPU away if it means i'm just going to get a new one which will die in 3 months too. A constant 3 month cycle of GPUs
 
Jun 24, 2021
2
0
10
Oh yeah, and with the GTX 1080, this GPU outlasted both the 3090 and 3070 combined, and never caused this error.

And yes, I have reinstalled all GPU drivers multiple times on both 3090 and 3070.
On the 3090 I also tried underclocking it, which didn't help so I haven't bothered with the 3070 yet