Hello, there.
I've was lucky to get a spot on the 3080 XC3 Ultra queue and have had the card since November last year without a problem... until now.
About a month ago I was playing HotS (Heroes of the Storm) when the screen suddenly went black. I could still hear my teammates on discord, they could hear me fine, but apparently I disconnected from the game. At first I thought it was a driver crash, but the screen never came back. I had to do the one finger salute. After a reboot I rejoined the match and all was well for a few weeks, when it began to turn into a common issue. It happened at least one time during the first match of every game. Since it got annoying, I took a look at the Event Viewer to see if there was anything there... but no, there's nothing. (Yes, there is, something about a webcam frame server thing... but it is not related.) HotS is not a GPU demanding game - more CPU heavy. Eventually, it happened randomly several times a match. Then it began happening in FF XIV, Cyberpunk, and more; to the point that now I can't event start up GPU heavy games.
I have a 5900X with an Asus Strix X570-E Gaming. I had recently upgraded the BIOS to 4002 which supposedly helped with the USB problems 570X's seem to be having. I have all the USB ports populated and indeed, sometimes the USB DAC would disconnect. Thinking that was the issue I rolled back to 3604 with which I had not had any problem other than random USB disconnects.The GPU issue was still happening - actually, I'm still on the 3604 BIOS; what I did is I disconnected all USB devices I was not using to see if that helped. But no, it didn't.
Then I went and changed all PCIe lanes to Gen 3, just to be sure. Nope, problem still there. Then I took the PCIe NMVe drive I had, making sure to just leave the GPU plugged in. Problem still there.
I have a secondary PC I use in my living room for couch gaming and VR with a 2080 Ti. So what I did was put the 2080 Ti on my main PC and the 3080 on the living room PC. Uninstalled drivers with DDU and installed them back on. And I have not had a problem with any of them since. This living room PC has an AMD 3600 and an Asus 470X-I.
So... the only variable I think I have to check but don't know how since I don't have spares around, is the PSU. The main PC has a EVGA SuperNOVA 850 G2 from 2016; the living room PC is an ITX system, so it is rocking a Fractal Design Ion SFX 650G. Which is interesting because for a second there I thought the "small" 650W PSU would not be enough to drive the 3080. Oh, and on the main PC I have a 4K display, while the living room is using a 1080p TV - I know, I know... I need a TV upgrade... I don't know if the TV not demanding as much power from the GPU keeps it from pulling too much power.
Anyway... is there a way I could pin point if the 850W PSU is the culprit? I don't want to needlessly buy a PSU in this current market. I wouldn't think 5 years is out of a PSU's lifespan but I don't know what else to point to. Because if it's not that, then I guess I'd have to RMA the GPU... which would be a bummer.
Thanks for any advice!
Edit: Oh, and yes, I have always used two separate cables to plug the GPU to the PSU
Edit 2: Just in case, I have 4 sticks of 8GB Crucial Ballistix 3600 MHz DDR4 running at... 3600MHz, hah! No overclocks there. Heck, there's no overclocks anywhere, not the CPU nor GPU. I tried running the RAM without DOCP which puts it at 2600 MHz, and even tried only using 2 sticks since I know having all 4 sometimes make computers misbehave. No change on the GPU issues.