[SOLVED] 3080 Artifacting in some games, is it toast?

[Wilshire90]

Specs
EVGA 3080 FTW3 Ultra
Ryzen 5 5600x
MSI Tomahawk x570
Corsair RMx 850 Gold PSU
Spectrix XPG D50 2x8 RAM

Ill start by saying my card is at stock clock speeds, and I have done a full DDU and driver reinstall. I have 3 dedicated PCIe 8 pic connectors to the card as EVGA recommends. Ive tried removing and reseating gpu to slot. I started getting these small flashing black bars all over my menu screen in Warzone which would always lead to a dev error 6068 crash before any match could start. I chalked it up to being a problem with the game since so many others suffer dev errors and I can play other games without issue, including Cold War. I recently started playing Titanfall 2 and noticed the same type tiny rectangles flashing all over some of my menu screens. In game looks fine and the game does not crash (yet). I can play many other demanding games without issue, but have seen these artifacts flash for a split second in games including Valheim when walking into a large structure. Is there anything else I can try, or is this an RMA situation?

Titanfall menu example: https://streamable.com/z3l69m

Valheim screenshot (zoom in and you can see pairs of small parallel boxes all over screen): https://imgur.com/a/HHG9Irk


Thank you

 

[Lutfij]

Welcome to the forums, newcomer!

Can you check and see what BIOS version you're currently on for your motherboard? Seeing that you should be on Windows 10, what OS version are you on? You will also need to state the age of the PSU in your build. Dedicated as in, there are no daisy chains from the PSU to the GPU? You speak of using DDU, can you check to see what version of Nvidia's drivers you're on right now? Make and model of your monitor(s) and how they're paired to your desktop?

 

[Wilshire90]

Welcome to the forums, newcomer!

Can you check and see what BIOS version you're currently on for your motherboard? Seeing that you should be on Windows 10, what OS version are you on? You will also need to state the age of the PSU in your build. Dedicated as in, there are no daisy chains from the PSU to the GPU? You speak of using DDU, can you check to see what version of Nvidia's drivers you're on right now? Make and model of your monitor(s) and how they're paired to your desktop?

Thanks! I will do my best to answer your questions, let me know if there is anything else I can provide

-Mobo BIOS version is 1.40, 10/29/2020
-Windows 10 Version 20H2 (OS Build 19042.928)
-I bought the PSU brand new in December 2020
-Correct, there are no daisy chained PCIe cables to the GPU. 3 completely separate 8 pin (6+2) cables from the PSU to GPU
-Game Ready Driver 466.27
-2 Monitors, main one is a Samsung Odyssey G7 32" connected with Display Port cable. The other is a ViewSonic VA2265 connected with Display Port to VGA adapter. The artifacts occur on both monitors, I've tried unplugging each and running on one monitor only, with the same results

 

[Wilshire90]

Can anybody help with this? My first build and I'm not really sure what to do. Don't wanna RMA my 3080 if there is a chance something else is causing the problem. Should I mess with updating the Mobo bios?

 

[digitalgriffin]

Can anybody help with this? My first build and I'm not really sure what to do. Don't wanna RMA my 3080 if there is a chance something else is causing the problem. Should I mess with updating the Mobo bios?

Block artifacting is usually result of a memory failure. Please try lowering the memory clock and see if the issue persist.

There's a known issue with memory on 3080 and 3090 overheating (over 110C at Tjunc) If turning back the memory clocks helps, you might have to RMA it.

 

[norcalsc]

I know with Warzone that was happening in the past but I thought that was worked out with the developer by now. have you tried to reinstall the games on a different storage drive?

I just to a look at the pics you provided and that is definitely artifacting and is likey an issue that will require a RMA unfortunately

 

[Wilshire90]

Block artifacting is usually result of a memory failure. Please try lowering the memory clock and see if the issue persist.

There's a known issue with memory on 3080 and 3090 overheating (over 110C at Tjunc) If turning back the memory clocks helps, you might have to RMA it.

I had tried lowering the clocks by -150 Core and -200 Memory in Afterburner previously when I was trying to fix Warzone. I was still getting the artifacts/crashing. Should I try different values? Was worried of going too low because I read that could cause more instability.

 

[thepersonwithaface45]

I would try using HDMI or display port cables. I didn't even think the 3080s had DVI.

 

[SkyNetRising]

I had tried lowering the clocks by -150 Core and -200 Memory in Afterburner previously

EVGA 3080 FTW3 Ultra is factory overclocked card (1800 MHz Boost Clock).
Reduce overclock to Nvidia reference clocks
Check GPU temperatures. Improve cooling.

Nvidia reference clocks:

Boost Clock (GHz) 1.71​

Base Clock (GHz) 1.44​

 

[Wilshire90]

I would try using HDMI or display port cables. I didn't even think the 3080s had DVI.

They are Display Port. I dont know why I brain farted and typed DVI, my bad

 

[thepersonwithaface45]

They are Display Port. I dont know why I brain farted and typed DVI, my bad

Oh okay lol, that makes more sense. My 3080 was acting strange when I first got it as well, but I bought the newest versions of display port cables and it seems to have resolved them. I also think adapters will always have a higher chance of screwing something up rather than a direct connection, but that's imho

 

[Wilshire90]

EVGA 3080 FTW3 Ultra is factory overclocked card (1800 MHz Boost Clock).
Reduce overclock to Nvidia reference clocks
Check GPU temperatures. Improve cooling.

Nvidia reference clocks:

Boost Clock (GHz) 1.71​

Base Clock (GHz) 1.44​

Thanks for the suggestion, I underclocked by -90 MHz Core Clock in Afterburner (1710 MHz boost clock) and still had the same issues. I am reading this may be a somewhat common thing to happen with the 3080/3090 OC EVGA cards...

EDIT: monitored temps in all my games previously and have never seen the GPU get near 80c

 

[dorsai]

I had tried lowering the clocks by -150 Core and -200 Memory in Afterburner previously when I was trying to fix Warzone. I was still getting the artifacts/crashing. Should I try different values? Was worried of going too low because I read that could cause more instability.

I would contact EVGA immediately for an RMA...the card appears defective. The only other outside shot is to try a different PSU, other than that I would go for an RMA as soon as possible because it could take quite some time to receive a replacement.

 

[sizzling]

Thanks for the suggestion, I underclocked by -90 MHz Core Clock in Afterburner (1710 MHz boost clock) and still had the same issues. I am reading this may be a somewhat common thing to happen with the 3080/3090 OC EVGA cards...

EDIT: monitored temps in all my games previously and have never seen the GPU get near 80c

It’s more likely a VRAM temperature than gpu temperature if it’s heat related. I only know of HWInfo64 that can report VRAM temperature on the 3080 & 3090. What is the VRAM temp while in game?

 

[Phaaze88]

I had tried lowering the clocks by -150 Core and -200 Memory in Afterburner previously when I was trying to fix Warzone. I was still getting the artifacts/crashing. Should I try different values?

I believe you can drop the memory clock more than that without issue, since there is no manual control over the voltage being fed to the Vram;
You can crash the memory from applying too high of a memory clock increase because the supplied voltage isn't enough.

I say, try a -450 on the memory clock, and if it still artifacts, RMA that card.

I only know of HWInfo64 that can report VRAM temperature on the 3080 & 3090.

Precision X reports Vram thermals with the FTW3 models.

 

[Wilshire90]

I believe you can drop the memory clock more than that without issue, since there is no manual control over the voltage being fed to the Vram;
You can crash the memory from applying too high of a memory clock increase because the supplied voltage isn't enough.

I say, try a -450 on the memory clock, and if it still artifacts, RMA that card.


Precision X reports Vram thermals with the FTW3 models.

Just tried with -450 memory clock, no dice. Same issues. Looks like ill be submitting an RMA 🙁 . Anyone have any idea how fast EVGA can turn around 3080 replacement these days?