GPU Driver Error 3080/3080ti fe

[keaton_hiveon]

After you’re on the latest image, if you get a driver error again click on the error and lower the mem clock on the problem card(s) and reboot. Repeat until stable

[anthonydinh]

Hi Keaton
I made it and run 14h. Is this right?

[keaton_hiveon]

Yep all of that looks good

[anthonydinh]

Hello Keaton.
After working with new driver 2days, today the rig stopped and rebooted. I cannot see any error from cards in order to adjust mem
You see

If you see, please tell me.
Thanks!

[keaton_hiveon]

You have 3 cards in the error, lower the mem oc on the cards and reboot(match the pcie bus locations from the error to the cards)

[anthonydinh]

Which one?

[keaton_hiveon]

0d, 01, 04

[Maxkaizo]

Hello, I’m having the same issue. It’s a new 3080 lhr. I’m already thinking on return it. But if you all have a solution I’ll gladly test it

[keaton_hiveon]

What driver/kernel? What clocks?

[Maxkaizo]

[Maxkaizo]

Thanks in advance, just posted some screns

[keaton_hiveon]

lower the mem clock on the problem card from your error each time it crashes and reboot. repeat until stable

[Maxkaizo]

I’ve reached 1900 on memory and still hanging, I’m returning it

[Zak]

Here is my stable settings for 3080 and 3080Ti

and OC Settings

And I’m using nbminer v.42.2

Hope this helps you!

[keaton_hiveon]

Use a locked core clock on all modern cards instead.

[Zak]

How to do that?

[keaton_hiveon]

Values above 500 are treated as locked core clocks in hive. The goal is to find the lowest value that maintains full hashrate