3080TI - Ethereum - OC settings

ok thanks

Yes thank you sir :slight_smile:

Thanks lee, by the way do you have the VISION ones?
Did you change the thermal pads on your cards?

Hello Big,

All visions Indeed and i changed nothing yet they are almost New ( bought it a month ago ) When do you change thermal pads ?

I did not change the thermal pads. I am just asking if you change them so thank you again for your answer. What do you mean you have to go step by step with the overclocking? Can you shed some light on this because I’m fairly new on handling lhr cards…

Hello,

Each GPUs can’t be overclocked the same way so you need to increase memory clock and core +50 each times for exemple to see how much ur card can handle. For example i have GPU which can go 3100 memory clock but i got one which can’t get more than 2340 and they are both 3080 ti vision so u need to find the “perfect” overclock for each cards

Look :


What pool are you using?

What pool are you using?

Ethermine =)

2 miners

HiveOS >= 0.6-211@211111 ( 2021-11-11)
T-Rex miner >= v0.24.6 (automatically updated with HiveOS)
nVidia driver = 495.44 (just execute “nvidia-driver-update” on “Worker Commands”)
Card EGVA 3080Ti FTW3 with latest VBIOS release 94.02.71.80.86
OC = core 1250 mem 2650 PL 290 (290 set required but consuming ~270W)
Autofan = Target Core TEMP 70 and Min fan speed 65

Hashrate 90~93MH/s
Consumption 255~270W

After set the OC settings, is very recommended to shutdown and power on the rig to see the "regular results"

no lhr locks? i get locks every 20 seconds…

Thanks for your help i’ll try but u have EVGA so maybe won’t work for gigabytes

Happened 1 time on only 1 gpu after 5 days booted so maybe u can try to set lhr tune
image_2021-11-27_131836

Autofan is pointless to use with Nvidia cards. When mining the memory junction temperature is what gets hot. The chips start to degrade around 100-110c. Nvidia doesn’t have the memory junction temperature reading for the linux drivers, so you can’t see them in HiveOS. Your core could be 50c and your memory could be 105c. It’s better to just use a fixed fan speed for nvidia cards around 80-100. Just my two cents, it will help out in the long run.

Agreed, fans are cheaper to replace than GPUs.

Got another problem, i have 5 3080TI cards, 4 the fans show up correctly in Hive, but 1 card looks like the fan is not working, but when i look at it it works. Is this a bug from Hive?
I have set the autofan from 80-100%

I am building a 3080ti rig. One card works great but as soon as I add 2 or 3 I get them all starting out at 90mh and then dropping to 50 then 20 then driver crash. Is this just an overclock setting that I have to figure out or is this a different issue?

Power, cables, risers, overclocks, etc.,

“driver crash”? check the logs?

I would suggest fixed fan rates for these cards regardless.

Not likely a “bug” in an OS for a single GPU.