Having an issue with one of my gigabyte 3060ti’s. A few minutes after booting up the hashrate drops from 60ish down to 20ish. Hiveos shows that a GPU driver error has occured. For reference I’m using NBminer. Any help is appreciated
Lower mem clock on the problem card and reboot.
Repeat until stable
Are you on the latest kernel/hive/nbminer?
I ended up dropping the mem clock to 2200 and downgrading hive to 0.6-215@220524 while using nbminer 41.3. So things are stable. I dont know how to check if im using an update kernel tho
have you flashed the latest stable image since april 23rd? if not start there.
Just upgraded to the latest hiveos 0.6-217@220528 with kernel 5.10.0-hiveos #110 and im still having the same issue. I’m also running the latest version of NBminer. Gpu5 for whatever reason has its hashrate drop ~40 megahash after about 30 mins
Looks like the dag verification for GPU 5 is failing? any idea whats causing this?
reduce memory oc and reboot. repeat until stable. i cant comment on nbminer as i use lolminer since it has a lower dev fee and in tests outperforms the rest currently.
@keaton_hiveon
Mind sharing your tests? For me lolminer was the worst performer of the bunch. I have only gddr6 cards (no x), could that be the difference?
Thanks. That is close to what I saw on the miner side. However, testing the actual rate, might be almost impossible unless you have a bunch of identical rigs.
Will try to give lolminer another chance once I get the pool to report a stable value. Changed a couple of days ago, and my bad luck followed me there. The pool went from being one of the most profitable ones, to the least now lol Also had a rj45 connector go bad on me, and was causing intermittent disconnects.
i only use lolminer for nvidia, and all my cards are on eth+ton. my amd cards are on trm doing eth+ton as its best for amd still imo
Im still having no luck. Updated to the latest version of hive (0.6-217@220528) and newest kernel (5.10.0-hiveos #110). Also lowered core/mem clocks. After about 30 minutes of mining the hashrate significantly drops off.
find the lowest locked core clock that maintains full hashrate, then when you have issues, lower the memory only and reboot. repeat until stable. if its not stable, continue lowering and rebooting.
hynix cards might not be able to handle those speeds.
I need to run the memory at 1150 on one, and 1750 on another hynix 3060 ti cards
Have the core at around 1350-1410
Start at much lower speeds 1200/1100 and work it from there
My samsung cards work much better, 1350/2250. It gives 59.5 Mh/s @ 119W. The hynix can’t do that much, even at 150W
You want stability over speed
I confirm. Better stability for 3060ti Hynix with lolminer, core OC 1300 and memory OC 800-900 (conservative, could be slightly increased)
@keaton_hiveon nice. I just did a 24 h test on both t-rex and lolminer. T-rex reports a higher value, and on the pool shows a closer hashrate to the reported value. However, could have just been the luck of the day, but lolminer had more shares. Let’s see how it goes during the next 24h.
This topic was automatically closed 416 days after the last reply. New replies are no longer allowed.