Just double checked with new drivers, new ones slow em down 3MH per card. Downgraded to version 199 and I got it back instantly. Until they fix this I highly recommend using the older drivers on any 3060ti rigs.
Thank you for all the info but I dropped it down and changed some settings and still getting these invalid shares. Very weird not sure why
Up the power to 130, that is too low. That alone can cause invalids from what I’ve read. Also set your fans to 85% those cards are on FIRE. Heck maybe even 90%. You can replace the fans easily if you wanted to run harder than that even and they send them for free most of the time warranty or not. 56 core is pretty high and those fans are so low none of it is making it to the back of the memory core juncture. You want to try and get them to 100 deg C memory temps, you are probably at or near 110 on a few of them.
1 Like
I have autofan on and have it set to 55 degrees. What temperature should they be around ?
That is your core temp, ethereum mines using memory core. The 3000 series have no memory temp in hive. Immediately turn off autofan, forget the temp thing ever existed and turn the fans to 85% or 90% ASAP. Only windows has an new version of HWINFO that will show memory junction temperatures which say for example my pc has a 3090 and the core temp is 48 but 102 memory clock(this is laying open next to AC vent mind you) in hwinfo. You think you aren’t hot but really my man those 60ti’s are on FIRE! They throttle at 110 (gaming doesn’t break 100 so Nvidia could care less if we burn out our cards.) and permanently damaged at 120. Also a big fan blowing constant fresh cool air between the cards so they don’t choke themselves out is key as well.
1 Like
The core temp is deceiving I aim to keep them all under 50, any more than that and I know the memory junction is over 102 C for sure. All I know for sure is you need to up those fans to 85% minimum because they don’t blow into the back where the memory is super hot very well so you need them on at least that to cool the memory cores. If someone knows better by all means, but my rig will choke itself to death and crash if I don’t keep proper cooling, fan speeds, and occasionally when hot AC blowing on it. 1.7GH gets hot though 
You are saying the memory temps are to high based off the coretemp? Familair with the cards themselves? That is interesting as I would not consider 50* core alarming at all. Not mining on 3060ti’s though.
Not at all arguing, genuinely curious.
Ethereum does not mine with the gpu core, so it is idle and cold. 3000 series run extremely hot. I would never let my 3060tis run at 56 they would choke out the whole rig and crash eventually. If you measure the difference in core and memory junction in windows you know what they are in hive at relative hash speeds. Mostly I am pointing out heat from fan speeds. 40% fan speed is 110 degree C guaranteed.
This is with the update settings please take a look. Also if Hive cannot monitor Core temperature how is that even useful then ? Can I see a screenshot of the settings you are running ?
It shows memory Temp also what is the target temp for memory ?
It does not see memory temperature for 3000 series cards so that is useless. Also the cards memory temp runs around 102 degrees minimum if you are doing it right, if your fans aren’t 85-90%(some do 100%) you are talking maximum temps of 110 degree Celcius. Don’t use that feature it is less than useless. Just set the overclock to 85-90% fan speed that’s what the 3000 series need, basically any card you are mining on needs heavy fans 85 is conservative even. This topic is heavily covered elsewhere. Good luck, you need to do a straight deep dive on the thermalling issues inherent in 3000 series cards. Many even replace thermal padding to improve chances of longevity and fix throttling. These are VERY HOT CARDS. I cannot stress it enough, you need to overclock your cards to 85% minimum and just leave them there. If you ran it in windows and ran HWINFO you would see the temps which are easy to top 100-102 just running optimal and it will say 47 degree core temp. The core doesn’t reflect anything other than if it is at all hot that means the memory temp is on FIRE.
ok thank you for the information, Can you please send me where this topic is covered heavily I would like to read up on it in depth and understand this
Thank you soo much for all the help!!! Much appreciate your time
Just look through reddit, search for 3000 series throttling, thermal issues, 3080 in particular has a lot of people modding it to make it run at all when overclocked. These cards took me 2 weeks to get stable and safe running for days at a time. I have 130 watt commercial 20inch fans blowing on each rig with the 3090 rig with two fans on it. It takes A LOT of airflow to ensure longevity with these cards. Do not trust the core temp, I want these cards to last 3 years and to do that I keep massive airflow, AC if hot, and powerful fans to cool them and give them chilled air so they don’t choke each other out with hot air from the back plate into the next cards fan. Red Panda talks about this happening with the 3000 series too. They are glorious once in a decade performance jump great but they run soo hot it takes effort to make sure you don’t degrade them quickly. Reports say 100-102 is fine but beyond that I just wouldn’t want to risk it. 120 is permanent damage so the throttle at 110 is already pretty close for comfort. If you keep getting invalids try boosting power level to 130, word is the extra few watts help it crunch when it otherwise would fail. Seemed to help me, good luck. If you have any questions feel free to ask.
This was most helpful. I have been using phoenix from the start and have always had lots of errors even without overclocks. I have been working through OC settings for almost 6 months trying to find the lowest number of errors. Never thought it could be related to the miner. Switched to T-Rex – no more errors! Thanks!
1 Like
Pro tip, downgrade to the .199 version of t-rex and it will give you an extra 3 MH if you are running 3060ti’s. If it is still available to downgrade to.
I did that all and still getting only 99.69% and with some stales shares still. I guess I will accept it as I can get 100%
i was having exactly same issue with 3060Ti… i found out my problem was my memory clock… ANYTHING over 2k and i was getting invalid shares… ive been at -500 , 2000 for 2 weeks now running perfectly… and yes i lose about .9 of a Mh/s but if it means 0 invalid shares then im happy.
Do you have delay on the OC? You could try fix the core to 1410 and PL should be 130. Temps should be below 50 degrees.
Invalid shares can also come from Wifi issues. I think if you plug all your rigs with ethernet cables using a switch like I did can be the solution.
Actually I verified with HiveOS doc and it’s more about the stale shares.
“* Large amounts of stale shares can affect your effective hashrate. Check your internet connection. Wi-Fi will cause more stales.”
1 Like
Lot of see invalid shares after update of HiveOs. I think GUI started visualising.
Look at the statistics of how many that are invalid… Should be 99 or above accepted or it might be an issue…