I got the fan to bug out to 0% fan speed consistently now. I THINK my issue was having an HDMI dummy plugged in. I’d suggest try to remove HDMI dummy plug and try it.
I got the 3060 to mine ETH at 50mh/s ONCE after the 0%fan speed bug. However, it was not stable and would have a damaged DAG and the miner will restart. at the same time, I cannot change the OC for the card.
Ever since then, I can consistently 0%fan, but cannot get the card to mine ETH at 50Mh/s after. it always drops down to the 20s.
I am not sure what i am doing differently after the 0% fan bug. Adamram, if you can enlighten me, that’ll be awesome.
I will also do more testing and update everyone here =D
From what I’m doing, it seems the limiter will always limits any coins once it hashing ethash. If I first boot with 3060 mining ethash, the limiter surely will kick in. In this condition, even I switch Flight Sheet to mine ergo, it will only hashing on half rate. So I always make sure before the fan (and also core+mem) dropped to 0, it should on full hashrate state with autolykos/octopus mining. A reboot needed to refresh the hashrate to full. After that I can safely switch flightsheet to mine ETH.
Restart methods are different between miner apps. Phoenix usually reboots, so it will breaks the glitch. T-Rex got a higher hashrate (49,6 MH/s) but sometimes freeze. I use Gminer because it more stable and when it lost connection to 3060, it restart and reconnect instantly and shares table will back to 0. Don’t worry because it just a counter from miner, just check pool dashboard to check real counter.
In my rig, if it’s already at halved hashrate state, then it will never back to full hashrate even I set overclock or flightsheet. I had to make sure overclock settings are at desired point, then I reboot to start ergo mining. Then wait for 0% to switch flightsheet.
One another critical point:
Maybe the glitch connected to memory clock settings. If it set 2200, it never dropped to 0 while mining ergo. I set it maxxed out to 2600 and it dropped quickly. A bit logic because ethash stressing on memory.
I was trying to test this with 1060 3gb mining etc and 3060 ergo and manage to bug the 1060 mining eth (after changing the flight sheet to eth) at 20mh…
I spent a couple hours testing & have tried many different configurations to enable this glitch without any luck. I haven’t been able to get the fan to drop to 0 no matter what I try. A fresh reboot was done between every test. Here are all of my different tests / variables.
All tests were done with this card:
Card: GeForce RTX 3060 12053 MB · EVGA
Memory: Samsung GDDR6
Bios: 94.06.14.40.45
All tests were done with this flightsheet, which enables nanopool / ERGO mining on the 3060. (I don’t know if anyone can do anything with my wallet addresses, so I blacked them out):
All tests were done repeatedly with the following variables:
Card plugged directly into 16x PCIe slot
Card plugged into 16x to 1x powered riser
Card connected to monitor
Card disconnected from monitor
OC: Fan 85 Core -260 Memory 2600 PL 110
OC: Fan 0 Core -260 Memory 2600 PL 110
Driver / HiveOS Version Tests:
Test 1
HiveOS Version: 0.6-203@210503
NVidia Driver: 460.73.01
Test 2:
HiveOS Version: 0.6-203@210503
NVidia Driver: 455.45.01
Test 3:
HiveOS Version: 0.6-202@210401
NVidia Driver: 455.45.01
Test 4:
HiveOS Version: 0.6-202@210401
NVidia Driver: 460.73.01
Hopefully we can somehow find the magic trick that unlocks these cards’ full potential in HiveOS. I have 2 of them running at about 26mhs right now.
adamram,
You’ve been extremely helpful in providing tons of information so far. Would you mind answering another question?
What type of riser card are you using for your 3060? Could provide a make/model and maybe even a screenshot from your order and/or a pic of the board? It may help in tracking down the solution for exploiting this bug.
The riser which 3060 using is a 16x v2.0 (not 3.0) and just a random ordinary riser bought from an online shop. I’m not making it in order, just plugged in to last remaining PEG slot (yellow one). And you see that randomly put GPU’s… 2x5600, 3060, 1660s, 2x5600… Not so cool and untidy. Hive OS read the order as my screenshot on previous post.
Lorenz, thank you so much for the tip! I saw your post & couldn’t wait to try.
I upgraded HiveOS, switched over to lolminer 1.29 & installed the 460.39 driver. I saw an immediate improvement on 1 of my 2 3060s. For some reason, the second 3060 doesn’t seem to start hashing while using lolminer 1.29. I tried swapping risers & connecting an HDMI dummy plug with no luck.
Anyways, I’ve currently got lolminer 1.29 running on my first 3060 with great results! I’m using gminer on the other 3 cards.
Well, strange. I suppose you need to play around a bit and you will found out. I saw many already mining with lolminer and I haven’t seem many issues.
Btw, my worker has been a bit unstable. I rebooted 4 times during the night. I believe was because my OC. Now I’m trying something less aggressive and so far is ok.
I tried to downgrade my nvidia drivers but no luck, keeps crashing, no GPU recognized - mess!
i did fresh install on HiveOS, same thing, i cant install the driver, or/and update since the installation file is with the old HiveOS version!
Anybody?
I tried few types of installation of the drivers, with full download address and the short command, none of them worked…i am giving up!
Switching back to Trex and waiting for stable release…
Please, if anybody has some thoughts, go ahead
Turn maintanence mod off and reboot.
Start the flightsheet with lolminer. It takes some time for every card to kick in (one card showed n/a hasrate for couple of minutes). But eventually it is working! Lolminer rebooted rig ones, will see what will be going on later.
Hive OS read the order as my screenshot on previous post.
