Wouldn’t holding the fans shorten their lifespan? I figured you can use a hairdryer, but you can also do it in 3 steps using 3 OC templates for 1 GPU, or 3+k templates for k GPUs.
Load template 1, and wait until the temp reaches 80+, then switch to template 2, wait for 5s, then load template 3. Done! I have also made a template 4 for the second GPU, and the same procedure goes in order of templates 4, 2, 3. For a third GPU, it would be 5, 2, 3. You may have to tweak the numbers a little.
I have 2 3060s on Win10 getting 49HR with Dummy HDMI plugs but both are connected to the PCIX Slots on the MOBO which support 16 and 8X for those two slots.
Works fine for me 
Plus i have 2 other cards on risers a 1660 Super and Regular getting the HR that are supposed to… 31.6 and 25ish
I for one am not attempting this Fan Glitch with 2 flight sheets… im waiting for LOLminer to come out with a permantent fix. Right now im using LOLminer and getting 37-38 HR on 2 3060s using risers… settings are 1430 GPU Core and 2300 Mem clock PL 125 bit since its Absolute Core the power for these cards are about 100 watts.
I tried 2350 on the memory and they reach almsot 40HR but it is unstable and crashes… so right now 2300 seems to be ok… if it contimues to crash i will drop back some… I see MINIZ miner has something for the 3060s also… but for now LOLminer is working somewhat fine for me…
I was able to get the glitch to work (0% fan) but having a hard time with the 460.39 driver on my rig. The card just can’t seem to get detected, or drops off from pci bus. But with latest driver (465.31), the card is recognized and works fine (but low MHs in ETH).
Anyone have any suggestions on other driver versions for Hive OS? Below is a snip of current config.
How would you set up a fourth template?
As far as I understand template one sets your fan speed of GPU0 to 30 and GPU1 to 70, after GPU0 gets hot you switch to template 2 (for what reason?) and with template 3 you want to get the 0 fan glitch right? how would the other templates look like?
You need to make the memory hot enough, but if you suddenly jump from 1 to 3, most often the fan will not be set correctly, so you will get an unstable temperature. Template 2 is giving enough time for the fan to reach the desired speed.
Template 4 is the same as template 1, but this time the other GPU is set at 30%.
Hi…
I guess you are running RTX3060 original version. The same that works on windows with 470.05 developer driver.
Im trying to bypass the new RTX3060 (GA-106-302).
Because this is a new chip the 470.05 driver does not find the card and wont install in Windows.
So i tried the Glitch method and also the LolMiner. But it also asks to downgrade to a previous driver 460.39 wich is from Jan/2021. As its an old driver, HiveOS won’t boot as well.
Does any of your RTX3060 has the GA106-302 chip? I guess its does not come named with LHR but it works equal.
yes my understanding is that the glitch will only work with the older NON LHR cards as the new ones will not be detected with the old driver that is required in order to perform the fan glitch method.
I do it even without holding the fan - just make errors and then switch the flightsheet - card inno3d 3060 x2 oc. But with much higher settings F 80 C -525 M 3475 PL 120 and hdmi dummy.
Doesnt work. My 3060 cant mine with 2800 MHz clock on Ram. When i switch to T-Rex the miner shut down after creating dag files and says that i have to lower the oc settings. I have 8 GPUs in a Rig and they cant mine 2800 MHZ clock on eth. with the glitch you cant change clocksettings after the glitch.
And for all the Windows Trolls around here: this is a Hive related forum we dont need solutions for things that are no problem. Windows is already solved and easy. 100% of the Internet knows the HDMI Plug and 470.05 Solution for hobby miners
Hi, seen a few comments on here since i posted,
I have been refining my settings across multiple 3060 cards and as per the comment from MarkusB you may need to play with the overclock settings to get a good setup. Ive now tried this about 10 times (moved the PC, tried other coins etc and its repeatable for sure across multiple cards).
One of my cards is unstable with the original overclock so dropped it to 2700 memory, the others are running at 2800, i get about 1 rejected submission per day on most of the cards but was seeing a lot more on that one card so dropped the overclock.
As people have said yes once the glitch happens you are locked into those settings. What i would say is make sure you have done the power limit too, trying to squeeze out more just causes more failiures (at least for me in testing).
Each setup will be unique based on ambient temps etc as you need it to glitch before the temp goes too high and it becomes limited due to temps etc.
One point i would make is one time i tried to rush it and didnt wait a few mins before doing the fan glitch and it just didnt work. Make sure that it has been mining for at least 2-3 mins MIMIMUM before attempting each card.
Again just trying to help others achieve what i have so dont shoot the messenger I will keep trying to help people as i can but not a pro just trying to share my findings.
Im trying better cooling to see if i can stay stable at 2800 as you see in the screenshot but think its unlikely ! but for those that want to see the full setup …
The PNY card (the last one) is the one im having issues with for failed submissions as you see there (ignore the 1080 as im still tuning that one). As you can see you can get 50 across the entire set of cards its been running now for about a week without issue so it can be done across multiple card versions etc. I hope you all have the same success. Also ignore the fan speed settings as they all go to 100% once the glitch is deployed.
EDIT : and for clarity, no i do not use any HDMI plugs (not needed on hive os) and windows restricts the beta driver to only work on 1 card and any other windows solutions are either non full hash rate or only work with one card. This thread is intended to help those on HiveOS with a multi card setup (although it will work with just one 3060 but windows option may be easier for those with that config).
After this procedure I got 50 Mh/s but every few minutes I got message : "TREX: Can’t find nonce with device [ID=0, GPU #3], cuda exception: CUDA_ERROR_LAUNCH_FAILED, try to reduce overclock to stabilize GPU state.
For this GPU settings are : Core clock 100 Memory clock 1500 Fan 70 PL 115
Any idea ?
that is an error you will get when you have clocked too high or the card has overheated. I have found a lot of 3060’s are very tempermental on overclocking. I would say drop down your memory clock even further to see if you can get a stable clock rate. If you are getting 50 @ 1500 then thats the perfect target, honestly i only had to clock mine that high to get the glitch to work, if you can get it lower then thats better.
Make sure that your PSU is up to the job as well, multi 3060’s take a lot of power, also make sure you are runnning the latest version of t-rex as i found some prior versions were a bit unstable.
Good luck!
is there anything i can do about the 100% fanspeed , the noise is killing me… ^^
Unforutnately not without modfying the card/cables.
The only way i know you could do it would be to put a manual control for the GPU fans but that requires getting adapters etc to fit the fan cables etc. Or to connect the GPU fans to the motherboard in order to allow for Software control but both methods require new cables etc to make it work. I may do it at some point and if i do i will post instructions on here but in the short term unfortunately there is nothing you can do.
I am having the same issue. I am able to hit the fan glitch every time but can’t find the sweet spot setting which stables my rig for hours. The best I got far is 40 min max, then the T-rex gives an error can’t find the nonce with the device.
My OC settings
clock -100
F 80
PW 120
Now for memory, I tried every 50 multiple from 2900 to 1500 still can’t get stable setup the T-Rex run for 5 min some time 10 mins some time 40 mins and error out as above please help me. I am using GPU GA106 [GeForce RTX 3060]. TRex version is 0.21.4
I can contact you on telegram also if required but need your help.
Thanks
Are you sure that you arent hitting any power limits etc, i have just built a second rig and while i have an 850w PSU the voltage delivery is clearly crap (it was a cheap PSU). This means in my new rig it runs as you say anywhere from 15 mins to an hour (for me) then craps out. When i try to get it running again it will give me cuda errors etc, sometimes wont detect the card, other times i get 0 hash rate.
I know its not a simple thing to check or test without investing in another PSU unfortunately but if the card will overclock and will run for a period of time and you are sure its not overheating then the most lilkely cause is PSU issues (IMO). There are im sure 100 other suggestions but having built PC’s for over 20 years 9/10 times the simple solution is the most effective one.
Hope it helps, and hopefully you can source another PSU to at least test with before buying an upgraded one.
Im running a rtx 3060 from EVGA with the Glitch in Hive os stable now for almost a day . 49.2 mh/s at 110watts in the Software.
And how does it work, there are no mistakes?
