Change mem speed on multiple 3060ti using Linux? [solved]

[Astromine]

I am mining ETH somewhat successfully, one rig with 3x 3060ti cards, running linux mint: NVIDIA-SMI 460.32.03    Driver Version: 460.32.03    CUDA Version: 11.2

Using coolbits and whatnot I can change the graphics clock speed for the cards, adjust power level and fan speed, but I can't change the memory clock speed.

I get errors like this;

Code:

$ sudo nvidia-smi -ac 6800,1440 -i 0
Setting applications clocks is not supported for GPU 00000000:01:00.0.
Treating as warning and moving on.
All done.

This seems to be accepted, but it doesn't budge the memory speed.

Code:

$ sudo DISPLAY=:0 XAUTHORITY=/var/run/lightdm/root/:0 nvidia-settings -a [gpu:2]/GPUMemoryTransferRateOffset[3]=500

Attribute 'GPUMemoryTransferRateOffset' (xxxxx:0[gpu:2]) assigned value 500.

Any ideas?
Thanks.

[Astromine]

Shortly after I posted this I discovered the solution. A bit strange since I have been looking for it for days.

This was the problem;
$ sudo DISPLAY=:0 XAUTHORITY=/var/run/lightdm/root/:0 nvidia-settings -a [gpu:2]/GPUMemoryTransferRateOffset[3]=500

This command adds 500 mhz to the memory speed - in Performance level 3. It seems my cards are operating in Performance level 4 and this did the trick;

$ sudo DISPLAY=:0 XAUTHORITY=/var/run/lightdm/root/:0 nvidia-settings -a [gpu:2]/GPUMemoryTransferRateOffset[4]=500

I wasn't even aware P.L 4 existed, but apparently it does.

This is the situation right now, up from 52MH/s - not too bad;

20210121 23:40:40 GPU #0: RTX 3060 Ti - 60.10 MH/s, [T:63C, P:129W, F:42%, E:466kH/W], 1283/1283 R:0%
20210121 23:40:40 GPU #1: RTX 3060 Ti - 60.10 MH/s, [T:58C, P:129W, F:30%, E:466kH/W], 1281/1281 R:0%
20210121 23:40:40 GPU #2: RTX 3060 Ti - 62.06 MH/s, [T:58C, P:129W, F:30%, E:481kH/W], 1283/1285 R:0.16%
 

Ps, the 2 rejected shares on GPU #2 came in at higher clock speed (64 MH/s) so I eased up a bit.

[ahnick]

What mining software are you using? (along with version number) 

I have a rig that is primarily GeForce GTX 1070s, but I just replaced one of the the cards with a 3060ti and I can't get it to produce valid shares.  It looks like it is mining (producing a hash rate), but every time it finds a solution I receive the message "gave incorrect result".  This happens even when using stock settings.  I'm also using Driver Version: 460.32.03 and CUDA Version: 11.2.