I had about 2.23 GH/s at first like most here. Took a look at the PCB and realized the fury chip was a bit choked to prevent heat and USB power issues. R15 next to the button was set at 22k. Read that a value of 10k would be fine. By paralleling another resistor you get R= 1 /(1/R1 + 1/R2). Set a tiny SMD pot to about 19k (careful not to set to low!) and bridged the the 22k resistor for a total of about 10k. Did a reboot and was seeing about 2.84 GH/s. High error rate and the heat sink was getting really hot. A little increase on the pots resistance landed me at at an Avg of 2.78 GH/s and a report of 2.61 GH/s. Error 4.6%

 
So you have an idea of what you should be listing in your rc.local (using Debian armhf on a SoC)
To make sure you are really listing the right device (a /dev/ttyACMx) on the port see what is there.

A fury will show:


This is even more important for ttyUSB devices since BE and BFL etc all will show up as /dev/ttyUSB* stuff.

Just change the ttyUSB0 to ttyUSB1,2,3 …n. In my case this is the cmd in my;



If ports start changing on you give the -S noauto a read

This is even more important for ttyUSB* devices since BE and BFL etc all will show up as /dev/ttyUSB* stuff.
Just change the ttyUSB0 to ttyUSB1,2,3 …n.

Using screen like this with the -dmS will just kick off the command and detach while the S parameter gives it a nice name “bfgminer”.  

Anyway I have my fury set at 2.7 GH/s now and that seems to be “the sweet spot” on this thing. More or less what I got with a fury chip setup I built running at 0.84V @ 2.5A clocked at 230MHz. This thing should be easy to chain fury chips on...idk. If I can get hold of some bare chips Ill give it a shot...that is if I don't get some ANT (BM1380) chips sent to me first for another project.

Somebody said something about "the pencil mod cmd" lol, sure that might work too. Not sure I personally would do it that way.