So, in theory, can plotting be done on multiple machines, and then transfered to the farmer manually? Say I plot on my workstation at work nighttime, and then move the files to an external HDD and take them home to the NAS?
From operating system standpoint, complete plots are just files not intended to be modified. You can use them, as long as you have fast enough read access and the farmer program is initialized with correct private (secret) keys to access their contents.
Plotting is a process of creating those files, that uses public keys. Can be done anywhere.
With MadMax plotter, plotting can run on any machine by using appropriate
--farmerkey command-line arguments.
I have this setup:
A farmer machine with lots of disks connected to it, running full node.
A regular workstation/desktop pc with lots of RAM and processing power that runs plotting software to fill up one HDD at a time. It takes several days to fill up a 18TB drive, so I spin up a “harvester” micro-service on it that receives commands from farmer node to search for proofs. This is so that you can make use of those plots that are already created, but not yet moved to the farmer machine.
Once the drive is full, I disconnect it from the plotter and walk with it physically to the farmer. Transfer speed reaches a staggering 18TB per 30 seconds, or ~600GB/s!!! Imagine if you carry 2 drives in your hands…