If you were able to run the other stations in 'headless' mode, it should be much faster, but it would require a 'free-standing' node that would be listening, and invoked by the Master station. Hoping your progress continues, as I am a little bit stuck at the moment.

Definitely up-vote this choice. It would allow a more efficient usage of resources as the fastest machine never sits idle. I even downloaded the Source code for Amleto, which I used to use with Lightwave. It is now Apache licensed and available for use. A little bit different methodology, a 'node' (or two) on each station and a Master plus a local node on the main station, which invoke the headless render on Lightwave, so it doesn't work 'inside' the application, but invokes it as headless render, which I know Blender can do. It is written in C#, inscrutable to me, and I wouldn't know if any of it is transferrable, but it would do "Tiles."

I use TightVNC on all three(3) of my stations, with Links on each station that allow me to view the other two. Once this is working well, I am well convinced that 'they' are all talking to each other. I sit at one station, and setup and monitor the other two using VNC. Make sure that 'Sharing' is turned on for the drives on each station as needed. I also use it to check the Task Manager on each station to see CPU and Memory usage. I find it a very useful way to 'know' that 'they' are all talking to each other. Good Luck.