@thatmattrogers
1.There is an NDI watcher, again this is CPU decoded so there is a reasonable overhead.
2. Generally no, the overhead comes from the capture process, what capture card are you using? If you were using some cheap capture dongle that sends a compressed H264 stream you ***may get less overhead on one machine by sending NDI from a second, but at that point it would be a far from optimal setup and getting a proper capture device would be advisable. There are blackmagic cards that have 4 adn 8 inputs, AJA cards that have similar, Bluefish cards that also have multiple inputs.
To do 7 HD inputs I would be looking at a threadripper machine and a beefy GPU like the 3080.