This is great. I want to see them implement some of this in hardware, using at least FPGAs to build some core architecture to see how much area savings they get.