Although it was a little bit of a long way to get to a point, I did somewhat appreciate the article's going into the idea of the circuit structure and look up t

Although it was a little bit of a long way to get to a point, I did somewhat appreciate the article's going into the idea of the circuit structure and look up tables. It made me think (and I don't know if this is an apt analogy), it's like if you have 2 versions of a machine responsible for sorting different size balls in a factory. One of them (#1) is built with pipes that exactly encode (in hardware, like a grid of holes) which balls are allowed to pass which size buckets on a production line. The other (#2) uses a robot that looks at each ball passing by, does some measurements, and decides whether it's the right size for a certain bucket. (#1) is not very flexible, but is really great at the job and can do it just by shaking the tray and works for thousands of balls at once. (#2) needs to take time to study each one passing by. But you can reprogram it more easily. Each has its purpose. And the FPGA tries to give you the advantages of #1 with a bit more flexibility to change the size of the holes, etc. Then, replace balls with input video files that need an efficient decoder in hardware, etc. and I see the link.