ConwayLife.com

Hi,
I'm planning to implement a massively parallel simulator of game of life on digital circuits for my class project. I have an FPGA with 6-million logic gates (effectively). I'm guessing the simulation would naively yield 1 million to 1 billion steps/sec and the grid will be somewhere between 100x100 to 1000x1000. I'm not quite happy with the grid size but I'll be thinking about how to trade in size for some overhead in speed.

I think GOL is really awesome but I don't know a lot about it (and I just joined the forum). So what I am trying to understand is:

1) What similar work others have done, and what are the fastest simulation speeds/sizes for cutting edge software (and maybe hardware if others have worked on the same veins).

2) What cool patterns could I run with this? Something to show non-enthusiasts how awesome GOL is. I'm looking at some kind of semi-useful (possibly numerical) computation and demonstrating how fast it runs. (for example I saw that someone has a pattern that produces fibonacci numbers encoded within gliders every 2100 steps, please point me to more of such awesome things).

3) What other cellular automata could I play around with this? It is actually easy to implement different circuit designs on it because it is all automated, so I am planning to experiment quite a bit with my kit.

Historically, the one big piece of famous hardware for this is the Connection Machine. There's a great article about Richard Feynman's involvement here:
http://longnow.org/essays/richard-feynman-connection-machine/

Stephen Wolfram got involved because of his interest in lattice gases, which you can read about in A New Kind of Science on pages 378-382. So lattice gases might be a fun thing to experiment with, there's some code here. (The system that Wolfram shows on page 378 is known as "FHP6".)

If you want something of more current interest, you might look at Quantum Cellular Automata, possibly, but it might be too hard. Otherwise, just pick a CA and see what speed you can get.