Hacker News new | past | comments | ask | show | jobs | submit login

I very briefly skimmed this, and it seems like it's using a pretty standard Von Neumann architecture, right? RAM, ROM, ALU, etc.

I understand that it's easier to build Tetris, or anything, using something we understand well, but I wonder what other approach would be better for building in GoL.

It's actually using a Harvard architecture. From Part 4:

> Harvard architecture, meaning a division between program memory (ROM) and data memory (RAM). Although this does reduce the flexibility of the processor, this helps with size optimization: the length of the program is much larger than the amount of RAM we'll need, so we can split the program off into ROM and then focus on compressing the ROM, which is much easier when it is read-only.

Well, cellular automata are grid-based, so maybe something that makes use of that spatial aspect? The only thing I can think of right now is Chuck Moore's greenarrays, which is a grid of dedicated Forth chips wired together:

> Chuck Moore discusses what it takes to program a 144-core asynchronous chip that consumes only 7 pJ/inst, the idle cores taking just 100 nW while the active ones need 4mW running at 666 Mips: tight coding to minimize the number of instructions executed, reducing instruction fetches, transistor switching, and duty cycle.


I don't think that in itself is a good fit, but maybe a good starting point to think about it alternative approaches?

As an aside, what you're saying about it being easier to build anything we understand well applies even more when working in groups, since it's a lot harder to find a solution to a problem without a shared baseline. So it makes a lot of sense that this approach was taken here.

Applications are open for YC Winter 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact