The hardware side of a Lisp Processing Unit

31 May 2010

A couple months ago, I was thinking about trying to build a processor based on Lisp.

I got some time and motivation again to work on this (who knows where it comes from some times). I think perhaps it was finding the excellently-crazy Relay Computer that Harry Porter built. Bill Buzbee made an amazing and more functional computer out of mere discrete logic TTL chips, also very impressive. (And I expect that’s the first time it’s ever been described as a more functional computer.)

A relay is a very big clunky version of a transistor. When power is supplied to a control line, it activates an electromagnet, which causes a switch on a spring to be shoved from touching one output line to touching the other. In this way, you can cause incoming electricity go one way or the other based on the switch.

But that’s it. That’s all they do.

Building a computer out of these large things captures the magic of creating a computation machine out of thin air. Everyone can understand the description of a relay, and has probably played with magnets and electromagnets. When you string a whole lot of these switches together and attach them in a specific way, you suddenly get a computer? How? <insert-Zen-moment-here>

Somehow in the TTL version (which of course is really very similar, except faster) it feels like the computation might be “hiding” in those chips somewhere. Of course they’re really just switches too, and they switch much more efficiently.

Maybe it’s just some silly human thing to enjoy the clicking noises that make it feel like “It’s really doing something in there.”


I currently know roughly nil about electricity, but was inspired enough to think it might help things if I tried to burn my fingertips with a soldering iron: I ordered Make: Electronics.

Experienced electronics dudes will probably sneer at the fact that I also ordered their kit that includes all the bits needed for the first 12 experiments. I’m sure it’s overpriced for the raw parts I’m getting, but hopefully will flatten the learning and frustration curve, not to mention me wasting money buying all kinds of not-quite-right bits.

How long can it possibly take before I’m prepped to make a relay computer, right? :)