Thursday, September 20, 2012

Random (Access) Thoughts

Still working on my computer design, off and on. Refining the instruction set, walking through theoretical problems to see which faculties are present, and which are needed. Tossing around some interesting ideas..like a nixie readout for the program counter, or how to decrease instruction time given a clock of 6 pulses/second. Everything gets written down, revised, some done on computer, others in notebooks. What gets put in? What gets left out?

Here’s an interesting thought. That nixie display for the program counter. Simple enough, right? You have 12 binary bits…convert them to decimal on the fly. Well…20 years ago they made chips that could do the conversion (they were actually pre-programmed ROMs). I’d need roughly 8.

1- this violates my relay/vacuum tube leanings but I won’t entirely rule it out. 2- you can’t get them anymore if you tried. 3- We’re looking at something like 2000+ relays in a brute-force method just to convert binary to decimal for a simple display. 4- today this would be done with a little microprocessor solution: Faster, cheaper, and more powerful than the entire computer itself. 5- In 1965 a paper was presented on how to do this conversion with solid state logic. It was found to be an affordable solution compared to other methods. It cost them $1300.

I really need to get to the Computer History Museum in Mountain View, CA before I get too carried away.