Perseverence is good.

A few of us have used true "hybrid computers" which interconnected digital and analog computers, but most of these were tossed out years ago,
The computer I'm planning on making is a hybrid, with non continuous time, but continuous values. Continuous time is more elegant in my opinion, and I'd love to see that in action, and even make it myself, but I've got a strategy for an oscillated logic chip that is actually potentially more hi-performance, but they are both hi-performance, but I get a little extra for putting the oscillator in, so I'm doing it that way I think at the moment, nothing is decided, until I've actually got the thing tho for real..
Why do we need oscillation? (or continous time.)
If your logic doesnt feedback the output back to the input, you dont need an oscillator... but if it does then, you can implement continuous time, or you can oscillate it. I had an idea for a quasi-continuous time one, that waits for the output caps to fill before they spark across a gap back to the input, and thats an idea for getting continuous time, but that actually isnt truly continuous, because it waits to spark.
Getting continuous time to work, theres a definite trick for it that you have to work out.
I dont believe that you just "throw these hybrid computers in the garbage" tho, If you have a one cycle whole frame advancement, thats a whole frame per cycle, not just a floating point operation per cycle, which they use to judge super computers these days, I dont want just one floating point operation, I want the whole program to be updated per cycle, and you can do that with continuous time, or hybrid oscillated, and you can get a "whole update" done, of all the instructions of the program in one hit.
If you do that, you get a squared performance! Then your kicking butt.
Heres an example of the whole frame of logic of ping pong put in physically. (In a virtual logic Sim.)
This thing actually gets the whole program done at ~4 HZ! so just imagine if it was a megahert, you would pump out the frames like mad in fast forwards! And as it is, you could probably make the logic out of wood, and it wouldnt go too fast for just hand cranked logic! =)
But putting in the 20 or so operation stages in (each stage being about 25 operations maybe) in physically, might get a bit big, so then continuous values comes in, and reduces the bus size to a single wire, and getting the operations in is more possible without a digital bus!
More down to earth -> even if your oscillating bog standard TTL and gate chips with a 555 timer, boring usual style with average transistors, if you get all the operations in physically, if its oscillating at only a megahert, u r still more hiperformance than a gpu, for the physical task you put in, because its a megahert per frame. not per operation!
FPGAs also grant this huge performance hit, I think.
And what im doing isnt much different than that, but I want a faster oscillation, and a smaller bus size to fit more operations in.
And thats the big deal with what I want to achieve... I wonder if I get there...
But the untold secret is, what is the purpose of my logic? what am I going to implement??
HAROLD - Ill go check out the maths on that - then Ill get back to you. thanx for helping me out, Ive got a feeling if I get past these slow formative stages Ill be flying fast like the flash very soon!!!