Gregory's ServerA Simple Universal Logic Element and Cellular Automata for Reversible Computing
by Kenichi Morita
https://www.cs.auckland.ac.nz/~cristian/UMCreadings/revcomputCA.pdf
 6 comments
 @neauoire in all of your research on reversible computing, have you found any evidence that reversible concepts can be applied to software to reduce energy usage on non-reversible hardware architectures? I’ve tried looking into this and have come up empty, but have not put as much time into it as you have. @jakintosh it's a complicated answer, but to put simply, no. It seems to be held behind blockers similar to why we don't have parallel data structures primitives in most common languages, it's many years before we have cohorts that are using future of coding type systems. There's a couple of shifts that need to occur before this can happen. It's just a fringe theorical idea atm. @neauoire so in other words, do we strictly need reversible hardware to see the energy saving benefits? or is it theoretically possible, but not plausible, with existing hardware? mostly coming from a perspective of planning for efficiency of existing hardware and not just praying for new architectures to do it for us @jakintosh Oh, yeah, no that's no compatible at all with how hardware works right now. It'll be a whole different architecture. @jakintosh and even beyond that, the current programming paradigms don't map well at all on the models. |
@neauoire Every invertible Boolean function is - necessarily - a permutation of its input bits over all argument cases. It follows that any such function can be described as a permutation index - which implies that, for example, any invertible 8-to-8 bit program can be described in 1683.996 bits. (see OEIS A000722)