On Fri, 14 Jan 2000, Eugene Leitl wrote:
> Matthew Gream writes:
> > Perhaps someone more versed in computability metrics, universal (turing),
> > theoretical (graph, numerical, parallel) and realisable (dna, biological,
> > silicon, etc) machines could shed more light into this area.
>
> DNA computations work by matching complementary patterns of hydrogen
> bonded bases. Solvated macromolecules molecules are not very fast, and
> are certainly not rapidly reconfigurable.
>
> 3d integrated molecular circuits (computronium) can run rings around
> DNA computing, even in the best case.
>
I agree 100% with Eugene.
[side bar, note this day in history -- while Eugene and I agree "generally"
on most things, we seldom seem to agree 100% on anything. :-)]
The 3d molecular circuits are going to be limited by the speed of sound
in the material determining propagation delays, but these will be much
much faster than diffusion in solution. Faster yet will be optical
computronium due to the decreased mass of the photons and faster
transit times. What remains unclear is exactly *how* small we can
get the transmitters & receivers of the photons (compared with atoms
or electrons). If we can get to the point of pumping single-atom electron
energy states reliably using single photons (which is different from quantum
computing), then we shall have really really dense and really really fast
computronium indeed.
Robert
This archive was generated by hypermail 2b29 : Thu Jul 27 2000 - 14:02:17 MDT