Are Optical Computers as PCs on the horizon?

[kyrathaba]

Today's silicon-based computers are limited in their speed by the speed with which they can transfer data and the speed with which that data can be processed via logic gates.  That being the case, it stands to reason that a primary factor limiting the speed of current computers is the speed with which electrons move around inside the silicon chips, which is roughly half the speed of light in a vacuum.¹  David Levy says that computing technology of the future will be impossible without a change in the way computers are made (materials, size, etc.)²  New conducting materials developed at NASA's Marshall Space Flight Center can act as computer logic gates that perform 1,000 times faster than today's fastest silicon chips, and are smaller.  And researchers at the University of Rochester believe that their work could lead to a computer more than a billion times faster than the fastest supercomputer was at the beginning of the 21st century.  Now that, friends and neighbors, is fast

Leonard Adelman, who coauthored the RSA encryption algorithm, introduced the idea of using DNA to solve complex mathematical problems.  And in 1997, researchers at the University of Rochester actually developed DNA logic gates.  I won't even get into a discussion of quantum computing. 

Here's my question: I want to know how powerful (memory capacity, processing speed) you guys think that PCs will be in 2050, and in 3000 A.D.  And, secondly, what everyday applications will such powerful computers, using A.I., be put to in those two years, respectively?



1. I.e., electron speed in our CPUs is approximately 93,000 miles per second.
2. Robots Unlimited: Life In A Virtual Age, copyright 2005

[f0dder]

Using DNA is not without problems - it doesn't have infinite lifetime, but will decay. Also it's a lot slower than our current silicon based computers, but has the advantage of being massively parallel. At least that was my understanding of it when I head about it a couple years ago... so the idea was that you could use a "DNA computing device" for some specialized tasks, rather than having a "DNA computer"... same goes for "quantum computer", really.

The idea of optical computers is interesting, but I have this hunch that there'll need to be some conversion to/from electricity in various parts, and that might end up being the bottlenecks.

[2stepsback]

Here's my question: I want to know how powerful (memory capacity, processing speed) you guys think that PCs will be in 2050, and in 3000 A.D.  And, secondly, what everyday applications will such powerful computers, using A.I., be put to in those two years, respectively?

-kyrathaba (April 25, 2007, 06:25 PM)

Speaking robots will be the next big thing, if you ask me.[Edit:]I didn't read the second reference at first, so it kind of looks silly 

Interested people can look at

1) http://rchi.raskince...php?title=Jef_Raskin
Jef Raskin
see the "MicroOptical" photo

2) http://cs.nyu.edu/~jhan/ftirtouch/
and http://www.ted.com/i...uY4IsCFQu2bgodsE4Hcg
Jeff Han

Jef Raskin and Jeff Han.
What's with these Jef's?

[kyrathaba]

There has been a several thousand-fold increase in the power of the PC since the first ones came out with 64 Kb RAM and no hard drive.  In the book by David Levy (referenced in my original post) that I'm reading, the author posits that by 2020 the PC will have the computing power of a human brain, and by 2030 it will have the computing power of an entire village of human minds. 

Do you think such estimates are wildly optimistic or not?  There are some very intelligent people, such as Ray Kurzweil, who believe we're approaching a technological singularity, a point at which our technological know-how reaches a critical mass, so to speak, and all sorts of things that now seem like science fiction will become realities.  For example, David Levy thinks that within the next few decades computing power and the sophistication of A.I. will advance so far that robots will be on a par with human beings in terms of intelligence.  He even thinks (and here I begin to arch an eyebrow) that there will come a point where robots will intermarry with humans, reproduce, have legal rights, etc.  My kneejerk reaction to such predictions is "yeah, right", but who knows?

[gjehle]

i'm lazy, i stick to moore ;-)

i don't see optical or biological computers in the _consumer_ market any time soon.
even tho you hear that researchers have "slowed down" or even "stopped" (aka stored) light the technology is no where near anything you could use.
it's still huge lab benches of high precision equipment in a controlled environment.

having an optical logic gate is only a very very very small step.
if you can't efficiently store the information for a prolonged time (think: registers, cache, ram, peripheral storage) you're nowhere near anything you could call 'computer'.

and yes, the "signal has to be converted to electricity at some point" bottle neck will be there.
there's no way you'll snap your fingers and have optical-everything just like that ;-)

i think the silicon based integrated circuit technology will go on for a bit longer, maybe not silicon based (some other mineral) but same concept.
as you might have already noticed for some time now, development took the direction of (massive) parallelism rather than higher clock speeds.

one speed limiting factor with the way logic gates work is capacity. every transistor has a capacity, every bus has a capacity, _everything_ has a capacity. what you need to do in order to "do" something is basically "fill it up", that's what you call "drive".
by decreasing the capacity (making the transistor smaller) you can get [ higher speed, less power consumption, more elements per inch, ... ].
actually, the little nano-meter value (structural width) that defines the 'technology' (as in: 65nm process, etc) is really the width of the poly-silicon stripe that is the 'gate' of the transistor.
making this smaller, you automatically decrease the capacity of said transistor.

well, silicon based computers aren't dead yet ;-)

[f0dder]

Iirc intel is down to 40nm now - that's pretty frigging small.

I wonder what the smallest scale we can reach are... I recall something about the laws of quantum mechanics setting some limits

[kyrathaba]

Well, there is a physical limit on how small they can get.  The smaller, the thinner.  Get the electron pathway too thin and it won't stand up to the heat generated by the electron flow.  It would melt.  So there must be a lower bound on size due to this factor, if nothing else.

[Deozaan]

the speed of current computers is ... roughly half the speed of light in a vacuum.
...
New conducting materials developed at NASA's Marshall Space Flight Center can act as computer logic gates that perform 1,000 times faster than today's fastest silicon chips, and are smaller.

-kyrathaba (April 25, 2007, 06:25 PM)

Are you saying that computers can be 500 times the speed of light in a vacuum with these new materials?

[kyrathaba]

No, although I can see where I may have given that impression.  Although electrons in chips can flow approximately half the speed that electrons in a vacuum can flow, there is also the matter of resistance in the material of which the circuits are composed (no resistance in a vacuum, on the other hand), and the distance they must flow inside a PC which, though small, do take time.  What I meant by the part that you quoted above is that the research at NASA's Marshall SFC has shown optical logic gates that can switch (on/off) 1000 faster than the current ones in silicon based chip technology.

[Edvard]

And, secondly, what everyday applications will such powerful computers, using A.I., be put to in those two years, respectively?

As fast and powerful as computers may get in the future, they will never outstrip the bounds of human nature.
In other words, I strongly suspect folks will still be finding ways to use computers (in whatever incarnation they may be then...) to find pictures of naked people and play some (incredibly fast) tetris.
  

[gjehle]

Iirc intel is down to 40nm now - that's pretty frigging small.

I wonder what the smallest scale we can reach are... I recall something about the laws of quantum mechanics setting some limits

-f0dder (April 26, 2007, 10:17 AM)

the (most) limiting factor right now is the photo process used to transfer the masks onto the waver.
structures are already starting to get smaller than the wavelength of visible light, which is countered by using shorter wavelength light (like UV) and special optical systems.

there have been experiments using electron rays or instead of light to transfer the masks several years ago.
it works, the structures are superbly sharp, but the process takes several hours for one layer.
not feasible.

another problem caused by getting the structures smaller and smaller is the fact that you use acid to get rid of the stuff you don't want.
acid doesn't know what sharp 90 degree turns are, everything starts getting rounded and blurred.
you also have to deal with acid not going straight down, but also sideways, basically eating away everything underneath the pathways you wanted to get by removing stuff around them.