Foresight Institute Logo
Image of nano

Ultimate computer noodling

from the we'll-settle-for-nanocomputers dept.
brian writes "Nice article in New Scientist that serves to establish both a physical upper limit to computing and a timeline. Kinda 'de rigueur' for Foresight but I find this kind of article useful for explaining singularity-like concepts to people. I.e. getting people to realize that one day (rsn) a 1 Ghz Pentium will seem as useless as my old 1 Mhz Apple ][+ seems today:

So here is where Moore's law must end, with a billion-degree laptop or an exploding submicroscopic black hole. "The truth is we have no notion of how to attain these ultimate limits," admits Lloyd. But don't despair–put your faith in human ingenuity. If the rate of progress doesn't slow, we'll reach these ultimate physical limits in just two hundred years' time.

Originally from Slashdot; see ensuing discussion if you like."

3 Responses to “Ultimate computer noodling”

  1. RobertBradbury Says:

    The difference between real and theoretical limits

    I found the article pretty poor because the emphasis was on theoretical rather than practical limits. It is quite possible that the theoretical limits may be unreachable. Most methods for computing at the subatomic level require the use of things like neutronium or muon atoms. These are either impossible to create or have very short lifetimes. Current methods would require huge amounts of matter and energy (e.g. particle accelerators) to construct small amounts short lived exotic computronium. The aggregate amount of computing that that computronium could do, might be much less than you would have gotten using that same amount of matter and energy for highly parallel computing using standard atomic matter for nanocomputers and/or photonic computing.

    I've been thinking about this problem for a couple of years, and it looks like using nanocomputers we can get about 1042 instructions per second using the entire ouput of the sun. We could hit that limit within a few years of getting self-replicating nanotech able to operate in space. It is much sooner than 200 years from now, I would put my guess around 2030, though 2020 is a distinct possibility.

    For more information see my Matrioshka Brain papers at /MatrioshkaBrains/

  2. redbird Says:

    …based on current knowledge

    What you are saying, though, is based on our current understanding of the universe. From what I can tell, we are somewhere between 18 months and 3 years from string theory finalizing (of course, I'm *hoping* that the breakthroughs are bound to come during this time frame based historical presidents). When this happens, I think that we will find ourselves with very different theoretical limits (given a bit of time to apply the new knowledge to computers). Even if string theory proves to be completely false, there is bound to be some other discovery in the works that will change those theoretical limits. Imagine if we had asked a 1950s scientist to predict the theoretical limits of computers? The results of this article will look just as silly as that 1950s' scientist's do today.

  3. kurt Says:

    Ultimate Limits

    This is rather off the wall. But who knows. When we get down to the molecular level in about 20 years (sooner if Reed and Tour have their way), i think things will stop for a while. Then when we get quantum computers, we'll have this big jump in computing power. Then another wait.

Leave a Reply