Self-Assembling Nanocomputers

A Semi-Anonymous Coward writes: "According to this article a researcher at Harvard University has developed techniques for self assembly of nanoscale wires that operate without resistance due to a property called ballistic conductivity. He hopes the research will provide an 'end run' around convential top-down circuit designs, allowing much smaller, faster and more energy efficient computers."

Comments

Re:Parent = +5 funny!

[Anonymous Coward]

Most MS trolls (Microsoft employees assigned to participate in public forums, pretending to be industry experts...sic), usually are...this is part of their 'profile', and one of the ways they attempt to endear themselves to neophytes....notice how the writing invites paraphrasing. Any number of rookies will take this trash as gospel, and regurgitate it around the water cooler for the next 6 months.

All part of the big lie out of redmund...it will be while before it fades back into the darkness from whence it came.

Nice to know they are still afraid, however :)

Re:Correct me if I'm wrong:

[madGenius]

Since there's no resistivity, that means that calculations will be almost instantaneous, right? And it will have very low power consumption, no waste heat, and be incredibly small?

I am afraid that most of the power in modern integrated circuits is capacitive not resistive, and though ballistic conductivity would reduce the dynamic heat disapated by signals and eliminate the static heat in the wires the overall difference would not be that great as most of the power is used to switch transistors from one state to another.

So this sort of thing could easily mean that we could have tiny computers that run for a long time on a single battery and are ninety billion times better than anything we currently have, right?
I am afraid not, though it may take a while longer to fry an egg on a processor implementing this technology :-)

Having said that any reduction in static (read useless) heat generation would increase the processor speed as you would be able to increase voltage and hence MOSFET switching speed with the same overall heat generation of a processor no using this technology.

PS: Moderators what are you on, the parent post may be inaccurate but it is NOT a troll.

Re:Correct me if I'm wrong:

[bradbury]

As others point out, even ballistic conduction doesn't get you past the speed-of-light limits. You are also going to have to cool the computers down to close to absolute zero if you want ballistic conduction over very long length scales (otherwise atomic vibrations will disrupt the flow of the electrons).

The limits on power consumption have relatively little to do with electrical resistance and a great deal to do with erasing bits. As Landauer and Bennett have shown, you can compute for essentially free but you have to pay a price of generating entropy (heat) when you erase bits. To achieve the really significant increases we have to move from non-reversible architectures (all current commercial computers) to reversible architectures that minimize the number of bits erased. Michael Frank is one of the leading people working in this area.

As Drexler discusses in Nanosystems, using reversible rod-logic nanocomputers, one should be able to get from our current 10^9 ops/sec chips to 10^21 ops/sec in 1 cm^3 before one hits the heat removal limits. So the anticipated throughput increase is ~10^12 ops which a trillion vs. your estimate(?) of 90 billion. But it isn't going to run on a single battery. Its consuming (and radiating) 100,000W. Interestingly enough, since such a nanocomputer has ~10^3-10^5 times the processing capacity of the human brain in 10^-3 times the volume such a computer is probably worth a million or more human brains (if we can figure out how to program it...).