backtop


Print 23 comment(s) - last by SiN.. on Aug 28 at 6:54 PM


The new research from IBM builds upon its discovery of light emitting nanotubes FETs, pictured here, which were developed in 2003.  (Source: J A Misewich, IBM)
IBM continues its march towards creating an optical computer with another breakthrough.

IBM is among several companies betting big on optical computing as the next big thing to replace traditional electricity based computing.  Optical computing is seen as a stepping stone for even more advanced computing technologies such as quantum computing.

In order to build a quantum computer, three key light-based components are needed:  transistors to form a CPU, I/O equipment, and memory/storage media.  IBM already has the I/O side well on the way, thanks to its advanced fiber optics research and switching breakthroughs.  It has also seen great gains in memory and storage media.  Finally, it could likely adapt its switching technology from the I/O research to form a transistor analog.

All the pieces are in place, though still in early stages, however; one critical component was missing -- something to make intense light pulses on a nanoscale.  IBM needed a nanolaser and that's exactly what its researchers have created.

The new nanolaser will likely be analogous to the clock in a full nanocomputer, producing the driving impulses.  Further nanolasers could convert electric signals from peripherals to laser impulses to be sent to the CPU.  The system could also be used more modestly in a shrunk version of fiber optic network on traditional electric systems.

The breakthrough device utilizes a familiar friend -- the carbon nanotube.  Carbon nanotubes, formed from interconnected hexagonal carbon rings, are also being studied as a material for traditional electric transistors, thanks to their great conduction, strength, and flexibility.

The new research, reported in the August 25 Nature Nanotechnology journal, uses a special nanotube-based field effect transistor to generate light impulses on a nanoscale.  The light impulses are then routed by a pair of tiny nanocavity mirrors.  By controlling the nanocavity mirrors, the wavelength of the optical emissions, the spectral and spatial distributions of the emitted light and the efficiency of the emissions could all be controlled.

An optical computer could theoretically perform computations at the speed of light (though only the speed of light in mediums such as glass, not the more commonly used speed of light in a vacuum).  This would allow for faster computers.  Also, the light impulses would likely generate less waste heat then electric circuits, which would allow for denser hardware.

A quantum computer could take these gains even farther, computing at faster than the speed of light, thanks to bizarre quantum effects such as entanglement.  However, a quantum computer would require manipulation of single electrons, where an optical computer only would require larger light-controlling components.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Speed of light
By Some1ne on 8/25/2008 5:46:48 PM , Rating: 4
quote:
An optical computer could theoretically perform computations at the speed of light


So what, precisely, is the speed of light when measured in computations per second?

The only "speed of light" metric that I'm aware of is measured in distance over time, and that's completely useless from a computational standpoint.




RE: Speed of light
By MozeeToby on 8/25/2008 5:54:44 PM , Rating: 2
Well, technically they could be refering to the time it takes a signal to travel from one part of the processor to the other. Unfortunatly, if this is the case, they are still wrong. You can't send information faster than the speed of light, even with quantum effects (An no, not even using entanglement, tunneling, and teleportation can you violate that rule).


RE: Speed of light
By MozeeToby on 8/25/2008 6:08:54 PM , Rating: 2
Whoops... I guess I have to retract the still wrong part, I read what they said incorrectly, hell they even qualified it with the speed of light through glass.

Just me, MozeeToby, saying... "My Bad".


RE: Speed of light
By SiN on 8/26/2008 4:56:45 AM , Rating: 3
I thought new research theoreticaly proved that you can in fact go faster than the speed of light. Now if only i could remember where i read this.

Personally i dont beleive we are bound by light. A theory is a theory not living proof.


RE: Speed of light
By The Boston Dangler on 8/26/2008 10:48:05 AM , Rating: 2
do you know what a hypothesis is?


RE: Speed of light
By SiN on 8/28/2008 6:54:15 PM , Rating: 2
if i say we cannot go faster than the speed of light, i guess thats a hypothesis. I didn't know what a hypothesis actually was untill you asked me, i had to check it out.


RE: Speed of light
By Xenoterranos on 8/26/2008 10:56:48 AM , Rating: 2
A belief is a belief, and you're entitled to have as many varied ones as you wish, but a theory's definition is pretty concrete.

Try here:
http://en.wikipedia.org/wiki/Theory


RE: Speed of light
By Fnoob on 8/25/2008 5:58:15 PM , Rating: 2
I assumed, prolly incorrectly, that electromagnetic energy (electricity) travels at the speed of light. So if it doesn't due to resistance, how close is it? Is it slowed down by some fraction of 1%? So this new tech would increase the speed by that amount? I'm underwhelmed and likely missing something here.


RE: Speed of light
By Some1ne on 8/25/2008 6:09:07 PM , Rating: 4
http://en.wikipedia.org/wiki/Velocity_of_propagati...

Basically, it depends on what sort of link the electricity is propagating through. It seems reasonable to assume that a propagation of at least 80% the speed of light in a vacuum is easily attainable.

So if the article is referring to the propagation speed, having signals traveling at the speed of light is only good for a gain of <= 20%. Not bad, but not revolutionary either.


RE: Speed of light
By 9nails on 8/25/2008 11:31:01 PM , Rating: 5
Ok, so the advantages of speed in copper connects vs. light connections is arguable.

The important thing to take away is that when you ramp up the density of those copper connections they tend to bleed data between each other, generate heat, resistance creates power inefficiencies, they are susceptible to external electro-mechanical interference, and can be taken out by odd effect of nano-whiskers causing short circuits. Circuits designed in light, theoretically, have none of these problems. The benifits may seem minimal, but the importance of this breakthrough are significant. This breakthrough allows one to tightly pack in circuits preforming at high frequencies with less power loss and heat. From a simple perspective; cooler circuits, more bandwidth and cleaner signals equate to faster computers.


RE: Speed of light
By JonnyDough on 8/26/2008 2:34:20 AM , Rating: 2
In addition to that, some further reading for those who require a better understanding of photons and electrons. I was doing some of my own research about size differences. Turns out, there is no relative "size"!

http://www.newton.dep.anl.gov/askasci/phy00/phy005...


RE: Speed of light
By AnnihilatorX on 8/27/2008 1:15:06 PM , Rating: 2
Well the propagation of signal base on electricity is equal to speed of light. Electron drift velocity within a conductor is only around millimetres per second. But that's irrelevant to speed which signal transmits.

An analogy of electrons drifting in a conductor is easily visualized as follows:
Imagine a tube containing a line of touching spheres without gaps. Pushing the ball one end will affects the one at the other end near speed of light due to electromagnetic interactions in microscopic level on sphere surfaces; though each sphere only travels at the speed you pushed them in.

Optical transmission is only better than copper not due to speed signal propagates, but solely due to 2 major advantages: The non-existence of leakage effects on a narrow channel and the much lower noise; especially when electronics are pushing to lower voltage levels at smaller manufacturing process to minimise heat dissipation.

You can see that resistance, hence heat dissipation, is only an indirect factor why optical is superior to copper.


IBM keeps surfing atop the wave
By DeepBlue1975 on 8/25/2008 5:04:37 PM , Rating: 3
Another advancement by IBM.
It's obvious that their decision to ditch their PC and laptop division was a smart one, as they are now totally focused on the development of new technologies.

That said, I'd like to see a scaled schematic of this, as my imagination can hardly keep up with laser generators, mirrors and light channels at such a microscopic scale!




RE: IBM keeps surfing atop the wave
By Murloc on 8/25/2008 5:43:04 PM , Rating: 2
that's true, in the future IBM will be the boss again imho.

Uhm because the light is faster maybe you can make fewer transitors, or not?
hehe I'm not strong on these things.


RE: IBM keeps surfing atop the wave
By KernD on 8/25/2008 6:16:20 PM , Rating: 2
You must have missed the news about how Intel is also going in that direction.

Just an example of some optical silicon they work on.

http://www.intel.com/pressroom/kits/research/4Gmod...

Did you seriously think for an instant that IBM was the only company that had future tech research in this field?


RE: IBM keeps surfing atop the wave
By nafhan on 8/26/2008 7:23:17 AM , Rating: 2
quote:
they are now totally focused on the development of new technologies


Don't forget about mainframes, consulting services, foundries, software, etc. It might be more accurate to say they are totally focused on enterprise level technology.

We just bought a Z10 here at work. :)


By DeepBlue1975 on 8/26/2008 10:04:00 AM , Rating: 2
Totally true, my bad!

It's just that it's been a while since I last worked near a mainframe. I'm more used to seeing "Cheap" hp-Intel servers all around :D


By Xenoterranos on 8/26/2008 11:00:54 AM , Rating: 2
I thought one doesn't really buy a Z10 -that one more or less rents it indefinitely, leasing more processing power as needed? Or is that just how the old IBM worked? Completely off-topic, sorry about that!


Carbon Nanotubes...
By MozeeToby on 8/25/2008 5:05:01 PM , Rating: 2
Is there anything they can't do?

Seriously though, this is one of the last few peices of the puzzle when it comes to optical computing. Unfortunatly, one of the few remaining obstacles is still one of the most daunting; how to manufacture the pieces cost effectively.

Some of the latest CPUs and GPUs contain over a billion transistors. Until the optical components are just as cheap for the same level of performance there will be only a tiny research market for them.




RE: Carbon Nanotubes...
By Clauzii on 8/25/2008 5:15:04 PM , Rating: 3
IBM Creates Tunable Nanolaser
By Ordr on 8/25/2008 6:12:27 PM , Rating: 2
...to attach to the nanoheads of the nanosharks, of course.




RE: IBM Creates Tunable Nanolaser
By Aloonatic on 8/26/2008 3:08:28 AM , Rating: 3
We'll be OK as long as they don't get into the hands of those long suffering sea monkeys.

Then they will be able to their revenge on all of us who have loved them for a short period of time before getting board, neglecting them and then eventually flushing them down the sink/toilet.

Revenge will be theirs :-s

Run for your lives :-o


Confusing article
By oTAL on 8/27/2008 5:58:17 AM , Rating: 2
Hey Jason,
By introducing quantum computation into the story you made it unnecessarily confusing.
The concept of light based computation has very little to do with the concept of quantum computation. It makes no sense to mingle both when the story is about an advance in light based computing.
Adding to the confusion a photon can be referred to as a "light quantum", and some other concepts have similar names because they are physically related.

Also, there are a few inaccuracies. For example, I believe this device would be useful for feeding a light processor the way electrons feed a common cpu (Vdd), and not just for clock signaling.
Please try to keep your articles simple and enjoyable =)
Keep the news flowing. ;)




"DailyTech is the best kept secret on the Internet." -- Larry Barber














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki