Page 1 of 1

Optical CPU?

PostPosted: Wed Oct 29, 2003 8:26 am
by Mithrandir
According to a Reuter's article there is a company in Israel that has developed an optical CPU. I know a lot of companies around here are working on the switching technology to integrate something like this into a man stream system, there aren't any officially on the market that I know of. Here's the article if you want to read it:
http://www.reuters.com/newsArticle.jhtml;jsessionid=JWF0QE5YPBHMMCRBAEOCFEY?type=scienceNews&storyID=3711894

PS: I don't think I'm quite as optomistic as their CTO about this being mainstream in 10 years...

PostPosted: Wed Oct 29, 2003 9:31 am
by inkhana
Whoa, that's cool...I'd like to see them get that out into the markets quickly...LOL ;)

PostPosted: Wed Oct 29, 2003 9:36 am
by TheMelodyMaker
I'm sure it would eliminate the need for CPU cooling fans, that's for sure. ^_^

PostPosted: Wed Oct 29, 2003 9:44 am
by Mithrandir
I dunno, think about the kind of cooling you need with that many lasers... Now, if you could get the bus lined up just right, then the optical CPU would just be reflecting light different ways. THAT would eliminate the need for fans. But then you would have to have repeaters... I dunno, I get the impression that fans will be around for a while. But what do I care, my G5 tower is nice and quiet. :)

PostPosted: Wed Oct 29, 2003 9:57 am
by TheMelodyMaker
I suppose you're right. In my mind, laser beams don't generate nearly as much heat as electrical current does; but maybe I'm wrong about that.

PostPosted: Wed Oct 29, 2003 9:59 am
by Mithrandir
I suppose that depends on the power of the laser and the speed of the current. Where's technomancer when you need a run down of heat loss phyics?!?

;)

PostPosted: Wed Oct 29, 2003 2:45 pm
by Technomancer
:)

Neat, I'll have to see if there's a write up in one of the journals, the article's a bit skimpy on details. I know people have been working on this for a while, but a lot of the work seemed to focus on optical busses, rather than transitors. We'll see how it works out, I remember the hype about holographic storage several years ago.

Unfortunately, I haven't done device physics in a long time (2nd year materials course), so I couldn't give you the really good details. Hmm. You'll eliminate resistances between devices, as well as parasitic capacitances. A lot will depend on the efficiency of the device itself, and I'm not realy up on diode lasers. I think I'll check IEEE and see if I can find anything useful.

PostPosted: Wed Oct 29, 2003 3:04 pm
by Mithrandir
:) Tuché. I've got an opticals specialist friend that owes me a favor, and now I'm curious. Hm... Where did I put that number...

PostPosted: Wed Oct 29, 2003 7:03 pm
by andyroo
That's pretty interesting. Does that by chance have any relation to "atomic computing?" I've read up on that before. The most I've heard one of those do was 3 x 5 = 15. Of course, that was something like a year or two ago.

PostPosted: Thu Oct 30, 2003 1:54 pm
by Straylight
I'm wondering, how do they implement logic gates and stuff like that in an optical system?

PostPosted: Thu Oct 30, 2003 2:12 pm
by Mithrandir
andyroo wrote:That's pretty interesting. Does that by chance have any relation to "atomic computing?" I've read up on that before. The most I've heard one of those do was 3 x 5 = 15. Of course, that was something like a year or two ago.


I'm guessing you are refering to the quantum computer projects, one of which was able to 'instantly' factor 15. I haven't seen much on them. The probablem with those is that you have to 'program' them with highly precise lasers, etc. I'm thinking they won't be mainstream for quite some time yet. I'll you all know if I run across any new developments, though.

PostPosted: Thu Oct 30, 2003 2:24 pm
by Technomancer
noz:

There's a pretty good write-up in the April/May 1996 issue of IEEE Potentials magazine. Basically, you first need to implement a transistor. To do this, some material with a non-linear refractive index is needed, and it must vary with the intensity of the incident light. In one region of operation (with low input intesity), the index is such that little light is transmitted. When operating in this region, small changes in intensity, result in small changes in the amount of light transmitted. Past a certain point, however you cross a threshold, and the amount of tranmitted light dramatically increases, and a new plateau is reached.

In order to make a transitor (transphasor), a laser beam with an intensity just below threshold shines continously on the crystal. A second, lower intensity beam can be used to modulate the output, when it is also shone on the crystal, pushing the incident intensity above threshold. This is pretty similar to how electronic transistors work, and once you've got transistors, you can then build logic gates.