digg

Join DiggAbout

Discover the best of the web!
Learn more about Digg by taking the tour.

See the original image at nytimes.com

Replacing Wire With Laser, Sun Tries to Speed Up Data

nytimes.com — The computer maker, which is based in Santa Clara, Calif., plans to announce on Monday that it has received a $44 million contract from the Pentagon to explore the high-risk idea of replacing the wires between computer chips with laser beams.

Bury
show profanity settings
expand all
  • ersatzphi, on 03/24/2008, -6/+43Excuse my transmission ignorance, but aren't fiber optics essentially the laser they speak of?
    • pianomahnn, on 03/24/2008, -1/+29No, not at all. My basic understanding of fiber optics is that they guide light along the physical cable/wire/whatever. These "lasers" are without such physical guidance element. I'm guessing speeds would be hella faster.

      But I'm also ignorant.
      • Gimpyfuzznut, on 03/24/2008, -0/+22Fiber optics send a beam of light across a physical cable as stated above. Basically the light bounces back and forth from top to bottom in the cable until it comes out the other end. The principle is called total internal reflection. This introduces a propagation delay which puts the theoretically maximum speed of fiber at 75% the speed of light.
        The idea behind this is that you would send a laser beam directly point to point (it is not being bounced/reflected back and forth inside the cable). This would technically be able to work at the full speed of light. It would also most definitely be several times faster than existing electrical connections used inside computer chips but wouldn't require the copper/fiber that optics require.
        • fkr3, on 03/24/2008, -4/+175% the speed of light just doesn't cut it. How many speeds of light will the lasers be?
          • Terr01, on 03/25/2008, -0/+1I can't tell if you're trying to be funny or are just failing in an attempt to one-up somebody with science.

            There are two major reasons that it's not "the speed of light". The first is that we typically refer to the "speed of light" as the speed *in a vacuum*, not in glass, plastic, water, etc.

            The second reason is that the laser beam in a fiber-optic channel is making oblique bounces. It's zig-zagging a bit. Because of that, it will take longer than it if traveled in a straight line.

            Both of these combine to make the effective speed of a signal less than the speed of light rays in a vacuum.
        • mitchlourens, on 03/24/2008, -0/+4one x the speed of light, you can't get faster using light.
        • dudettea, on 03/24/2008, -1/+0I learned about total internal reflection in my physics class.
          • Onyxblaze, on 03/24/2008, -0/+1And, it has little to do with fiber optic cables!
          • dudettea, on 03/27/2008, -1/+0Careful, your ignorance is showing.
      • FluffyWolf, on 03/24/2008, -6/+1Yes, you are (and thank you for stating that yourself). Glass has approximately the refractive index 1.5, meaning that the speed of light in glass is 1/1.5 of the speed of light in vacuum. That is not a significantly lower speed, at least not one that will lead to greatly improved computing speeds.
      • DrDragun, on 03/24/2008, -0/+6Believe me the speed of light is not the limiting factor for speed in anything, even computer chips. Even the speed of light through a dense glass or plastic fiber optic cable would not be the limiting factor.

        Really it's about crosstalk as ComputerGod posted about below, and the amount of power density you can get in a small space. Fiber optics still require a source light, and you can't exactly have a bunch of incandescent bulbs generating the light to send through the fiber optics. Lasers are the most dense power to get into a small space.
    • DrDragun, on 03/24/2008, -1/+11Yes but just much messier. Lasers are all 1 wavelength, and most optics have different focal lengths for different wavelengths. So focusing white light is a messy business. Now LED's offer a pretty tight band close to 1 wavelength but LED's cannot be made very powerful.
      • dajuggernaut, on 03/24/2008, -6/+3"Lasers are all 1 wavelength"????
        Looks like someone slept through physics class
        • TomFrost, on 03/24/2008, -0/+7I think he was just summarizing that good quality lasers produce light of a single wavelength rather than the multiple wavelengths you get from any other source -- not that all lasers are the same color.
          • JayD16, on 03/24/2008, -2/+1But there's no such thing as a single wavelength. There are an infinite number of wavelengths in the spectrum because its not discrete.
          • Terr01, on 03/25/2008, -1/+2@JayD16: Even as sophistry goes, that's sort of lame. "Single" is being used to refer to "unified" or "practically identical", not "digitally discrete".
          • JayD16, on 03/25/2008, -1/+1How you're misusing words is irrelevant to me. The fact is there are an infinite amount of wave lengths between 200nm and 201nm just like there's an infinite amount of discrete numbers between 200 and 201.
          • Terr01, on 03/25/2008, -0/+1Neither TomFrost nor I are misusing words: You're simply equivocating over a normal language ambiguitity.
    • computergod, on 03/24/2008, -0/+5Yes, it's exactly what they are talking about. People have been talking about this for as long as I can remember. The big reason for it is that as frequencies get higher and wires get smaller and more close together is that there will be more and more crosstalk, necessitating a new technology free from this interference. They are not faster in themselves, they but they do allow faster communication. Talk of chips, not interconnects, that work on light for much the same reason have also been purposed.
    • krebcycle, on 03/24/2008, -1/+4Fiber optics are large; they're used for transmission of data, not for computation. They're talking about using lasers for transmission inside a processor itself, no more silicon. If they could achieve the above-mentioned 75% of the speed of light with silicon this would be a non-issue.
      • Vlatro, on 03/24/2008, -0/+3Yes and no. They would be creating an optical BUS, not an optical processor. It's a step in the right direction, but not quite what you are envisioning.
    • Vlatro, on 03/24/2008, -1/+3Fiber optics in basic terms is "Sending Light down a strand of fiber". The fiber is usually a glass-like flexible clear tube that reflects and refracts light so it keeps it's course from one end of the strand to the other. There would be no additional latency in using fiber optics as opposed to this proposed method. The biggest issue is that light guided by fiber optics can turn corners. Unguided light relies on Line of sight, which is significantly limiting if they are trying for a compact and cost efficient design.

      The speed increase with what they propose would be negligible to you or I in a home PC, In fact, it would likely seem slower, but spread across a distributed network of thousands of processors, the reduced latency could shave weeks off of complex calculations. Until light is used in photonic micro-circuits (not what they are proposing), it's unlikely there will be any market advantage to this. That's why the article cites it as risky, and only the government will finance it.
    • cannonball, on 03/24/2008, -0/+1I don't see how this could speed up processing. Lasers use electronic chips, so wouldn't that prevent lasers from being faster than the chips that control it?
    • Genma, on 03/24/2008, -0/+1consider their pupose to shrink the tech similar to fiber so that it can replace the integrated circuits on your motherboard. it's more about bandwidth than speed, the problem they're trying to solve is getting data to each processor so that it can keep up with the speed of the silicon. right now processors don't scale because we don't know how to share the data between them fast enough. that's why most parellel computing systems you see today are seperate machines with single or few processors each sharing data between each other, and chip makers are so focused on increasing cores. it's the only method we have so far that scales at all, and it's limited to how small they can etch the circuits into the chips.
    • JayD16, on 03/24/2008, -0/+1"However, the new approach, which is based on the company’s ability to accurately align chips with high precision making it possible to transmit beams of light across the surface of the chips in ultranarrow channels called wave guides, could have a much bigger impact."

      Its not fiber optics.
  • nik3t, on 03/24/2008, -15/+18HALP LASER MOUNTED SHARKS ARE NEXT.
    • Godlike, on 03/24/2008, -8/+3ITS NOS TJUST THE LAYSR MOUNTED SHARKS BUT THERE TEETHS
    • dafragsta, on 03/24/2008, -0/+1Pffft... laser mounted sharks? Call me when they have shark mounted lasers.
    • dave_colorado, on 03/24/2008, -0/+1you *might* get sea bass.
    • TodoInTX, on 03/25/2008, -0/+4No, It's "SPARCs with freakin' lasers on their heads".
  • Magic0, on 03/24/2008, -9/+2the end of the world as we know it..
  • mGARANDEUR1, on 03/24/2008, -5/+2I don't think this will be available anywhere in the near future.
    • Vlatro, on 03/24/2008, -0/+3It's being financed by the pentagon and has no market viability at this time. No one's talking about when it will be available, just remaking on the technology used and the conceptualization of such a process.
      But no one on digg these days will care much unless it comes standard in the next iPod or x-box.
    • theaceoffire, on 03/24/2008, -0/+1Depends. How near are we talking about?
      We will probably see light being used on a chip within 10 years, even if it is not used efficiently.
      • Vlatro, on 03/24/2008, -1/+1You make the same argument as the editor of Boot Magazine, back in 1995. Optical systems on the consumer market have several major hurdles to overcome. There is not presently a cost effecient means of making optical systems as small as required on a mass scale. Even assuming a niche market popped up to buy these at a high price, repair is out of the question should a circuit become damaged, replacement is the only option. While this is standard practice for you and I, the people who buy $6 million machines expect more. Until the can serve that market effectively, there is little funding allocated to the research that addresses the manufacturing problems inherent with bringing it to a mass market. It usually takes something like this (grant money) to get the ball rolling, and many years of research before confidentiality agreements are lifted and enough research has been done to further approve it. By the time we see this technology, it will have evolved to the point that it only in principal resembles what we're discussing here today. It is coming, but I won't hold my breath waiting for it.
  • Hobbes24, on 03/24/2008, -2/+11"high-risk" idea? ...i fail to see how developing the whole laser transfer thing (which has all ready been in development for awhile) is "high-risk"
    • mediaspree, on 03/24/2008, -1/+5if you stare directly into the lazer you go blind!
    • DrDragun, on 03/24/2008, -1/+1Miniaturization and creation of complex arrays in a small space.
    • trispear, on 03/24/2008, -3/+8It's called economics. Any entity has a limited amount of resources. The entity Sun, decides to put resources into investigating this unproven technology and either a)not in other promising techs or b)not in how existing can be incrementally improved.

      If this Tech works -- Sun wins new patents for its profile and can make sales on new fangled widgets. If it doesn't, it may be lag behind competitors for a business cycle or two, lose money, etcetera.
      • Vlatro, on 03/24/2008, -1/+3While I agree with you assessment, and applaud your contribution for being logical and well spoken, it seems people are digging you down. Next time try a one liner involving "X Politician", a 3-letter internet acronym, and a pointless joke that has no bearing on the topic at hand. You get a digg + from me, but the idiots are harder to convince and have the advantage of greater numbers.
      • SpaceMonkeyZero, on 03/24/2008, -0/+3The risk is on the Pentagon, not Sun. They may have just thrown $44 million in order for Sun engineers to go on field trips to see Laser Zeppelin.
        Or the Sun engineers will do something revolutionary, and get their R&D paid for. Double Win!
      • elscorcho717, on 03/24/2008, -0/+2It's a risk for both the Pentagon and Sun but not a big one. This seems promising.
    • krebcycle, on 03/24/2008, -1/+1i can near-hypnotize my dog with my laser-pointer.. perhaps mind control is the fear
    • fahrvergnuugen, on 03/24/2008, -0/+11"high risk" in this case is referring to economics, not physical safety - that is, there is a good chance they may spend millions of dollars on research just to find out it won't work for one reason or another.
    • daborg, on 03/24/2008, -0/+1"i fail to see how developing the whole laser transfer thing (which has all ready been in development for awhile) is "high-risk""

      RTFA.
    • blackjack75, on 03/24/2008, -0/+1They won't need to put warning signs to prevent people from walking over the 'cables'.

      If they do it will cut their feet.
  • DrDragun, on 03/24/2008, -3/+2I'm a little skeptical but hell, they already said this was a high-risk, high-payoff technology.

    Floating dust particles in your comp could block the laser beams, and even if transmitting the light is super-fast and precise, you still have to excite the laser gas to generate that light in the first place. It doesn't seem like a faster physical phenomenon than computer switches but then again I'm a complete n00b on this subject.
    • SpaceMonkeyZero, on 03/24/2008, -2/+3Pentagon putting something in the vacuum of space? Inconceivable!
    • mOdQuArK, on 03/24/2008, -0/+41. If they're using the lasers inside the chips, then dust shouldn't be too much of a problem (since they already assemble the chips in a ultra-clean room). 2. Semiconductor lasers don't use gas.

      The main issues will be, can they make them small enough to fit on a chip with the other circuitry, can they tune them precisely enough so that they can control how they are modulated with the data, and can all this be done without generating so much heat that it melts the chip...
    • vertinox, on 03/24/2008, -1/+2Dust inside your hard drive disk would kill that too.
  • mediaspree, on 03/24/2008, -2/+18Chips with fricken lazers.
  • Skull0Inc, on 03/24/2008, -5/+0read: Mind at light speed: A new kind of intelligence.....it predicts this and even steps in computing beyond optical. However 'we' can only process information so quickly and therefore we have a bottleneck with digesting such large amounts of data.
  • brandon81211, on 03/24/2008, -2/+6Imagine the amount of training needed to open up and work on a MB if this works... That's a whole 'nother book for the A+ cert!
    • geodescent, on 03/24/2008, -0/+2I can't wait for the job opening requiring 10+ years of experience...
  • alex1015, on 03/24/2008, -0/+31High risk? They mean high awesome
    • wafflesomd, on 03/24/2008, -0/+3I can't stop laughing.

      Then again, wake and bake.
  • jmkiii, on 03/24/2008, -0/+6What is that picture, a stuffed radar gun?
  • 955701, on 03/24/2008, -0/+1Why not fiber? It can be multiplexed w/ non-interfering frequencies of light. Imagine a 128 lane bus on one strand.
  • Teck64, on 03/24/2008, -0/+7Title should be : Replacing Wire With Freaking Laser beams, Sun Tries to Speed Up Data.
  • Rwned, on 03/24/2008, -2/+4The Singlularity = one step closer.
  • ysmski, on 03/24/2008, -4/+3Did anyone else read the title and think "Does the Earth's heat source really use wires?"
  • morg666, on 03/24/2008, -3/+4All Sun needs to do is go to the Congo with a gorilla that knows sign language to get the proper diamonds.
    • Ometoch, on 03/24/2008, -0/+2Gotta love obscure movie references. cheers!
  • iwannagofast, on 03/24/2008, -1/+2IBM is ahead of the game with its optical interconnects
    -The main advantage of all of this is heat. Also it can remove many components between processors.
  • houndeyex, on 03/24/2008, -1/+3Hi, I'm a bird, flyin' in your laser beam.
  • Topher06, on 03/24/2008, -2/+3I am sorry, but its been a very long time since I looked at Sun as a technology innovator. IBM already announced optical interconnects, Sun is doing their usual trick as of late and that is to take someone else's innovation and put their brand on it.
  • SirZRX, on 03/24/2008, -0/+3pew pew mobos =D
  • dodgyd55, on 03/24/2008, -0/+1id like a really fast processing computer. but they will take ages for us to get them T_T
  • mcse2k3, on 03/24/2008, -2/+1But WHY?
  • isaactwito, on 03/24/2008, -0/+1I hope lasers replace all wires in this world, it'd be so trippy.
  • pinoyboy82, on 03/24/2008, -0/+2just as long as they name the new processor "The Shark"
  • Shadowgamers, on 03/24/2008, -0/+1Affordable supercomputers, COMING TO A STORE NEAR YOU
  • DeFex, on 03/24/2008, -0/+3I wonder how much they will have to pay laser LED patent granny
  • 10GunSalute, on 03/24/2008, -1/+3Anything that makes their horrendous Java applications any faster is fine with me.
  • bincoder, on 03/24/2008, -1/+1Why not 256 processors at 1 Ghz each running as a cluster? Thats faster and cheaper than any single chip limited by the speed of light.
  • buddamus, on 03/24/2008, -0/+1I was talking about that with a friend of mine doing a PHD in computers, I should have thought of it sooner
  • eric2006, on 03/24/2008, -0/+0The real motivation, of course, is that if you put it in a clear box, it will look SWEET.
  • jazzdonkey, on 03/24/2008, -0/+1Wait...can someone explain the difference between this and IBM's theory using light pulses between cores?
  • alvarezg, on 03/24/2008, -0/+1Surely not open-air laser beams! Ever see all the dust bunnies inside a computer?
  • BobaFett2ha, on 03/25/2008, -0/+0"The technology, part of a field of computer science known as silicon photonics"
    Computer science...? No...

    I actually work at the UCSD lab doing this stuff.
  • truthseeker69, on 03/25/2008, -0/+1Umm....photonics...ftw!
  • callmeTroy, on 03/25/2008, -0/+1Awesome! its like Doom3!
  • bdbElysian, on 03/25/2008, -0/+1"the high-risk idea of replacing the wires between computer chips with laser beams." High risk? I'm sure It'll work. I'll bet a 95% success rate not 50.

    And I'm also sure that any alignment issues will be resolved by auto alignment systems suitable for machinery that take massive abuse which includes residential PCs/notebooks.
  • strypersarmy, on 03/25/2008, -0/+1would it be easier for the gov to spy with lasers?
  •  

    Login or register to add a comment

    Create a new account or login to join in the conversation on Digg. You'll also be able to Digg stories to help promote things you like.


© Digg Inc. 2008 — Content posted by Digg users is dedicated to the public domain.
DIGG, DIGG IT, DUGG, DIGG THIS, Digg graphics, logos, designs, page headers, button icons, scripts, and other service names are the trademarks of Digg Inc.