Engineers Report Breakthrough in Laser Beam Tech

petralynn writes to tell us the New York Times is reporting that Stanford engineers have discovered a method to modulate a beam of laser light up to 100 billion times a second. The new technology apparently uses materials that are already in wide use throughout the semiconductor industry. From the article: "The vision here is that, with the much stronger physics, we can imagine large numbers - hundreds or even thousands - of optical connections off of chips," said David A.B. Miller, director of the Solid State and Photonics Laboratory at Stanford University. "Those large numbers could get rid of the bottlenecks of wiring, bottlenecks that are quite evident today and are one of the reasons the clock speeds on your desktop computer have not really been going up much in recent years."

Comments

More informative article:

[TripMaster Monkey]

The NYT story is pretty light on the technical details....a more detail-oriented write-up can be found here... and you don't have to register to read it.

Re:More informative article:

[AKAImBatman]

I was just about to point out a fun trick of searching Google News for the article to read it without registering. Then I clicked on the link and realized that it doesn't require registration. How odd...

Re:More informative article:

[olddotter]

Thanks. I hate the lite on science in main stream reporting!

Re:More informative article:

[fishybell]

Well, that link is a little better. I recommend this one instead, since it has a coherent summary and has numerous comments from industry analysts.

Re:More informative article:

[amliebsch]

Or, if you really want technical, skip the write-up altogether and read what some of the Intel engineers had to say.

The Silicon Solution

Re:More informative article:

[Anonymous Coward]

Don't you EVER get tired of being a fscking frist-psoting, karma whore??

Re:More informative article:

[Guppy06]

Come now! Industry analysts at least get paid for not having a girlfriend!

Re:More informative article:

[Anonymous Coward]

Dudes own write-up...

http://www-ee.stanford.edu/~dabm/304.html

Re:More informative article:

[JJman]

What? You hadn't already read the article in IEEE Spectrum?
What kind of a geek are you?

Re:More informative article:

[macdaddy357]

If they want to do something really cool with "laserbeams", they need to put them on sharks' heads!

Re:More informative article:

[orac2]

Hee. You just made my day. (I'm an editor at Spectrum!).

Does that mean...

[Signal_Noise]

That I can disco dance a billion times faster?

Re:Does that mean...

[Anonymous Coward]

That I can disco dance a billion times faster?

No, but it does mean we can now install gatling laser turrents on our corvettes to intercept missiles from those pesky planetary defense bases!

(Master of Orion reference)

100 billion hits per second

[thewiz]

Looks like the NY Times servers can't handle the /.ing!

Moore's Law Finally Broken?!?!?!?

[Anonymous Coward]

just kidding. I hate all the people that post that without knowing shit about it. As this applies to optics and not semiconductors, it really doesn't have anything to do with moore's law.

Re:Moore's Law Finally Broken?!?!?!?

[Anonymous Coward]

I hate all the people that post that without knowing shit about it. As this applies to optics and not semiconductors, it really doesn't have anything to do with moore's law.

From the article:

Several industry executives said the advance was significant because it meant that optical data networks were now on the same Moore's Law curve of increasing performance and falling cost that has driven the computer industry for the past four decades.

Doh! Don't you hate it when you get all high and mighty posting about people who don't know what they're talking about and then find out that you don't know what you're talking about?

Re:Moore's Law Finally Broken?!?!?!?

[Anonymous Coward]

Since WDM the optical data rates actually increase at a rate faster than Moore's Law for electronics!

Re:Moore's Law Finally Broken?!?!?!?

[Salvage]

It's not all that accurately worded, but it is relevant. The lack of accuracy is likely due to trying to keep that comment short.

In any case, while Moore's Law is specific to transitor based circuitry, the pattern is applicable to other technologies, such as Kryder's Law which covers rigid magnetic media (hard drives). In fact, looking at these cases in general within a field of technology suggests a more abstract pattern. After all, the original component technologies with which Moore worked when he made his observations have been replaced over the years, some of them multiple times, with the the common thread to all of them being that they ultimately deal with transitors.

If optical technologies get pulled in by the same economic factors that drive Moore's and Kryder's Laws, they'll very like fall into a similar pattern: doubling of a particular characteristic over constant intervals.

Of course, all of this also depends on how how close a class of technology is to its fundamental extreme physical limits. For instance, density of transistors is ultimately limited by the size of atoms; the limit there may be somewhere around a "one molecule transitor." In the particular case of the article, the technology is optical modulators and the measure is switching rates. For that, one limit may be the frequency of the transmitted light. The visible spectrum runs from 384-769 THz, with the higher frequencies more difficult (in general) to generate. All this in turn suggests an upper limit of around 700 trillion switchings per second. With a Moore's or Kryder's Law like rate, say doubling bit rate every two years, today's 10 billion bps goes to 700 trillion in about 33 years.

Re:Moore's Law Finally Broken?!?!?!?

[ZachPruckowski]

Yeah, but the hope is that in 33 years, we have something newer. I mean, by then, we'll hopefully have three-dimensional chips, or quantum stuff, or something we haven't even thought up yet. And I'll be reading this article's clone on Slashdot, and we'll have the exact same discussion. Except I'll have taken over the world by then.

Re:Moore's Law Finally Broken?!?!?!?

[Buelldozer]

Well gee, if you're going to be ruling the world...could ya' take a minute or so and fix the dupe problem?

Then you wouldn't have to read the article again(!) sometime in the mysterious future...oh and bump my karma up some! :-D

Re:Moore's Law Finally Broken?!?!?!?

[ZachPruckowski]

Don't worry, it's on the list. But first I have to exile Bill Gates, cure world hunger, etc... But rest assured that Slashdot is, was, and always will be, on that list.

BugMeNot shortcut for 'ya ...

[xmas2003]

NYT registration required to read this John Markoff (infamous at Slashdot because of his "sensational" coverage of Kevin Mitnick) article ... but fortunately, BugMeNot comes to the rescue with username/password of "twernt/twernt"

This work was funded by Intel and DARPA with some assistance from an HP researcher and uses something called the Quantum-Confined Stark Effect with primary application in optical networking gear ... but hey, maybe we'll see a 100 GHz PC in the not-too-distant future.

The halloween webcam is up ... but X10 technology isn't capable of 100 Billion times/second updates ... ;-)

WTF is Standford University?

[winkydink]

Is that the one across the bay from Berkly?

Re:WTF is Standford University?

[Reality Master 101]

And, more importantly, does Standford's football team suck as much as Stanford's?

Re:WTF is Standford University?

[Anonymous Coward]

It's the West Coast's answer to Harfurd.

Re:WTF is Standford University?

[FuckTheModerators]

It's right down the street from those three electronics factories: Magnetbox, Panaphonics, and Sorny.

Re:WTF is Standford University?

[nherm]

No, it's STandFord University.

Have you seen their T-skirt?

Re:WTF is Standford University?

[kclittle]

I keep correcting you guys, but nobody listens! It's "Bezerkeley"! I know, I went there. (At least, I'm told I did -- I really don't remember much about those years.)

Re:WTF is Standford University?

[SoCalEd]

Yep. Call themselves The Standford Cardinals.

Re:WTF is Standford University?

[Cyclon]

Don't you mean "Berzerkeley"?

Modulated the power of my laptop

[jurt1235]

At about the same frequency. Now you can do a lasershow in the cloud of smoke. Does that count as research too?

Modulating Laser...

[mcsestretch]

But will it pop a huge jiffy-pop container in my professor's house by shooting it from a plane?

Re:Modulating Laser...

[DarthStrydre]

I know I remember seeing the end of this movie at one time, with the laser, airplane and popcorn, but Google isn't turning up anything, and I cannot remember what the movie was...

Can anyone give me a hand identifying the parent's reference?

Re:Modulating Laser...

[nullset]

Real Genius is the movie you're looking for, starring Val Kilmer. I love the promotional poster of Val Kilmer wearing an Einstein shirt, and Einstein wearing a Val Kilmer shirt.... (God) Kent, have you been touching yourself? Kent: yes....I mean NO!

Re:Modulating Laser...

[SatanicPuppy]

Real Genius

All I wanted...

[parasonic]

...was chips with frickin' laser beams!

Re:All I wanted...

[Admiral Ackbar 8]

[holds pinky up to mouth] 100 billion modulations!

Speed of light vs. speed of electrons in wire?

[sdo1]

I guess I naively thought that the speed of electrons in a wire was roughly the same as the speed of light, or at least "nearly" the speed of light. Can someone explain what the real advantage is?

-S

Re:Speed of light vs. speed of electrons in wire?

[cheddarlump]

Actually, I believe the speed of electrons moving through wire is roughly 1/10th the speed of light in a vacuum.

Re:Speed of light vs. speed of electrons in wire?

[Anonymous Coward]

The heat and electromagnetic noise generated by wires are a problem. Also wiring picks up noise which can lead to fliped bits.

Re:Speed of light vs. speed of electrons in wire?

[Incongruity]

You need to differentiate the drift speed of the particular electrons (this can be quite slow, esp. in AC) and the speed of propagation of energy, which if I recall is damn fast (near C, but not there...granted, 1/10 of C is still astoundingly fast, so my poor memory of freshman physics may not contradict you, though I think your guess is off)...the real advantage is that the switching speed is far beyond what we can do with current metal/electron based circuits (rtfa) . Additionally, this is big because using electrons generates more heat and is subject to induction/capacitence effects that light isn't. So those would be the main advantages, as I understand it... but I only play a physicist on /. so feel free to correct me, cruel world.

Re:Speed of light vs. speed of electrons in wire?

[DarthStrydre]

The speed of the electrons is on the order of cm/s, and is related to the current density.

The electromotive force, or voltage, travels at about the speed of light.

Picture a hose of water. The water (electrons) takes a long time to get from one end to the other... but the effect of putting water in one end is immediately seen at the other end (within reason).

With AC, electrons never really gain ground in a balanced load situation. Back and forth and .. . . .

Re:Speed of light vs. speed of electrons in wire?

[Detritus]

The drift velocity of electrons in a conductor is actually very slow, 0.1 mm per second is a typical value. It depends on current density.

Re:Speed of light vs. speed of electrons in wire?

[Red Flayer]

The modulation. The signal travels at about the same time, but you can turn it on and off much much faster... so the density of bits per unit of time is much higher.

Normal signal: ____----____----____----

0 1 0 1 0 1

New hawtness: _-_-_-_-_-_-_-_-_-_-_-_-

010101010101010101010101

Both took the same amount of time to travel down the pipe. But one conveyed 4x the information.

Re:Speed of light vs. speed of electrons in wire?

[mrtroy]

Standford engineers have discovered a method to modulate a beam of laser light up to 100 billion times a second.

So, you still have the same "speed" but you have way higher modulation than what is possible through a traditional chip.

So, if you have two highways, both going the same speed, but one is filled with dinky cars and one is filled with transport trucks....in which highway can you have more total cars get through?

So the data gets there in the same length of time, but you have data sending/arriving much more frequently, so total throughput is up.

Re:Speed of light vs. speed of electrons in wire?

[joe_bruin]

The speed of electricity in a wire is not really the issue (it's about half the speed of light, I think. I'm sure someone will correct me). The real issue is signal propagation. When a transistor switches from closed to open or back, the electrical signal travelling through the wire is not a perfect on/off. The voltage ramps up or ramps down as some function of the length of the connection, width of the wire, conductivity, leakage from the transistor, inductance, ... The system needs a bit of time to "settle" into the new high or low state. This is a big limiting factor in the clocking of modern CPUs. For communication off the chip, it's far worse. Now the lines are no longer 90nm (or whatever the chip was made at) in width, and have to go through a far longer distance. That's why today's processors are limited at around 1GHz to the outside world, while internally they can be faster.

Optical interconnects alleviate many of these problems. With a laser, the ramp up time is significantly shorter, there's no capacitance in the system, and it is far less prone to interference. So, on a 100 GHz optical link you can multiplex 100 1GHz pins (essentially running a P4's FSB on two wires instead of something like 180), thereby significantly reducing the pin count. Or you could run the pins 100 times as fast, meaning much less processor waiting on RAM or bus data.

Re:Speed of light vs. speed of electrons in wire?

[Chris Burke]

Yeah, that's not true. I don't know how fast an electron moves (I'm assuming not the speed of light, since they have mass, and that quantum physics I know little about probably comes into play), but in a normal conductor they don't move very far before slamming into something. Individual electrons don't move that far or fast on their own, it's the aggregate and resulting field that really moves.

But that's not really the problem. Transmit time is still quite low (I've heard 1ns per 6 in of trace on a board). Latency isn't really the problem. The problem is -- how fast can you change the signal? That's bandwidth. Here electrical conductors suffer because of parasitic capacitance and inductance, skin effects, reflections, induced current from nearby conductors, and a whole host of other signal integrity issues. It gets worse the longer the channel is and the more things you have connected to it. If you're wondering why the MP Pentium 4s have been on a 100MHz QDR front side bus since they were released, this is why. It's also why even point-to-point interconnect like AMDs has only recently broken 1 GHz.

Optics don't really have this issue. Two fiber optic cables next to each other don't interfere with each other. You don't have to overcome the capacitance of the channel to change from one value to the next. You just send photons of one frequency, and then switch to the next. As fast as you can switch is how much bandwidth you can get.

Alright, I'm not really liking this explanation anymore. To just directly answer your question: the advantage is 100 GHz interconnect in a way that could potentially be built into chips.

Re:Speed of light vs. speed of electrons in wire?

[Rob the Bold]

The speed an electric signal will propogate in a transmission line is somewhat less than 1C. The value of 0.1C in a sibling post is a good rule of thumb. Think of your transmission line as a bunch of inductors in series and a bunch of capacitors in parallel (imagine a ladder with inductor legs and capacitor rungs). At each step along the way you need to charge up the capacitor before current will move to the next inductor, where your current will charge up the magnetic flux and then on to the next cap, etc.

You can build what's called an "aritficial transmission line" in just such a manner. It simulates the effect of a much longer pair of wires for lab purposes.

Re:Speed of light vs. speed of electrons in wire?

[9Nails]

They have not increased the speed of light. They however increased modulation. That is the light flicking on and off. If you figure that each time the light flicks on, that's a 1. And each time the light flicks off, that's a 0. So now you can send a whole lot more 0's and 1's to a remote computer. In essence, they've created the laser that you will need in order to improve Internet bandwidth. Assuming that they can get it to shrink down and fit inside a router, and run over fiber, without the expense of heat or added power, etc...

Re:Speed of light vs. speed of electrons in wire?

[IvyKing]

First off, the electron velocity in wire is much less than the propagation velocity through the same wire.

Now for the fun part - What is the velocity of propagation?

For frequencies were the inductive reactance of the conductor is significantly larger than the resistance of that conductor at that frequency (think skin effect), then the velocity of propagation is c divided by the square root of the effective relative dielectric constant. This is often referred to as an LC transmission line since propagation is dominated by the series inductance and shunt capaitance. LC lines have a propagation velocity independent of frequency (at least to the first order). As an example, coaxial cable with a solid polyethylene dielectric will have a propagation velocity of 0.66c, which would be valid from a few hundred kHz to several GHz.

When the the conductor resistance is greater than the inductive reactance, then the line becomes an RC line where the "propagation velocity" is dependent on frequency (dispersive) and the time for a transition to propagate is proportional to the square of the line length. The effective "propagation velocity" is going to be a lot less than c. Turns out that the interconnects on chips are RC lines - and it is often necessary to insert inverters on a line to speed things up (recall that propagation time varies with the square of the line length) - a good rule of thumb is to space the inverters so the the propagation delay equals the gate delay.

The RC problem is why loading coils were put on phone lines - the inductive reactance of the coils is larger than the resistance and the line becomes an LC. The loading coils are bad news for DSL - and an unloaded line looks like an LC line at the frequencies used by the DSL modems.

A good reference for this is High Speed Digital Design, a Handbook of Black Magic by Johnson and Graham.

Re:Speed of light vs. speed of electrons in wire?

[freidog]

Well, on that scale, the wire acts as a capacitor. You have to put x amount of charge into the wire before you get to your desired voltage on the far end. 10 years ago, the delay associated with charging a global wire (something on the order a few milimeters long) was insignifigant compared to the delay from charing a single logic gate. Today, the dealy of that same wire is several times that of a single logic gate. In a 45nm process it might be greater than an entire stage of logic (15-20 gate delays).
Obviously, light has no such pre-charge requirement.

Using light - at least on the longer wires in a chip - could mean going back to a time when transmitting signals 'long' distances in a chip has an insignifigant cost.

Now, that's just one of several problems that exist with wires - particularly copper wires - in modern processes, but it is a big one.

Re:Speed of light vs. speed of electrons in wire?

[Anonymous Coward]

Good explanations all, but there's more to it than just the speed.

If you can interconnect optically, you eliminate a wide variety of problems that limit the speed of electrical interconnects:

• inductive effects (since current flowing in a conductor has electromagnetic effects on adjacent conductors)
• capacitative effects (since adjacent conductors at potential but insulated from each other form a capacitor whether you desire one or not)
• microphonic effects (due to inductive and capacitive effects above, if the substrate, conductor or cable allows conductors to move relative to each other, you've effectively made a primitive microphone) - this can also be called electromechanical noise

The above items largely limit how fast you can clock nearby conductors, and how close you can put them together (thus limiting overall size)

There are probably other positive attributes (better slew rate, maybe? lack of thermal effects?) and probably more than a few problems specific to optical interconnects (traditional silicon bottleneck at each end of interconnect, optical equivalents to interconnect noise, etc), but this looks like pretty good technology!

Re:Speed of light vs. speed of electrons in wire?

[Anonymous Coward]

You have obviously never taken a shower at my house. I turn on the shower and can read a few slashdot articles before my water comes out.

Re:Speed of light vs. speed of electrons in wire?

[osobear]

The speed of the electrons is on the order of cm/s, and is related to the current density.

Slightly more correctly, the drift velocity of electrons in standard copper cable is on the order of (tens of) cm/s. Actual electron velocity is close to c (as they bounce around in a cable), and electron drift velocities can be on the order of 10^7 m/s in some media.

Re:Speed of light vs. speed of electrons in wire?

[Antonymous Flower]

Nothing moves at the speed of light (in a vacuum) as the energy required to accelerate near the speed of light increases (presumably) asymptotically. Current is a measure of the speed of electrons in a conductor if the number of free electrons per volume of the conductor is known. Because the capacity will be constant, one can get a general idea of current as proportional to the speed of electrons in a conductor.

Re:Speed of light vs. speed of electrons in wire?

[budgenator]

It helps to think of the wires as a capacitor, which take a while to charge up to a trip voltage. Presently CPU's are trying to flip the bits about as fast as the wires can be charged, so they are making the wires smaller to charge faster and run cooler. Smaller wires means more bad chips and more expensive cpu's, using light for signaling gets around big parts of the problem.

Re:Speed of light vs. speed of electrons in wire?

[vertinox]

Can someone explain what the real advantage is?

Heat.

I could be wrong, but I would think an optical chip would produce less heat since it would have less resistance and would scale better in power consumption.

I'm not engineer so I could be wrong about that assumption.

Re:Speed of light vs. speed of electrons in wire?

[hr raattgift]

You just send photons of one frequency, and then switch to the next. As fast as you can switch is how much bandwidth you can get.

Most optical signalling is done with blinking on and off rather than other modulations. This is mainly because collimated coherent light sources make this approach really easy on both the transmit and especially the receive side.

However, through a medium like glass fibre, wave propagation speed varies nonlinearly with frequency. In other words, two identical-length pulses on different frequencies transmitted at the same time may be detected at different times. This becomes important when stepping up the pulse rate, because signals sent earlier on one frequency may be detected later than those sent later on a different one. However, even within one frequency, the pulse intensity will spread out over time because real light sources are not truly monochromatic, and because the signals may be rotated in terms of polarization, so that a rapid off-on-off may wind up looking like it has more than one "on" pulse. Although these effects increase with distance, and the distances within a single system (or especially a chip-sized package) will be small, they also increase with signalling rate.

These limitations don't look particularly scary in comparison with electronics, especially since there are other media -- including vacuum -- through which one could send the optical energy. However, at the moment there is very little real work that can be done in a purely optical system, so there will be E->O conversions (which is what the article is about) and O->E ones. Converting high bandwidth optical signals into something useful for electronics is currently a harder problem, and in telecommunications is essentially sidestepped with multiplexing.

On the scale of metres to megametres, optical pathways are clear win, mainly for the reasons you listed.

However, the reason that EO-OE links interconnecting chips via fibre aren't more popular is that on the scale of a few centimetres to a few millimetres it's an expensive engineering choice even in comparison to the difficulty in shrinking the distances between chips (to the point of putting multiple chips in a single package) or increasing the multiplex fan-out across more pins/paths/traces while maintaining a relatively slow clocking rate.

Re:Speed of light vs. speed of electrons in wire?

[fyngyrz]

I'm thinking you wouldn't want vacuum over your chip... acts as an insulator, so heat would accumulate.

Re:Speed of light vs. speed of electrons in wire?

[grimJester]

The speed of electricity in a wire is not really the issue (it's about half the speed of light, I think. I'm sure someone will correct me).

Both the speed of electricity in a wire and the speed of light in fiber are about two thirds of the speed of light in vacuum. No, not an issue.

There's still a problem...

[josephdrivein]

Well, actual technology has a maximum bandwidth of 10-100Ghz... How do you elaborate such a quick signal?

Re:There's still a problem...

[Sebilrazen]

... How do you elaborate such a quick signal?

DO NOT TRY THIS @ HOME

1. Get a box(es) of toothpicks (flat preferably), a fast car with the rear passenger door removed, and a trusty sidekick.
2. Proceed to place the toothpicks at one toothpick's width intervals for a significant distance down a straight quarter mile stretch of road.
3. Starting from 1/4 mile away, while lying in the backseat with your head out the door hole and your accomplice driving, accelerate to a good fraction of the vehicle's max speed.
4. Try to catch glimpses of the pavement through the toothpicks.
5. Profit???

Rawghs.

[Brantano]

What does this mean for laser pens?!

Re:Rawghs.

[mikael]

It's obvious - you will be able to write much faster, but will take the same amount of time to read.

Re:Rawghs.

[Anonymous Coward]

nope - it's called meaurement. You see by telling us that they can modulate at 100gbps theyre also saying that they can measure this data rate with a detector. Hence, it can also be read at that speed.

Re:With great bandwidth comes great responsibility

[Anonymous Coward]

No, no no...

With great bandwith comes great amounts of porn.

I'm still betting on qubits

[neologee]

Having a look at Quantum computers development, i still think that this will be the next Big leap in performance!
By 2010-15 (that's already pretty far out) i could imagine processors will need a complete new mechanism to go any faster.

Re:I'm still betting on qubits

[GameMaster]

Quantum computers are great, in theory, but even if we are able to figure out how to build one that actually works they are only capable of solving certain types of problems. Our present understanding of quantum physics tells us that you can't design a quantum computer that can do all the same math problems as a generic Intel/AMD CPU (e.i. run Windows; play Counterstrike; etc.).

That being said, the problems that can be solved by quantum computers tend to be the ones that would take a regular CPU until the end of the universe to perform (break strong encryption, large traveling salesman problems, etc.). At some point, if we can make a quantum computer compact enough, we might end up having quantum co-processors built into out PCs but we'll probably never see the CPU of our PC replaced by a quantum computer.

The tech being discussed in the article would be directly applicable to making generic PCs run faster (though it could also have the potential to improve communication speeds with a hypothetical quantum computer as well). Another tech that will probably be leveraged to make generic systems faster is the replacement of silicon in computer chips with diamond. Since diamond can handle vastly higher temperatures than silicon, without melting, it is theoretically possible to push the clock speed on a diamond based CPU much higher than on today's silicon CPUs.

-GameMaster

Re:I'm still betting on qubits

[Heembo]

Mommy, can I have $25,000 to buy the new 10k Ghz Dell Diamond Dimension pleeeeeeeeeeeeeeeeeease?

Re:I'm still betting on qubits

[Lobachevsky]

> Our present understanding of quantum physics tells us that you can't design a quantum computer that can do all the same math problems as a generic Intel/AMD CPU (e.i. run Windows; play Counterstrike; etc.).

I'm going to have to debunk you on that. QCs require the ability to:
1. CPhase (rotate amplitude on the real-imaginary 2D plane)
2. CNot (quantum entanglement)
3. Mix (reset amplitude to random)
4. Measure (collapse of the probability wave)
and, importantly, 5. NAND (not-and classical gate, fountainhead for all classical and contemporary logic).

The operations 1-4 allows us to model quantum effects - remember, the whole idea for QCs spawned when Richard Feynman asked his CS buddies to model quantum effects and despite their CS prowess, none could model it efficiently, which Feynman rightly concluded as indicating that the Universe had a faster computer to predict and enact quantum decisions than the computers we built; hence, he reasoned, we're using an inefficient construct and tapping into the Universe's compuations of quantum effects will allow us to build a more powerful (time-complexity-wise and not just boosted by a constant-time coefficient) computer. There is nothing a classical machine can do which a QC cannot. A QC must have NAND, else it's rather pathetic with just CPhase, CNot, Mix and Measure.

And the "Windows/Counterstrike" argument is a bit lame - even Sparc processors can't run Windows/Counterstrike, that hardly makes them inferior to Intel/AMD. Assembly-interoperability and Mapping-reducibility are very different things. Transmeta went for the asm-interop route, but any computer scientist will tell you that as long as you can simulate a NAND and have O(1) random access memory, any functionality of a modern day Opteron is mapping reducible to your simulation.

> That being said, the problems that can be solved by quantum computers tend to be the ones that would take a regular CPU until the end of the universe to perform (break strong encryption, large traveling salesman problems, etc.).

Agreed. But a QC is never worse in complexity than a classical computer. It is always equal or better complexity-wise. It may have a horribly constant coefficient, making it 1/100th as fast for operations a classical computer has equal complexity. However, keep in mind that a mere constant of 1/100 or 1/1,000 or even 1/1,000,000 pales in comparison when a QC can complete in O(n) time what a classical computer requires O(n^1.5) time.

Re:I'm still betting on qubits

[GameMaster]

I will admit, I'm not a physicist. However, while I was studying Computer Science at college I had two occasions to study quantum computers. One occasion was a guest professor at my parallel computing class who covered how software would be developed for a hypothetical Quantum Computer and the other was a colloquium run by my CS department with another guest speaker discussing almost the same topic. The question of playing Doom on a quantum computer came up both times and they both agreed that quantum computers are incapable of performing all of the basic ops needed for a general purpose CPU. This was, supposedly, one of the reasons that there were such a limited number of algorithms in existence to justify the creation of a quantum computer in the first place (Shor's factoring algorithm and a few others).

As for running Windows/Couterstrike, it is perfectly possible to run both of those pieces of software on a Sun Sparc system. I am not now, nor was I then, suggesting that either piece of software would run on a Sparc out of the box but it is completely possible to take them and modify them to run on a different platform. The basic algorithms can be run on the Sparc CPU to produce the exact same output. My point wasn't to suggest that Intel/AMD processors are special compared to any other CPUs (Sparc, Alpha, PowerPC, PIC, Cell, etc.). I simply used Intel/AMD as examples of two of the most well known general purpose CPUs. Truth be told, I didn't feel the need to list all general purpose CPUs ever invented nor did I think myself capable of listing them all without missing some. The point I was trying to make was that, while I can convert Windows/Counterstrike to run on a general purpose CPU like Intel/AMD/Sparc, I couldn't do the same for something like a nVidia Gforce GPU because, while they are very programmable and parallel processing friendly, they are missing some core ops needed to run generic software (this may have changed recently with the newest generation of GPUs). As I said before, my experience in hearing the guest professors discuss the specific topic of algorithm development for quantum computers led me to the belief that the same holds true for quantum computers.

It has been maybe six years since I heard those professors speak about quantum computing, so I am more than willing to grant that things may have advanced in the field of quantum computing algorithms and new ways may have been found to emulate the missing ops. However, I haven't seen anything myself to suggest that that is the case.

-GameMaster

Re:I'm still betting on qubits

[Lobachevsky]

I doubt they would have said quantum computers are incapable of ever running Doom or such; they most likely qualified their statement, indicating that they would be slow by some constant time for such tasks, or they would cost too much in the near future to be practical. However, mathematically, it would be unscholarly of them to say without qualification that QCs are incapable of general purpose computing. See here:

Practical quantum computers
David DiVincenzo, of IBM, listed the following requirements for a practical quantum computer:
scalable physically to increase the number of qubits
qubits can be initialized to arbitrary values
quantum gates faster than decoherence time
Turing-complete gate set
qubits can be read easily

It's important to note the inclusion of Turing-complete gate set -- this with O(1) memory access guarantees the ability to run a VMWare-like app and simulate any x86 processor, albeit by some constant time performance penalty.

Who??

[Maradine]

petralynn writes to tell us the New York Times is reporting that Standford engineers. . .

That's awesome. I can't wait for Hraverd and Yalle to catch up.

Re:Who??

[Procrastin8er]

Yes, MTI too.

Re:Who??

[MyHair]

"So, you went to Yale?"

"Yes, I yust got out last week!"

Re:Who??

[IceFoot]

Make that Hah-vahd and Yael.

I can't wait until Intel makes a chip with it

[Anonymous Coward]

It'll probably put out as much heat an light as the sun.

Re:I can't wait until Intel makes a chip with it

[hcob$]

It'll probably put out as much heat an light as the sun.

At least then you'll never need to get up to get rid of that monitor tan and you can fix a grilled cheese sammich at the same time!

BRRRRRRRRiiiiilliant!!

Re:I can't wait until Intel makes a chip with it

[stevel]

Well, there is this...

Dear mister engineer

[Killjoy_NL]

I love you

Sincerely,

a hardware enthousiast (and laser lover)

Obligatory Reference

[Stibidor]

So can they attach them to sharks' heads yet?

Re:Obligatory Reference

[sameerdesai]

Actually they were dolphins (Google cache)

Desktop power not going up much?

[Omega1045]

and are one of the reasons the clock speeds on your desktop computer have not really been going up much in recent years

This sounds silly to me since desktop power (say a $500 system - discounting monitor and keyboard) is increasing exponentially, doubling every two years compared to the price. The machine I built this spring was twice as powerful than a system I built in 2003 for the same money, but 8 times as powerful as a machine I built just 6 years ago and is about 128 times as powerful as the machine I had when I went to college in 92. And I am only considering pure clock speed, not increases in the efficiency of chips, growth of RAM and disk for the price, etc. While Moore's law concerning silicon chips will start faltering as we approach 2020, I have been nothing but impressed with how desktop performance continues to improve.

These new laser improvements, and things like molecular computing, will help us continue on after the 2020 mark with our current exponential growth.

Sorry to go off, I just got done reading The Sigularity Is Near

Re:Desktop power not going up much?

[gnuLNX]

Actually clock speeds really haven't done much in the last 2-3 years.

I think the increase in clock speed game is pretty much up. From here on out we will see multicores and other such innovative technology...but probably we won't see 10Ghz chips anytime soon...which if clockspeed was doubling every 18 months like it used to we would see it in about 3-3.5 years.

Re:Desktop power not going up much?

[robertjw]

Actually clock speeds really haven't done much in the last 2-3 years.

I'm not so sure about that. Speeds may have only increased from 3Ghz to 3.8Ghz in the last 1-2 years, but that's still a respectable 27% increase over a 1 year period. Of course you are right, the biggest increases have been with the multicore and 32-64 bit architecture changes. Clock speed may not be doubling every 18 months, but a 27% annual increase is nothing to sneeze at either.

Personally, I believe part of the reason we haven't seen a huge increase in clock speed is the lack of demand. Most people are finally satisfied with the performance their computer is giving and don't feel the need to upgrade constantly.

Re:Desktop power not going up much?

[timeOday]

I disagree, the slowed progress in PC speedups in the last couple years has been disappointing. Things really started to fall off at about 3ghz. 3ghz was released in 2002(!) over 2 1/2 years ago and we still haven't hit 4ghz, that says it all with respect to clock speed.

More efficient processors are only just closing in on 3ghz... pretty bad when the P3 (also reasonably good IPC) came out at 1GHz *5 years* ago.

Intel and AMD have clearly indicated that the good old days are over by introducing dual-core chips... nice if your workload needs that, but complicating the programming model (to multithreaded) is a concession to the physical limitations that are imposing themselves.

Re:Desktop power not going up much?

[a_ghostwheel]

Moore's Law has nothing to do with clock speed I think. If i remember correctly, it states that number of transistors on the chip will double every 18 months. Improved clock speeds are just side effect.

Re:Desktop power not going up much?

[Omega1045]

I am not referring only to pure clock cycles. I actually think the trend towards multi-core is a good thing. First, in modern computer enviros where multiple threads are running, multi-core systems should prove to be very effective. Second, multi-core systems will use less power than a single core system with the same total processing power. This is simple EE - multi-core means power consumption goes up in a linear style instead of as a square function. It does not matter how it is technically getting done, system are still increasing in FLOPS at the rate I mentioned. I applaud faster machines via this new laser tech, but my point still stands. Computers are still increasing in processing power in the exponential terms I mentioned. I just compared the system I built earlier this year (in Feb), and I can almost get a chip 50% faster for the same price now.

Re:Desktop power not going up much?

[Omega1045]

Actually clock speeds really have done just as much in the last 2-3 years, but let me explain in more detail exactly what my point is.

I only use Moore's Law as a side note here, so don't let that old argument take you away from my real point. I am talking about the PC you can buy for a fixed prices, $500 in my case. I wish my scanner was not borked, or I would scan in a chart I have sitting right in front of me proving that this trend of being able to purchase twice the machine for the same price every two years is dead on correct. I did some googles, and only came up with opinion pieces so I cannot provide a link. I am sure this data is out there somewhere, as the author is quoting a number of technical sources including Intel. The chart quotes real data through 2004. I cannot defend anything through 2005 with the data I have in front of me, but I believe, from looking at the prices earlier this year I have scribbled down for my new desktop (built in late Jan - I called that spring in my earlier post) I believe that trend has also continued through this year. But I will stand firm on 2003 & 2004, 2 of the 3 years you mention. I have hard data in front of me from a very respected author, Ray Kurzweil, proving my point.

Re:Desktop power not going up much?

[georgewilliamherbert]

Clock speed was advancing more rapidly than process shrinks; now, it's probably going to be a small factor on top of the process shrink improvements.

We probably have at least 3 more generations of process shrink before any sort of "wall", and quite possibly several beyond that. The SIA roadmap isn't mostly red until after 2011, and many of those problems are solvable in less than the 6 years we have left.

Re:Desktop power not going up much?

[RicktheBrick]

My question is this. What would you rather have? A 100Mhz 386 computer with broadband or the latest dual core computer with dial up? I know I would choose the 386 with broadband. Computers are faster but it is only the increase in speed of communications that are making them more interesting. Please give an example of a program that helps the common man that runs on today's computers that would not run on a 386. Games are entertainment and I consider them little help to the common man.

Re:Desktop power not going up much?

[budgenator]

I once paid almost $500 for a scanner, now dell gives them a way if you pay shipping and ink or vica versa, basicaly you can easily buy a printer/scanner for what you'd have paid for just shipping a few years ago.

Re:Desktop power not going up much?

[Omega1045]

You have a valid point of your own. The growth of the Internet and bandwidth both follow the same model as I discussed. I am not making an argument against the increase in bandwidth in my previous comments, so I am not sure how to take your comment. A more generalized version of the comment I was making extends to bandwidth.

To address your question about benefits to mankind, I think you will soon find that a number of huge questions will be answered via distibuted computing. Project like SETI's screensaver require no more than a dial-up connection, but they do count on the processing power of each node. Other projects have followed this model, including analysis of cancer via the huge processing power of millions of desktop machines. Personally, I would also take the 386 with broadband, but for the benefits of curing any number of diseases, I would go for the other approach. In our world, we do not need to make this choice.

Re:Desktop power not going up much?

[Anonymous Coward]

I can't believe you'd quote anything by Kurzweil! The guy is a loony.

Re:Desktop power not going up much?

[khayman80]

You're right- desktop computers have continued to get more powerful over the last few years. But the point is, the clock speeds have begun to slow their frantic increase (yes, I know that this isn't related to Moore's law). There are lots of examples of this, but I'm way too lazy to look them up. I'm pretty sure that if you plot the highest frequency attained by consumer-grade CPUs over the last 10-20 years, you'd see it rising quite fast, then starting to level off near 2000 or quickly afterwards. Recently there's been a drastic example of this- AMD stopped reporting clock speeds because they weren't increasing very fast, and Intel suffered a similar defeat recently. I can't recall the exact codename of the Intel chip, but several times in the last couple of years Intel has been forced to drastically scale back their roadplan expectations of the highest attainable clock speeds for their best chips.

The reason that this failure isn't more obvious is because computer engineers have gotten very clever and inventive lately regarding ways to increase perceived performance without actually increasing the clock speed of the chip. Things like making a chip perform more calculations in parallel, adding two CPUs to a chip, predicting memory requests more intelligently so as to make better use of the chip's cache, etc. Optimizations like this will eventually hit a point of diminishing returns, which is why so many groups are researching ways to make faster transistors or looking for entirely new computer architectures that will replace silicon (eventually) when it finally hits a hard-and-fast performance limit.

Re:Desktop power not going up much?

[ivan256]

More efficient processors are only just closing in on 3ghz...

Who cares? They're more efficient. They don't need to run at 3ghz to be faster that the old stuff. Just because the clock speed isn't there yet doesn't mean the performance hasn't gone up. Look how many times AMD has pulled ahead of Intel in performance, and they've never even shipped a 3ghz CPU. The only thing that has fallen off is the power of intel's old marketing. The only reason there's a 3ghz number to "catch up to" is that so much performance was given up to hit those timings.

IPC isn't really that good a measure of efficiency either. What kind of instruction? How much work does that single instruction do? How long did it take to get the data for that instruction? IPC numbers are always calculated with "fast" instructions that have no, or few, wait states.

If all you're comparing is clock speed, or "IPC", you're not getting a very good performance comparison.

Intel and AMD have clearly indicated that the good old days are over by introducing dual-core chips...

Sounds more to me like the good ol' days are finally here. Multi-core has been a goal for years, but was available only to those with the deepest pockets. Intel and AMD bringing multi-core to the masses doesn't mean they've run out of ideas for increasing single core performance so much as it means they've figured out how to throw a few more cores on a die in a cost effective manner. As for multi-threading (a far too simplistic way to describe what you need to do to properly process in parallel), the fastest computers have always been highly parallel. We know how to do that stuff now. It's not rocket science. End users don't typically program their machines anyway, and we could do with weeding a bunch of the crappy engineers out of the job pool.

Re:Desktop power not going up much?

[gnuLNX]

Which was basically my point. We won't have to continue increasing clock speed to continue gaining improved performance. If we could get the system bus to match the speed of the processor we (I am a scientific programmer) would notice huge performance gains.

Standford?!

[mrm677]

Ok, I usually don't gribe about spelling mistakes, but STANDFORD?!!! First we get Taco's rant about his online gaming debacles and now we get this...Slashdot has stooped to new lows today.

Digital Bandwidth?

[dada21]

A week or so ago, I mentioned decommissioning analog & digital TV broadcast spectrum to use for ore wireless data. I mentioned how fiber was just on serendipidous discovery away from massive data rates. I was shunned as "everyone knows" there are limits to light.

While this may not be THE discovery I was alluding to, it proves that the door surely isn't closed.

While science can find use in this discovery, I'm more interested in profitable consumer uses. What are the possibilities there?

Re:Digital Bandwidth?

[amliebsch]

You might want to check out this article that appeared in IEEE Spectrum magazine:

The Silicon Solution

It describes what I believe is the same breakthrough in considerable detail. The Big Deal is that lasers can now be made from standard CMOS silicon fab processes, meaning you can integrate the lasers and optoelectronics directly into the chip without needing radically new chip fab techniques. Really interesting stuff!

Re:Digital Bandwidth?

[Detritus]

The problem isn't bandwidth, it's cost, getting those high data rates on and off the fiber at a reasonable price. Wavelength division multiplexing can be used to attain insanely high data rates, if you have enough money.

The fundamental limits haven't changed.

Re:Digital Bandwidth?

[dada21]

Dead link?

10.4.5 404 Not Found
The server has not found anything matching the Request-URI. No indication is given of whether the condition is temporary or permanent.

Re:Digital Bandwidth?

[phalanx]

Just copy and paste www.spectrum.ieee.org/oct05/1915 it works

Re:Digital Bandwidth?

[xTantrum]

To make the modulator, Harris and Miller's research group had to buck the conventional wisdom that physics wouldn't allow it, says student researcher Yu-Hsuan Kuo.

I love it when we do that. always so tired of the overly intelligent who know everything about everything spouting their verbiage on the possibilities of possibilities.

Redundancy is their linguistic bottleneck

[dfn_deux]

Maybe they can get rid of "the bottle neck of bottle necks" but can they reduce the reduncancy of repeated overuse again and again also?

Have they future-proofed their technology?

[Anonymous Coward]

Is this Internet 2 and Web 2.0 compliant?
With the rise of Internet 2 especially on college campuses, I hope they have made sure that their application can be used in an Internet 2 environment. At the same time I am concerned that these 'lasers' may not be compatible with the new Web 2.0 standard that everyone has been talking about of late.

materials..

[greywire]

OK, so this is a solid-state light switch that goes really fast.

I've been wanting to know for some time if there is a material that can switch from transparent to reflective? It would need to be pretty fast (or slow, if you could also slow down the speed of light, which I have read somewhere can be done)

Re:materials..

[ParadoxDruid]

I only took 3 materials science classes in Undergrad, so this won't be a full answer, but it might get you started on the right track.

I recall that some crystalline materials exhibit very different refractive and reflective properties when put under mechanical strain. Materials that do this but with electricity are how we make accelerometers these days. So a crystal that either transmits the light or refracts it off into a random direction depending on strain may be what you're looking for. No clue what rate you could achieve, though.

Re:materials..

[greywire]

nope, that wont do it! I need something that can reflect normaly, or be transparent. Although I suppose being able to reflect normaly, or reflect at a slightly different angle, would also work (instead of being transparent). It needs to be pretty precise.

Yeah, I know about MEMS and DLP devices, but I want something solid state.

Re:materials..

[toma-kun]

Essentially what you need to do is adjust the angle of the material with respect to the light beam such that it is larger than the critical angle for the material/medium interface. This results in total internal reflection (meaning zero transmittance). Probably the best way would be to use a Peizoelectric material, which will mechanically deform under an applied voltage.

Re:materials..

[servognome]

Not sure if this is exactly what you are looking for, but acousto optic modulators uses acoustic waves to change the refractive properties to diffract the incoming light to a known specific angle. So by sending a pulse through the material the beam changes angle and you can then reflect this part of the beam back.

Re:materials..

[2.246.1010.78]

I hope you have your notification-email settings turned on - that is if you are really interested.

I think the answer pretty much depends on what you call "transparent". In the case of optical communications, often it is sufficient to just modulate (or switch) one particular wavelength (frequency). In this case, for example every (direct) semiconductor will do. The idea is as follows:

• neutral semiconductor: light with a photon energy smaller than the semiconductor bandgap will not be absorbed.
• neutral semiconductor: light with a photon energy bigger than the semiconductor bandgap will be absorbed and generate charge carriers.
• with an applied current the conduction band of the semiconductor is not empty anymore. If the electron states near the conduction band edge are filled, photons at the bandgap energy cannot cause electron transitions anymore. As a result, the light will not be absorbed and the material is transparent again.

Of course, the problem with this is the speed at which you can modulate such a structure, as the electron lifetime is the limiting factor.

Re:materials..

[greywire]

Interesting, thank you.

But if I understand what you are saying, you are talking about a semiconductor that is either transparent or absorbant, not reflective. I need something that can literaly be like a mirror in one state or like glass in the other. Alternately, if it could be reflective in one state like a normal mirror, or reflect at a different predictable direction in the other state. Although I imagine if there were such a material known they would be using it instead of MEMS in DLP devices already.

obsolete clock rate

[Anonymous Coward]

Why would the clock rate need to go up if the busses were all huge, or .... why not a tiny serial bus with a petahertz clock.

Good but not great?

[grumpyman]

OC-768 laser runs at 40 giga bits per second. Also I wonder how such modulation will behave in a WDM system... more prone to non-linear effects and such.

Re:Good but not great?

[Anonymous Coward]

yeah, and it's nice that you can modulate at 100 GHz, but if there is no laser source in Si or Si/Ge to modulate, what's the point.

Re:Good but not great?

[Anonymous Coward]

100 Ghz seems particularly bad, considering the base signal itself is approx 193.1 Thz, with the wavelengths separated by 50/100/200 Ghz spacing. Could cause some very interesting non-linearities.

What wavelength?

[Orthogonal Jones]

If I remember correctly, QCSE uses excitons to absorb light.

What is the wavelength of these excitons in SiGe? If it's significantly different than 1.3 microns - 1.5 microns, then this is a short-haul play -- like inside a box. In any case, 100 Gb/s is generally fragile stuff anyway over long distance, so it's highly unlikely that this is part of some global supercomputer, as the article suggests.

That's OK, though. This might be great stuff for optical interconnection buses.

BTW, D.A.B Miller is a big name in the field, so this is likely a big deal.

Re:What wavelength?

[kebes]

I'm reading the actual Nature article now (Vol 437|27 October 2005|doi: 10.1038/nature04204, refer here for those who have access). The structure they have built is a multilayer of Si and SiGe (10 nm Si and 16 nm SiGe, repeated ten times). You are correct that there are exciton peaks in the range of 1.3 microns to 1.5 microns. Specifically, there state:

Clear quantum confinement is seen, with strong exciton peaks that we assign to electron-to-heavy-hole (e-hh; ,0.88 eV at 0 V) and electron-to-light-hole (e-lh; ,0.91 eV at 0 V) transitions.

(0.88 eV = 1.41 microns, 0.91 eV = 1.6 microns) The impressive thing is that when they change to bias the, say, 4V, there is a huge change in absorbance across the 1.3 - 1.5 range. At 1.46 microns, the absorption goes from nearly zero to orders of magnitude larger. It's an impressive result.

Re:With great bandwidth comes great responsibility

[orderb13]

Innovate or die. Welcome to capitalism, now either get out of my way with your namby pamby feelings our I'll claw my way over your corpse to climb the corporate ladder.

Re:With great bandwidth comes great responsibility

[Anonymous Coward]

Quote: "social problems such as the rampant piracy on the internet"

That's not a social problem. That's Hollywood trying to own the culture and dictate WHAT you can see, WHEN you can see it, and HOW you see it. Screw 'em.

Great...

[Munk]

if somebody can just get this technology to Geordi, then we will have a chance against the borg.

Re:Great...

[kryten_nl]

Yes, but in the meantime our networked computers make us vunerable to attacks from the Cylons. (welcome to the 21st century)

Re:Great...

[TwoTailedFox]

They did. They're called "Tachyons".

Or it's the janitor's room, hell I don't fuckin' know.

Will lower barriers to new data providers

[James McP]

....by reducing the cost of fast switching. There's plenty of dark fiber http://en.wikipedia.org/wiki/Dark_fiber out there for anyone who can afford the hardware and this may take OC12 fiber cards from ~$6000US to a couple of hundred.

At the very least, it will make it possible for gigabit ethernet switches to use an optical brain to handle much larger total loads and likely at lower costs. (No, I don't know if this is cheaper to make but I figure the low grade parts that don't run at full speed will be sold cost-competitively to recoup losses, ala the original Celeron)

Quantum Optical Laptops

[scovetta]

maybe we'll see a 100 GHz PC in the not-too-distant future.

What are you talking about? I've had a 6.8 GHz laptop for over a month. It does quantum-optical calculations damn quick, and runs Duke Nukem Forever!

Re:Quantum Optical Laptops

[Anonymous Coward]

6.8 100 == true

Re:Quantum Optical Laptops

[systemic chaos]

I don't see why anyone would want to run Duke Nukem for more than a few minutes, much less the time span you propose.

Re:Quantum Optical Laptops

[mrselfdestrukt]

Now THAT was good slashdot humour! Mod this guy up.

Publicity

[humina]

Like most unreleased technologies, I am skeptical. Many research groups publicize the possible miracle's their technology could fix while downplaying the downside of the technology. This is done in order to get more research dollars spent to fund your research. This sounds like publicity from a research group in order to get more funding. In that respect I think it is working.

I'll stick to Journal articles to see if the technology actually works though.

Re:Publicity

[ScottyB]

Uh, it's in this week's Nature. That's a pretty important Journal.

Re:Publicity

[Red Flayer]

RTFA:"He acknowledged, however, that there is a significant gap between research results and commercial availability of devices based on scientific breakthroughs.
Other designers working in the field were also cautious about direct applications of the technology. Alex Dickenson, chief executive of Luxtera, a Carlsbad, Calif. start-up firm that announced a 10-billion bit per second optical modulator using a different silicon-based approach earlier this year, said that he believed there would significant hurdles to the commercialization of the Stanford discovery.
He cautioned that while the display was interesting from an academic perspective, the researchers had yet to prove that the effect works at the standard frequencies of light used by the telecommunications industry."

The article actually bothered checking with other experts in the field to see how close we are to commercial apps.

Re:Publicity

[Anonymous Coward]

As someone from the Stanford research group (not on this project or the Nature paper), I had to laugh at the Luxtera quote. The QCSE modulator is small, runs at low voltages, and can be made in dense 2D arrays for vertical emission off-chip. The equipment wasn't that special; hundreds or even thousands of companies or universities already have it. So the modulators could be on chips within a year, probably 2. I suspect the holdup for conventional CPUs won't be the modulator, but the fact that there aren't cheap, optically interconnected circuitboards yet. That'll take some time to ramp up.

Re:An Open Letter to Solid State and Photonics Lab

[$RANDOMLUSER]

Dude.
Your tinfoil hat slipped loose.

Does this mean...

[eno2001]

...Quake 4 will play at decent speeds on my next computer? Lord knows gaming really pushes the envelope of power.

Re:Does this mean...

[sagenumen]

Maybe, but you still won't be able to play it on Windows Vista.

Re:Does this mean...

[Sandbox Conspiracy]

If by "Quake 4" you mean "porn," then yeah. I agree.

Re:Does this mean...

[eno2001]

If I meant porn, I would have said porn. And I also would have said, "Lord knows that porn pushes the envelope of acceptable sexual behavior (if you're a conservative)". ;P I'm a kidder...

Fun fun

[jtroutman]

I can't wait to see this at the local rave...100bpm? That's nothin'!

Re:Fun fun

[hardcorey]

100bpm and 100 billion times a second. Something tells me there is a bit of a difference between the two. Anyways, raves are silly.

yep...

[Anonymous Coward]

.....Ahhh we are now closer to the lightsaber....yes... the darkside will dimish......

Re:yep...

[dogmatixpsych]

The first thing I thought of is: laser guns from Star Wars. Nothing beats a good blaster at your side kid.

This will even inprove on Star Wars - we could have some wicked fast automatic laser rifles. I think I'll take a laser gun over your lightsaber and just pick you off from a distance. The only reason lightsabers were effective in the Star Wars movies is because the Storm Troopers were about as accurate as a Dvorak opinion. Their accuracy was much worse than chance.

Re:yep...

[Anonymous Coward]

a light sabre being a really powerful and fat laser that ends artificially after just 3 feet? roughly how does this make us closer to that?

Heat dissipation?

[andyo]

It would also be interesting to know how much heat is generated by the absorbtion of the light. How does this compare to electrical units' heat?

Re:Heat dissipation?

[birge]

Heat from absorption of an optical signal (at the levels used in communications) is negligable. First, the signals are very low, measured in mW, and the absorption per meter in fiber is incredibly small. (I can't remember the actual number, but you need to go tens of miles before you even lose half your light.) As you can imagine, dissipating a few mW over a hundred miles doesn't generate any heat.

OK, I'll bite

[42Penguins]

...the bandwidth of a beowulf cluster of old Korean/Soviet sharks with frickin laser beams on their heads in a station wagon going down the highway next to Natalie Portman with hot grits.

An Open Letter to Anonymous Coward

[montreal!hahahahah]

Anonymous Coward has recently made a number of people very, very angry, including me. However, as anger serves no function in a successful rebuttal, I will simply state objectively that Anonymous Coward's brethren have an almost identical mentality, as if they all had been cloned from a single insufferable prototype. I would like to start by discussing Anonymous Coward's press releases, mainly because they scare me. The thing I'm the most frightened about is that the question that's on everyone's mind these days is, "Will the world ever be free of negligent, disgraceful doofuses like Anonymous Coward?" After days of agonized pondering and reflection, I finally came to the conclusion that Anonymous Coward has planted its worshippers everywhere. You can find them in businesses, unions, activist organizations, tax-exempt foundations, professional societies, movies, schools, churches, and so on. Not only does this subversive approach enhance Anonymous Coward's ability to destroy that which is the envy of -- and model for -- the entire civilized world but it also provides irrefutable evidence that I frequently wish to tell it that its witticisms serve no purpose other than to declare a national emergency, round up everyone who disagrees with it, and put them in concentration camps. But being a generally genteel person, however, I always bite my tongue.

Anonymous Coward's morals leave me with several unanswered questions: Why do we put up with it? And what in perdition does it think it's doing? These are difficult questions to answer, because it is always prating about how all major world powers are controlled by a covert group of "insiders". (It used to say that human beings should be appraised by the number of things and the amount of money they possess instead of by their internal value and achievements, but the evidence is too contrary, so it's given up on that score.) In particular, Anonymous Coward says that it has mystical powers of divination and prophecy. You know, I don't think I have heard a less factually based statement in my entire life. Anonymous Coward can't help it; it just loves to keep us perennially behind the eight ball. Anonymous Coward extricates itself from difficulty by intrigue, by chicanery, by dissimulation, by trimming, by an untruth, by an injustice.

Even by Anonymous Coward's own account, if you think you can escape from its harebrained indiscretions, then good-bye and good luck. To the rest of you I suggest that we must reveal the truth about Anonymous Coward's pranks. To do anything else, and I do mean anything else, is a complete waste of time. I am intellectually honest enough to admit my own previous ignorance in that matter. I only wish that Anonymous Coward had the same intellectual honesty. If one dares to criticize even a single tenet of Anonymous Coward's revenge fantasies, one is promptly condemned as egocentric, repugnant, biggety, or whatever epithet Anonymous Coward deems most appropriate, usually without much explanation. Something that I have heard repeated several times from various sources -- a sort of "tag line" for Anonymous Coward -- is, "We should go out and make empty promises. And when we're done with that, we'll all tear down everything that can possibly be regarded as a support of cultural elevation." This is not a direct quote, nor have I heard it from Anonymous Coward's lips directly, but several sources have paraphrased the content to me in near-enough ways that I feel fairly confident it actually was said. And to be honest, I have no trouble believing it.

As a matter of policy, sniffish insincere-types should not bar people from partaking in activities that cannot be monitored and controlled, but this has never stopped Anonymous Coward. The impact of Anonymous Coward's balmy, overbearing put-downs is exactly that predicted by the Book of Revelation. Evil will preside over the land. Injustice will triumph over justice, chaos over order, futility over purpose, superstition ove

Re:An Open Letter to Anonymous Coward

[Anonymous Coward]

Netcraft confirms it. Slashdot is dying

            "This site has ceased to be amusing," says one time Slashdot fanatic Morton Raven. "Something that used to be fun is now about as lively as a funeral. In the past, you could come here for sometimes informative articles, and always amusing mirth-filled comments in the discussion section. Now, anything remotely funny just gets modded down as troll, and one of the world's best venues for amateur comedy is gone."

          He continued, saying "The sometimes off-color and angry banter back and forth between trolls and the newbies who fed them was a great source of amusement, but has rolled over and died, just like Stephen King, Marlon Wayans, or J.K. Rowling and what a shame! Now, it is too dull to even hold my attention for two minutes. Slashdot will be sorely missed."

It is official. Slashdot it dead.

I was thinking more of...

[Spy der Mann]

"All your hertz are belong to us." :)

Opportunity for new "PCB" makers

[G4from128k]

This tech will mean a new opportunity for a new kind of "PCB" maker. Circuit boards with embedded optical traces will replace (or layered on to) traditional electronic circuit boards. New optical chip-to-board interconnects will also become a new, growing business. I know that people do make all-optical circuits (I've seen these at Lucent's museum in NJ), but it looks like the current tech is very expensive (etched channels in a sliced wafer).

The first company to develop a low-cost, high-quality tech for "printing" optical traces will make a mint once these interconnects become common. I'd bet that the ultimate technology will be a sandwich of resins with etched channels and vapor-deposited reflective layers, walls, corners (or high-index resin filling). For most applications, the optical interconnect can be single-layer because the non-interference on crossing beams will let two traces/channels cross each other with interference.

Inventions like this one are a great start. But until they find away to make cheap circuits to route optical connections on a board, this tech won't see widespread adoption.

Re:Opportunity for new "PCB" makers

[randyest]

That was all very interesting lay speculation, and I do wish I could take you up on your bet.

See, no one is going to "print" optical traces. Unless you consider gluing fiber to a board "printing." Fiber optic cables are cheaper than PCB by a long shot, which is why they are used for optical interconnect now, and will be in the forseeable future.

modulation

[sgt scrub]

I'm sure it is impossible but....

What I would be more interested in is "modulating" a lazer so a portion of it appears to stay suspended in the air. An example would be the stream of water from a sprinkler. It would be the biggest step towards a true 3d display.

All find and dandy

[Whyzzi]

but what kind of weapons can be made from this technology - and how long is it going to take?

Re:to all you anti-us guys that i argue with in...

[Anonymous Coward]

add another one the running tally of things this "stupid country" has done to change the face of the world.

At least that's ONE of you that's willing to admit you're changing the face of the world. Now how about joining the rest of us in trying to preserve that face while we still have a chance?

Genetic engineers wanted...

[jpellino]

...to come up with a frickin' shark that can keep up wih these new lasers.

Or use NYT Link Generator!!

[antdude]

Non registration link from the NYT Link Generator. Use it!

Re:An Open Letter to Solid State and Photonics Lab

[Anonymous Coward]

Do these long posts have some kind of coding inside them?
I mean, does someone use slashdot to store their pr0n?

Re:to all you anti-us guys that i argue with in...

[CDPatten]

lol. very good. Its not often you side makes me laugh...

Progress is exponential. News @ 11

[Saeger]

Another breakthrough; another day the exponential march of progress continues.

And it won't be long before we're building chips in three dimensions vs just two.

Its Leyland Stanford Junior University

[iambarry]

Full name : Leyland Stanford Junior University

Bah!

[DarthVain]

Thats not a real break through... a real breakthrough would be dolphins with freckin' posion dartguns attached to their heads... oh wait... nm.

moores law my hindquarters

[McNihil]

Back in 1989 I had a 50 MHz machine MC 68030 (Amiga) and if Moore law about 18 month doubling would be true we should have had 5.7 GHz machines today. Incidentaly my next upgrade is when it hits 6 GHz because other than a magnitude faster machinery it isn't worth upgrading.

Re:moores law my hindquarters

[Omega1045]

That is like comparing PPC to x86. How many FLOPS did that run compared to a high-end machine today?

Re:moores law my hindquarters

[Bastian]

Moore's law has absolutely nothing to say about whether or not we should have 5.7GHz CPUs by now. Moore's Law refers to transistor counts, not clock speed.

Re:moores law my hindquarters

[McNihil]

Transistor count and thus transmission lengths most definitely dictate how fast a CPU can work. By doing good CPU design one can match the speed quite close to the "count" BUT as you seem to be an x86 Intel brainwashed techno child it isn't supprising that you defend the historicaly poor performance of Intel processors. All this is realy nothing BUT that you get 5 mod points is a travesty and mods here should get a clue stick.

Re:moores law my hindquarters

[McNihil]

Why shouldn't we compare different processors types? All in all look at the Power 5 processor and that is where we all should be at regarding computational power. We are behind the evolution curve because the x86 cow needs to be milked to the bones.

I wonder:

[kurbchekt]

If this'll mean that CD's will be around a lot longer?

Re:Its Leyland Stanford Junior University

[Anonymous Coward]

that's "leland", dumbshit

speed * time

[Bazman]

How quick is 100 billion times per second? Well, if my calculation is correct then light, moving at 3*10^8 m/s, will, in 1/(100 billion) = 1/(10^11) seconds, have moved 3mm.

At that rate, the universe is almost stationary.

Baz

Re:Its Leyland Stanford Junior University

[iambarry]

that's "leland", dumbshit
Nice to know that Stanford graduates read my posts. Thanks for not ending the sentence with a preposition.

Sorry for the spelling error, I stand corrected.

--Barry

Great!

[Digital Vomit]

We can finally cut through the Borg's shields now!

mod parent up!

[sbma44]

This is an important point that is frequently overlooked: quantum computers will not speed up traditional computing, they will just let us solve classes of problems that are intractable, at the moment.

One Question:

[karnal]

Can loving lasers hurt?

gosh a 100 billion

[Anonymous Coward]

Wow 100 billion sounds much more impressive than 100GHz!

Need a little help here guys.

[thevil]

I've run into some sort of trouble. The stupid ray start diverging almost immediately. I've followed the instructions, but it still doesn't work. Any ideas?

Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Malum quidem nullum esse sine aliquo bono.

100 billion times per second?

[elgatozorbas]

Why write '100 billion times' per second instead of 100GHz? My cell phone 'generates an electromagnetic field that changes orientation almost 1 billion times per second'. Wow!

Re:100 billion times per second?

[Anonymous Coward]

Because Hertz are units that express cycles per second for periodic functions.

Since the modulation of the laser is not periodic, but rather is modularly controlled, it's not a periodic function, and so can't be expressed in Hertz.

Re:100 billion times per second?

[elgatozorbas]

Not quite. You are only interested in the bandwidth (expressed in Hz). If you can turn that light on and off at 100 times per second you can also do it at 10 times per second (but maybe not at 1000). In other words: you can modulate it with a bandwidth of 100Hz (*), so it makes sense to call this '100Hz' and not '100 times per second'. IMHO no engineer would ever do this because it takes some time to figure out how many GHz '100 billion times per second' represents (especially for non native speakers).

(*) I know that turning off/on _instantly_ requires a higher bandwidth because of the steep flanks but that is not the issue here.

Overstated results

[PhysicsPhil]

Somewhere between the lab and the press release things got overstated. Since my PhD is in silicon-based optoelectronics, I am familiar with this kind of work. A few thoughts crossed my mind after reading the paper.

What these guys have found is a physical effect that possibly could lead to fast modulation of light. Neglected in the press release are a few fairly important issues:

• They haven't demonstrated any time-resolved optical effect, and are inferring it strictly from what might be possible. I have no doubt they can modulate, but the operational speeds are still guesstimates.
• The effect that was demonstrated is not within the 1550 nm wavelength window used for telecom traffic. Their current work shows the effect right in the middle of an H2O absorption peak. Can the effect be shifted? Probably, but these sorts of things are always more work than expected.
• From a practical standpoint, other Quantum Confined Stark Effect devices often show a strong sensitivity to the polarization of the input light. Ensuring a known input polarization is a major problem right now in optoelectronics. Lord knows it was (still is, actually) a major hassle in my research
• This device is not quite as CMOS compatible as might be hoped. Building strained germanium quantum wells on a silicon substrate requires depositing atoms layer by layer, and is a slow process. Process throughput will no doubt be an issue.

All that being said, this is still very exciting. It is a new physical effect demonstrated in a silicon-based material, and a physical effect that has been used elsewhere to do useful things. Hopefully a real modulation device will come along shortly.

Re:Overstated results

[birge]

Your points all seem valid, except for the polarization one. I think most modulators are polarization sensitive. You just polarize the input and accept the losses. In the case of the modulator attached to the laser (the usual case) the laser output is polarized.

Don't worry...

[danratherfoe]

Don't worry ... if they try to use it as a weapon, we can always recruit Val Kilmer to reprogram the laser to destroy its designers' houses.

A Dislexic Would Say

[Ranger]

Standford? Berkly? I don't get it. What's so funny? And here I just learnt how to spell Lebalebanon correctly.

Re:A Dislexic Would Say

[Fulcrum of Evil]

"There are more birth defects among Born Again Christians than any other religious group."

That's not a very nice way to talk about their children.

Re:A Dislexic Would Say

[Anonymous Coward]

it's false too.

closer to data's positronic net...

[Anonymous Coward]

here comes AI...

...another small step for machinekind.

[Seraphnote]

I hadn't seen any comments about Skynet yet!

So does this bring us one step closer to AI capable hardware running the Windows "Hasta La Vista Baby" version?

Warning stickers...

[TiggertheMad]

Will there be a warning sticker on the chip that says:

ABSOLUTELY, DO NOT OPEN THE CHIP CASING AND LOOK AT THE FIGGEN' LASER, STUPID. IT WILL MELT YOUR FACE!

Because, I'm thinking that will be needed for something that modulates (holds pinky to mouth) 100 BEEEELLION times per second!

Re:With great bandwidth comes great responsibility

[TerranFury]

Climbing the corporate ladder != Innovation

Innovation?! C'mon! This is a culture in which people really do use words like "synergy" and "value-added" with straight faces! I know; I've worked with them!

Each time I've worked in a corporate environment, I've been thoroughly appauled. People don't pursue good ideas! Rather, they make sure that they have all the right "check marks" on their "report cards." At the last place I worked, there were so many half-assed useless projects lying around -- wastes of time and money -- which could have been made useful if the resources had been put into them to do them right. But they weren't, because that's not how the incentive structure works.

When the end of the quarter comes around, you're faced with a choice: Have I "met my goals" (your immediate supervisor will be inclined to say that you did, because it'll make him look good), and pick up a fat bonus -- or do you finish the job right?. Of course you choose option 1; you play the "incentive structure" for all it's worth.

You make the right moves. You cozy up to the right people. You do everything you can to look good. You do not investigate great new ideas.

Political scientists speak of "collective action problems." The corporation is a legal construct, and the laws that govern it seem tailor-made to create collective action problems. The individual incentives that corporatism puts in place spur individual actions which do not sum to positive collective action. That is, each worker puts the right checkmarks on his report card, but the company does not pursue goals - like investing in new technology - on which its future ultimately depends.

It's because of the "incentive horizon." People pursue goals "within their horizons." Investment culture, and legal obligations to shareholders, dictate that the incentive horizon is approximately three months long. Why don't we have decent broadband in the US? Because infrastructure takes time and has delayed returns. Successful cultures emphasize the importance of 'delayed returns,' but corporatism as it is currently practiced does not. There's a famous explanation in political science for why hereditary monarchy is rationally preferable for a people than is a series of dictatorships by unrelated people: The monarch has a larger "incentive horizon," and so will seek to build a country that will serve him and his decendants. He will tax at the maximum level which does not significantly harm economic growth, because, integrated over time, this represents his largest possible profit. The despot, in contrast, has a shorter incentive horizon, and so it is not rational for him to pursue delayed returns: He taxes everything immediately, seizing farms and industrial equipment. His actions mean that soon the public will not be able to generate new tax income to tax, but, in the short time-span in which he is operating, that is entirely rational. The problem is that modern corporatism creates this second incentive structure. Other countries, like South Korea and Japan, have succeeded in developing good broadband because they have succeeded in using government regulation to effectively change the incentive structure for corporations. The incentive horizon is longer for them. Probably still not optimally large, but longer.

So what is an innovator to do? Certainly don't get caught up in the mess that is corporate culture. Me, I'm seriously thinking about a PhD and research. I've been nothing but impressed with academic scientists.

Do they run... whatever?

[Anonymous Coward]

Wake me up when I can run GCC on that iron... or whatever it is.

Yippee!! . . .oh, dang

[Shrike9]

I thought they'd finally gotten to light sabres . . . Now I have to go back and find my popgun.

Registration free

[Anonymous Coward]

Here /. editors ought to automatically change the links.

No more Gigabit Ethernet

[jgaynor]

Bring on the 93.1322575 gigabit/second ethernet!

This has nothing to do with CPU clock speeds..

[StikyPad]

"Those large numbers could get rid of the bottlenecks of wiring, bottlenecks that are quite evident today and are one of the reasons the clock speeds on your desktop computer have not really been going up much in recent years."

I'm pretty sure the wiring "bottleneck" has, uh, absolutely nothing to do with why clock speeds haven't been going up. CPUs can run at whatever speed they like, independent of the bus. (Well.. an arbitrary multiplier of the bus; not independent strictly speaking). The problem is that they start turning into space heaters as their frequency increases, not because of anything to do with the bus speed.

Now bus speeds definately could benefit from using fiber connections, but it doesn't really matter what speed your bus runs at if the devices can't keep up. Memory is much more of a bottleneck on the FSB, and the local bus just got an upgrade with PCIX, so I'm really not sure what bottleneck the author is referring to.

Desktop Clock Speeds

[LostBurner]

"bottlenecks ... are one of the reasons the clock speeds on your desktop computer have not really been going up much in recent years."

Funny, I could have sworn the reason my clock speeds haven't been going up is that I've been broke! That and I haven't done any overclocking. If their laser tech developments will make my computer faster, I'm all for it!

Imagine larger numbers!

[mpfife]

Standford engineers have discovered a method to modulate a beam of laser light up to 100 billion times a second. From the article: "The vision here is that, with the much stronger physics, we can imagine large numbers - Those large numbers could get rid of the bottlenecks"

Image new numbers different than 100 billion?? Like 100,000 million! or 100,000,000 thousands! Awesome! Stanford's l337 rox0rs more than MIT's 'large numbers'!

KARMA WHORE

[Anonymous Coward]

Not that being a karma whore is different than any other TMM post.

I don't know what's worse - that this guy has orgasms over being a karma whore or that there are apparently legions of idiot mods who have orgasms over modding this idiot up every f**king time. Just listen for the collective, global groan and you can tell when TMM has posted something.

Go on, mod me into oblivion. Accuse me of being jealous. (My actual account has had excellent karma for over a year.) Whatever. That doesn't make the fact that he's a karma whore any less true.

bah

[Anonymous Coward]

The problem with computers is if I may, the *&^&*^ING CLOCK! Who's brain dead idea is it to keep using a *&(*ING clock! Make a computer clockless and what happens, it doesn't feel like a giant nuclear reactor reaching criticle, any hardware optizations make the whole thing better etc.

Quantum-Confined Stark Effect

[ScrewMaster]

That must be what kept that one Farscape character's head from exploding.

How does laid fiber scale?

[G4from128k]

See, no one is going to "print" optical traces. Unless you consider gluing fiber to a board "printing." Fiber optic cables are cheaper than PCB by a long shot, which is why they are used for optical interconnect now, and will be in the forseeable future

Yes, fiber is cheap for point-to-point routings, but I doubt it scales well. What happens when a motherboard becomes 100% optical interconnect -- with virtually every chip and attached device using optics to communicate? Optical connections would run from the CPU (maybe each core of the CPU) to memory controller, cache, main memory banks (perhaps one fiber per optically connected RAM card), I/O controllers, mass storage devices, I/O ports, expansion bus slots (again, one fiber per slot), etc. A single motherboard might have easily have dozens of optical interconnections. I'm suggesting that printing the channels becomes cheaper than mechanically laying dozens of fibers.

With discrete fibers, doubling the number of interconnections doubles the cost because each added fiber must be laid into place during the manufacturing phase. With printed optical traces, the cost is nearly independent of the number of interconnections -- just add another trace in the mask during the design phase. It's the same logic that means that people use PCBs and not wire-wrap boards for mass-produced electronics. If N is the number of interconnections, then printing is cheaper than routing fiber for some value of N.

Perhaps you and I have different expectations for both the cross-over value of N (where printing makes sense) and the likely future value of N for intensively optical motherboards.

Re:How does laid fiber scale?

[randyest]

Yes, fiber is cheap for point-to-point routings, but I doubt it scales well. What happens when a motherboard becomes 100% optical interconnect -- with virtually every chip and attached device using optics to communicate? Optical connections would run from the CPU (maybe each core of the CPU) to memory controller, cache, main memory banks (perhaps one fiber per optically connected RAM card), I/O controllers, mass storage devices, I/O ports, expansion bus slots (again, one fiber per slot), etc. A single motherboard might have easily have dozens of optical interconnections.

Your doubts are unfounded. Fiber scales better than PCB traces because it has a higher bandwidth density (more bandwidth for a given amount of space occupied) and it's easy to cluster tighly without fear of crosstalk and signal integrity issues that plague PCBs. If it can be done on a PCB, it can be done with fiber, more easily and less expensively (from an interconnect standpoint -- the optical drivers are still way expensive compared to simple electrical drivers, but this article is about how that might be changing with on-chip optical drivers.)

I'm suggesting that printing the channels becomes cheaper than mechanically laying dozens of fibers.

And I'm telling you that you're wrong.

With discrete fibers, doubling the number of interconnections doubles the cost because each added fiber must be laid into place during the manufacturing phase.

Have you ever done any PCB design? Because "Just another trace" is kinda funny to someone who has. See, there are design rules and signal integrity issues that must be considered in PCB design. You can't just cram in as many wires as you want -- traces need a minimum spacing that depends on their width and frequency of operation. Adding more traces eventually requires more PCB layers so they can all be routed, and every new layer adds a lot of expense.

With printed optical traces, the cost is nearly independent of the number of interconnections -- just add another trace in the mask during the design phase.

First, there is no such thing as a "printed optical trace". You're making it up without even realizing what a huge hassle such a thing would be to work with (do you think light likes to travel down the 90-degree and 45-degree angles common on PCB layouts, or were you going to add a mirror at every corner? Maybe use some circular design rules and bezier all the routes? lol!) Second, even in the case of traditional traces on PCBs, the cost is directly related to the number of layers and design rule flexibility of the PCB, which are both dictated by the number of interconnections required. So, the idea that "the cost is nearly independent of the number of interconnections" is pretty naieve.

It's the same logic that means that people use PCBs and not wire-wrap boards for mass-produced electronics. If N is the number of interconnections, then printing is cheaper than routing fiber for some value of N.

There's more reason to shun wire-wrapping than simply cost of manufacturing/labor intensiveness. Wrapped wires are inductors, and any signal rate above a few MHz is going to go to crap immediately in a wire-wrapped circuit. I get the impression that you think fiber connections are manually made, like wire wrap. This is not the case. It's as automatic as mounting integrated circuits.

Perhaps you and I have different expectations for both the cross-over value of N (where printing makes sense) and the likely future value of N for intensively optical motherboards.

Well, yeah, since the crossover you're seeking doesn't exist for fiber vs. PCB trace. Excepting transimitter/receiver cost, fiber is always cheaper than PCB (and faster and more noise immune) for any N.

But only for transport speed

[Solus Stultus]

sadly this won't help raise the speed of your CPU, you still can't build a transistor with purely optical materials.

What really is moving - the electrons or the hole?

[SonicSpike]

My high school physics teachjer would go on rants all the time about what actually moved down the line, the "hole" or the electrons themselves.

His name was Dr. Troy Soos and worked for Los Alamos for a while. Then he decided to write baseball murder mysteries (apparently that makes more money than being a research scientiest).

Here is a link to his books:
http://www.amazon.com/exec/obidos/tg/detail/-/1575 664550/qid=1130379198/sr=1-2/ref=sr_1_2/103-264399 1-8895002?v=glance&s=books

Oddly enough he still teaches high school in Central Florida "just for fun".

Re:Speed of light vs. speed of electrons in wire

[birge]

When sending signals electronically, you're not really moving electrons. More like shaking them. Electrons don't actually travel through the wire in a net current. The way signals are propagated in an electrical transmission line is actually light, with some motion of electrons in local currents. It's just a difference in frequency, that's all. I think RF waves (the electrical signals you're talking about) generally travel around half the speed of light in vacuum. On the other hand, that's not all that much slower than the speed of light in a fiber. The whole point of transmitting information optically has nothing to do with the whole "speed of light" deal. It simply has to do with how much information can be encoded in an optical waveform, and perhaps more importantly, how much more bandwidth is available in optical media as opposed to RF media. That's really the big difference: the bandwidth of an optical fiber is huge, on the order of hundreds of THz as opposed to tens of GHz for the best RF transmission lines.

Not CPU speed.... BUS speed....

[Teancum]

This has nothing to do with CPU speed, but rather the bus speed that connects the CPU to other components. The last "major" upgrade on a common bus was increasing PCI frequency to 66 MHz from 33 MHz... and that took 10 years to accomplish, not the 18 month doubling of "Moore's Law" that everybody talks about. Even PCI-X is an "older technology" by many standards. And think about that too: If the bandwidth going to a peripheral card is limited by the fundimental bus architechture, why should peripheral designers try to push the bandwidth of their products as well? Gigabit ethernet is nice, but as a practical matter you can't get that gigabit bandwidth of data across the PCI bus to the CPU. PCI-X at full bandwidth is 66 MHz @ 64 bit parallell = 4 Gigabits/s (roughly). Keep in mind that you also have to include protocol overhead, bus interrupts, and other factors that also substantially reduce the actual throughput, not to mention legacy support for the 33 MHz devices as well.

An optical "bus" would indeed be a huge improvement, and something in the realm of where you can put TTL logic to optical converters and have them mean something. A 100 Gigabit/s bus would mean a 25 x improvement in bus throughput that would paint the bullseye on system speed improvement to memory designers, just as you have suggested. And memory designers are also hampered by bus bandwidth as well (memory bus in this case) because refinement of the memory chips designs themselves are trivial compared to CPU improvements.