Since microbes reproduce much more quickly, everyone can settle this debate using them. Over time, strains of bacteria become resistant to antibiotics. This is pure Darwinism. So it should be possible to prove evolution using evidence generated after a challenge is offered.
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.
http://ieeexplore.ieee.org/search/wrapper.jsp?ar nu mber=775342
Ionization of the argon atoms apparently imparts a phase shift to the ionizing pulse. Therefore, the argon gas acts as a dispersive medium.
Farhad and Hosain Hakimi demonstrated the same idea using short optical pulses in dispersive optical fibers. The fiber acts as a lens, generating the "far field" response -- i.e., the Fourier transform. They also demonstrated a practical use for their temporal gratings -- continuously-variable true time delay of optical pulses.
Among the useful applications of true time delay is optical logic and phased-array radar.
"ZFS, the dynamic new file system in Sun's Solaris 10 Operating System (Solaris OS), will make you forget everything you thought you knew about file systems."
I thought silicon had higher carrier mobility than diamond (carbon), and that SiC would have a mobility in between.
Therefore, yeah, it would take more heat, but it would be slower. As it's already been said, this might be OK for a space app or power app. But I doubt it makes sense for computers.
Also, combo semiconductors like this (or, say, GaAs) are subject to the antisite defect -- which they're saying is lower than ever in this case. However, for silicon, it doesn't exist at all.
Given that many of the Linux programmers who contributed to the core code aren't Americans, it is a bit disingenuous to say that Linux promotes the transfer of American intellectual property to parasitic non-American companies.
Thanks for the information, especially for pointing out that autocompensation doesn't limit the range.
As far as the SNR and BER, it seems that we're thermal-noise limited here, so we have to cool the receiver quite a bit. Bummer.
Even if the thermal noise can be subtracted, the photons incident on the receiver are presumably still governed by Poisson statistics, so it seems a little hard to have both one photon per bit and good BER
OK, I am not a believer in quantum cryptography for one big reason -- fiber loss. Someone please enlighten me if I'm wrong.
The loss of standard single-mode fiber is about 0.1-0.2 dB/km. Therefore, unless the distance is short (as in this demonstration), the transmitter must send multiple photons to ensure a decent probability of providing the receiver with one photon.
For example, if the span is 100 km long (20 dB loss), then on average only 1 out of every 100 transmitted photons will reach the receiver.
The situation is worse for autocompensating quantum-crypto systems (e.g., polarization-based encoding), because the photons must survive a round trip through the fiber.
Therefore, the relatively high power at the transmitter implies that an attacker can tap into the fiber near the transmitter, subtract (on average) only 1 photon, and remain undetected by the receiver.
Furthermore, typical optical amplifiers add noise (3 dB noise figure for your standard erbium-doped amplifier). The added noise photons would screw up the link, so amplifiers are out.
In the end, it seems to me that quantum crypto is good for table-top demos, and maybe for short jaunts across a metro area. But it is NOT absolutely perfect, at which point computationally difficult encryption is more attractive.
Sun cannot compete with Linux/AMD64. Hopefully Microsoft did not buy IP ownership rights for Java, because Sun ought to open-source it before the company expires.
I can bench press a lot, me and my near-future self will bench press each other. Then we'll get around the near-future earth.
Of course, we would then have to get around Bizarro Earth. Personally, I'm assuming my Bizarro self is a terrific dancer and extremely wealthy, so I plan on crashing on his couch.
I'm an American-born ethnic Indian, and I've been there many times.
It *is* a big difficulty to live there if you weren't born there. Most American-born Indians don't like it.
So if all you white people are repulsed by the idea of moving there, thank God for his mercy.
I remember living in San Diego and seeing Orange County engineers diffuse in. These people started demanding the removal of evolution from the teaching curriculum, and in general started throwing their weight around.
The average Indian wants your money, not you. Please keep your white superiority and proselytizing here in the North American Wal-Marts, where it belongs.
This is a 2 Gb/s modulator, whereas III-V semiconductor modulators above 40 Gb/s are commericially available.
A modulator by itself is nothing new, and not the whole story. You need optical waveguides with bending radii much smaller than currently available for routing, and optical logic gates which are an even worse problem.
The article doesn't describe the technology -- is it electroabsorption? Mach-Zehnder?
Nevertheless, a small and fast silicon modulator has obvious commercial value, even if it isn't the greatest thing since sliced bread.
As an American-born ethnic Bangalorean, this trend gives me mixed emotions. America been BERRY, BERRY good to me. On the other hand, it's nice to see my cousins not be poor. In fact, they act like they won the lottery.
Whereas India may be great for R&D, it is a one-trick pony for now -- Desk Jobs R Us. They have poor power, roads, water, and government. So their mechanical engineering is still stuck in the 1950's. America should switch over to that field -- robotics, materials research, etc. You'll have a much harder time outsourcing those.
The endgame of the IT revolution is just around the corner. Stop talking about how to get it back.
On the other hand, the U.S. actually has a tendency to fight wars quite often. It has a need for new materials and robotics. The current military has maxed out its use of IT for non-hierarchical combat, but that's still only good for surgical strikes. Once you get into occupational mode, the army reverts back to 1970's Vietnam. Reducing body bags in that mode requires new technologies.
So, in a way, the war in Iraq and the outsourcing trend are the perfect storm -- universities should be getting more research dollars to crank out relevant technologies for our soldiers in the field.
As someone already mentioned, a good technical demo, but some distance from usability...
Optical computing of this kind has been around for at least 11 years. I know, since I was working on it for part of my Ph.D. thesis (disclaimer -- I am an optical engineer). This stuff was big at UCSD. The primary funders are military, since they're always DSP-limited in their image recognition, detection algs., etc.
Some of the difficulties have been thermal/vibrational stability and contrast ratio of the spatial light modulators. I see they're using GaAs MQW modulators in reflection mode, so I would guess the contrast ration is about 20 dB (any better guesses?).
It looks like the output intensities are summed on the photodetectors, so there must be an array of A/D converters at the back end. This brings into question the signal to noise of the optical sources --> detectors links.
All in all, I'd say well done. But this is not (and is not intended to be) a general-purpose computer.
I'm writing this at home on one of my two Linux/Alphas.
My first contact with one was in 1997, when I was working in Lincoln Laboratory. I bought two (for $30K!) to do a hero experiment. Put Linux on them, played with TCP parameters, and got a sustained 1 Gb/s TCP/IP session between them over an 850-km optical link. Back then, it was a world record. We tried it with a Sun server. Couldn't get the 1 Gb/s. Ditto with Intels.
Six years later, that kind of performance probably wouldn't cost a thousand dollars. But to see it then was breathtaking.
I've gone on to work with many DEC engineers. They are some of the brightest people you'll ever meet. But I've heard that their marketing sucked donkey dick. If you once worked in DEC marketing, you would NEVER put that on your resume. Pricing a computer at 15 times the next competitor is insane, no matter how good it is. That's no way to own the market.
So whether I should or not, I blame the management at DEC for sinking what was a true technical achievement.
The article mentions DWDM systems with 100 Gb/s per wavelength. That's bogus.
I am an optical engineer at a 40 Gb/s startup. The jump from 10 Gb/s to 40 Gb/s is huge. Many signal degradations (chromatic dispersion, polarization mode dispersion, nonlinearity,...) become a LOT worse when you jump from 10 to 40 Gb/s. The jump to 100 Gb/s would incur an even greater penalty.
Compensating for chromatic dispersion, PMD, et. al. requires optical components which DO NOT follow Moore's law. These components are handmade specialty devices.
While a business case can be made for 40 Gb/s, the jump to 100 Gb/s is commercially pointless. If you are building a DWDM system anyway, just spread the same data across more 10 Gb/s channels.
What the hell is "Directions", anyway? It sounds like sci-fi fluff meant to entice VC's.
Diamonds Won't Replace Silicon
on
The Diamond Age
·
· Score: 2, Insightful
Diamond semiconductors have already been produced by several countries -- South Africa, Israel, and the former Soviet Union, among others.
The good things about diamond semiconductor are its thermal conductivity and high bandgap. The high bandgap especially makes it good for satellite applications, where radiation hardness is needed.
However, higher-bandgap material has lower carrier mobility, which translates into slower transistors.
So, yeah, diamond may be more heat-tolerant than silicon. But it would have to be -- its gate voltages would be higher. In any case, don't expect to see any GHz-class chips made in pure diamond anytime soon.
Slashdot Mirror
Since microbes reproduce much more quickly, everyone can settle this debate using them. Over time, strains of bacteria become resistant to antibiotics. This is pure Darwinism. So it should be possible to prove evolution using evidence generated after a challenge is offered.
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.
http://ieeexplore.ieee.org/search/wrapper.jsp?a
Ionization of the argon atoms apparently imparts a phase shift to the ionizing pulse. Therefore, the argon gas acts as a dispersive medium.
Farhad and Hosain Hakimi demonstrated the same idea using short optical pulses in dispersive optical fibers. The fiber acts as a lens, generating the "far field" response -- i.e., the Fourier transform. They also demonstrated a practical use for their temporal gratings -- continuously-variable true time delay of optical pulses.
Among the useful applications of true time delay is optical logic and phased-array radar.
Apology accepted, if not successfully parsed. And by the way, I doubt this is an American. But I'm just judging the "Engrish"...
Don't worry. This dothead wouldn't piss down your throat if your heart was on fire.
For once, I'd like an OLD superman, who'd rather put his feet up and watch bowling.
And maybe for Lois Lane, we could have Courtney Love with a dye job. Or Margot Kidder. What's the difference, anyway?
"ZFS, the dynamic new file system in Sun's Solaris 10 Operating System (Solaris OS), will make you forget everything you thought you knew about file systems."
What is a....."file-system"?
Excuse me, but Liberians can't be President of the United States.
If he's running for the Liberian presidency, that's a different matter. In that case, I wish him all the best.
In such an event, what is his plan for restoring order to Liberia?
I thought silicon had higher carrier mobility than diamond (carbon), and that SiC would have a mobility in between.
Therefore, yeah, it would take more heat, but it would be slower. As it's already been said, this might be OK for a space app or power app. But I doubt it makes sense for computers.
Also, combo semiconductors like this (or, say, GaAs) are subject to the antisite defect -- which they're saying is lower than ever in this case. However, for silicon, it doesn't exist at all.
Given that many of the Linux programmers who contributed to the core code aren't Americans, it is a bit disingenuous to say that Linux promotes the transfer of American intellectual property to parasitic non-American companies.
Thanks for the information, especially for pointing out that autocompensation doesn't limit the range.
As far as the SNR and BER, it seems that we're thermal-noise limited here, so we have to cool the receiver quite a bit. Bummer.
Even if the thermal noise can be subtracted, the photons incident on the receiver are presumably still governed by Poisson statistics, so it seems a little hard to have both one photon per bit and good BER
OK, I am not a believer in quantum cryptography for one big reason -- fiber loss. Someone please enlighten me if I'm wrong.
The loss of standard single-mode fiber is about 0.1-0.2 dB/km. Therefore, unless the distance is short (as in this demonstration), the transmitter must send multiple photons to ensure a decent probability of providing the receiver with one photon.
For example, if the span is 100 km long (20 dB loss), then on average only 1 out of every 100 transmitted photons will reach the receiver.
The situation is worse for autocompensating quantum-crypto systems (e.g., polarization-based encoding), because the photons must survive a round trip through the fiber.
Therefore, the relatively high power at the transmitter implies that an attacker can tap into the fiber near the transmitter, subtract (on average) only 1 photon, and remain undetected by the receiver.
Furthermore, typical optical amplifiers add noise (3 dB noise figure for your standard erbium-doped amplifier). The added noise photons would screw up the link, so amplifiers are out.
In the end, it seems to me that quantum crypto is good for table-top demos, and maybe for short jaunts across a metro area. But it is NOT absolutely perfect, at which point computationally difficult encryption is more attractive.
Sun cannot compete with Linux/AMD64. Hopefully Microsoft did not buy IP ownership rights for Java, because Sun ought to open-source it before the company expires.
I can bench press a lot, me and my near-future self will bench press each other. Then we'll get around the near-future earth.
Of course, we would then have to get around Bizarro Earth. Personally, I'm assuming my Bizarro self is a terrific dancer and extremely wealthy, so I plan on crashing on his couch.
I'm an American-born ethnic Indian, and I've been there many times.
It *is* a big difficulty to live there if you weren't born there. Most American-born Indians don't like it.
So if all you white people are repulsed by the idea of moving there, thank God for his mercy.
I remember living in San Diego and seeing Orange County engineers diffuse in. These people started demanding the removal of evolution from the teaching curriculum, and in general started throwing their weight around.
The average Indian wants your money, not you. Please keep your white superiority and proselytizing here in the North American Wal-Marts, where it belongs.
We had enough of you people last century.
Disclaimer: I am a Ph.D. in fiber optic physics
This is a 2 Gb/s modulator, whereas III-V semiconductor modulators above 40 Gb/s are commericially available.
A modulator by itself is nothing new, and not the whole story. You need optical waveguides with bending radii much smaller than currently available for routing, and optical logic gates which are an even worse problem.
The article doesn't describe the technology -- is it electroabsorption? Mach-Zehnder?
Nevertheless, a small and fast silicon modulator has obvious commercial value, even if it isn't the greatest thing since sliced bread.
He said it, so it's done! I can't wait to see Bush on Mars running a surprise 4th of July barbecue for the troops!
Whatever it takes to bring democracy to the oppressed microbes there...
As an American-born ethnic Bangalorean, this trend gives me mixed emotions. America been BERRY, BERRY good to me. On the other hand, it's nice to see my cousins not be poor. In fact, they act like they won the lottery.
Whereas India may be great for R&D, it is a one-trick pony for now -- Desk Jobs R Us. They have poor power, roads, water, and government. So their mechanical engineering is still stuck in the 1950's. America should switch over to that field -- robotics, materials research, etc. You'll have a much harder time outsourcing those.
The endgame of the IT revolution is just around the corner. Stop talking about how to get it back.
On the other hand, the U.S. actually has a tendency to fight wars quite often. It has a need for new materials and robotics. The current military has maxed out its use of IT for non-hierarchical combat, but that's still only good for surgical strikes. Once you get into occupational mode, the army reverts back to 1970's Vietnam. Reducing body bags in that mode requires new technologies.
So, in a way, the war in Iraq and the outsourcing trend are the perfect storm -- universities should be getting more research dollars to crank out relevant technologies for our soldiers in the field.
As someone already mentioned, a good technical demo, but some distance from usability...
Optical computing of this kind has been around for at least 11 years. I know, since I was working on it for part of my Ph.D. thesis (disclaimer -- I am an optical engineer). This stuff was big at UCSD. The primary funders are military, since they're always DSP-limited in their image recognition, detection algs., etc.
Some of the difficulties have been thermal/vibrational stability and contrast ratio of the spatial light modulators. I see they're using GaAs MQW modulators in reflection mode, so I would guess the contrast ration is about 20 dB (any better guesses?).
It looks like the output intensities are summed on the photodetectors, so there must be an array of A/D converters at the back end. This brings into question the signal to noise of the optical sources --> detectors links.
All in all, I'd say well done. But this is not (and is not intended to be) a general-purpose computer.
I'm writing this at home on one of my two Linux/Alphas.
My first contact with one was in 1997, when I was working in Lincoln Laboratory. I bought two (for $30K!) to do a hero experiment. Put Linux on them, played with TCP parameters, and got a sustained 1 Gb/s TCP/IP session between them over an 850-km optical link. Back then, it was a world record. We tried it with a Sun server. Couldn't get the 1 Gb/s. Ditto with Intels.
Six years later, that kind of performance probably wouldn't cost a thousand dollars. But to see it then was breathtaking.
I've gone on to work with many DEC engineers. They are some of the brightest people you'll ever meet. But I've heard that their marketing sucked donkey dick. If you once worked in DEC marketing, you would NEVER put that on your resume. Pricing a computer at 15 times the next competitor is insane, no matter how good it is. That's no way to own the market.
So whether I should or not, I blame the management at DEC for sinking what was a true technical achievement.
The article mentions DWDM systems with 100 Gb/s per wavelength. That's bogus.
I am an optical engineer at a 40 Gb/s startup. The jump from 10 Gb/s to 40 Gb/s is huge. Many signal degradations (chromatic dispersion, polarization mode dispersion, nonlinearity,
Compensating for chromatic dispersion, PMD, et. al. requires optical components which DO NOT follow Moore's law. These components are handmade specialty devices.
While a business case can be made for 40 Gb/s, the jump to 100 Gb/s is commercially pointless. If you are building a DWDM system anyway, just spread the same data across more 10 Gb/s channels.
What the hell is "Directions", anyway? It sounds like sci-fi fluff meant to entice VC's.
Diamond semiconductors have already been produced by several countries -- South Africa, Israel, and the former Soviet Union, among others.
The good things about diamond semiconductor are its thermal conductivity and high bandgap. The high bandgap especially makes it good for satellite applications, where radiation hardness is needed.
However, higher-bandgap material has lower carrier mobility, which translates into slower transistors.
So, yeah, diamond may be more heat-tolerant than silicon. But it would have to be -- its gate voltages would be higher. In any case, don't expect to see any GHz-class chips made in pure diamond anytime soon.
Or motorized unicycle. I would **LOVE** to see an aging baby-boomer on one of these, especially in San Francisco.
I wouldn't mind paying for the pyramid scheme of Social Security if I could get my reality entertainment that way.
Oh, hey -- that brings something to mind. Can we at least get rid of Social Security now that the workforce is overseas?
We have about 145 Linux machines here. Windows desktops of laid-off employees are re-formatted and added to the penguin cluster.
But those aren't desktop machines. Most of the 60 or so desktops are Windows.
I have 6 Linux desktops scattered around the lab for lab control and display. And the one in my cube is, of course, Linux.
Just easier in an engineering firm, I guess.
How 'bout some classes on... 1) Distributed Scheduler Design 2) SMP, Spinlocks, and Cache Coherency 3) Rules-Based Simulator Design 4) Debugging Multi-threaded Assembly Language