Slashdot Mirror
Optical Computer Made From Frozen Light
neutron_p writes "Scientists at Harvard University have shown how ultra-cold atoms can be used to freeze and control light to form the "core" - or central processing unit - of an optical computer. Optical computers would transport information ten times faster than traditional electronic devices, smashing the intrinsic speed limit of silicon technology. This new research could be a major breakthrough in the quest to create super-fast computers that use light instead of electrons to process information. Professor Lene Hau is one of the world's foremost authorities on "slow light". Her research group became famous for slowing down light, which normally travels at 186,000 miles per second, to less than the speed of a bicycle."
I am not sure what you meant by this. Modern photolithography (used in production) has optics which works well at the 193nm wavelength. EUV which is lot more complicated has optics which works all the way to 13nm wavelength.
The speeds here are limited by the gate speeds of the electronics, just like normal computers.
I think you meant interconnect delay and not switching speed of a transistor. State of the art and next generation transistors can switch in a fraction of a picosecond. On the other hand interconnects don't scale well and are the bottleneck.
Optical interconnects can break even for clock distribution were skew & crosstalk are important and the network has lot of capacitive load. That, in my opinion, will be the first place where optics will enter into microprocessors.
It's not the same.
The c in E=mc^2 (or E^2 = M^2c^4 + p^2c^2) refers to an intrinsic property of spacetime. Bose Einstein Condensates and so on don't really alter that. One way to think about it is to stop with the 'slowing down light thing', and instead conceive it as the BEC swallowing up photons for a while, storing the information, and then reconstructing a new photon which is exactly identical at the end. This is pretty much the same, because in QM, you can't really track anything exactly, and you definitely can't distinguish between objects with the same properties.
From Wikipedia: 'In a sense, any light travelling through a medium other than a vacuum travels below c as a result of refraction. However, certain materials have an exceptionally high refractive index: in particular, the optical density of a Bose-Einstein condensate can be very high. In 1999, a team of scientists led by Lene Hau were able to slow the speed of a light beam to about 17 metres per second, and, in 2001, they were able to momentarily stop a beam.' Slowing light down is nothing new, it happens every time light travels through a medium other than the vacuum of space. Atmosphere, glass window, diamond, etc. It just so happens that we can now create in a laboratory these BEC's, a so-called "superfluid" which is basically a substance cooled to the point where nearly every atom collapses to the lowest quantum state (like, close to absolute zero). This gives it some interesting properties, like zero viscosity and an extremely high optical density. Hope that helps.
1) Wavelengths are too big: 1 micron is now a large number, and optics doesn't work much smaller than this.
.3 um range. Result: you can move the light around in smaller pipes, but the evanescent modes decay quickly, on the order of centimeters.
Not precisely correct. Most of the optical switches that Intel was developing back in 1999-2000 used evanescent modes to propagate along phosphorous-doped silicon waveguides with widths in the
2) There are no good nonlinearities. Anyone can make a linear OR gate optically, but to function as an effective digital technology you need nonlinearity and level restoration. This is missing in pure optical systems, except at very high power levels. The high power levels imply low density. There are some optical gates which process data in "femtoseconds," but ask them how long it takes to get to the next gate. Maybe someday someone will invent a great, low power, fast, optically nonlinear material. Don't invest in it yet.
Can you expand on this a bit? I'm confused as to how releveling implies high powers. Are you saying that the need for additional power input in order to improve the eye is prohibitive? Are you talking power input or optical power density?
Secondly, the gate region of a MOSFET (if doped appropriately to make the energy levels right) is an optically nonlinear material that makes a great switch. By setting the appropriate bias levels statically, one can change an optical OR gate into an AND gate into a NAND gate on the fly. While the switching rate is in tens of gigahertz, the reconfiguration rate is much slower, in the megahertz range, because you have to bleed off the common-mode biasing caps with another circuit and this takes longer. You can even have a buffered feedback circuit that does dynamic pre-emphasis over a few bits at a time. So what you have is an electrically biased and reconfigurable switch where the data path is all optical.
3) The serious workers are now mostly working in combined electronic/optical modes. The speeds here are limited by the gate speeds of the electronics, just like normal computers. You have to then ask if optics is a good (cost effective, space efficient, low power...) replacement for wire. Ultimately, the answer is probably yes, but there's an awful lot of work to do before that's true (for the distances of a few centimeters in high density computers, that is).
I agree. I'll bet that Intel's trying to perfect on-chip semiconductor lasers fabricated in their existing process. I'm pretty sure that they've nailed optical recievers in their process already.