New device could lead to computers thousands of times faster

[Rohith_Kumar_Sp]

Researchers have worked out how to control light beams around curves that are twice as tight as ever before, a crucial step that will help them build super-fast computers that send data on light beams, instead of electrical signals. 
 

We already have materials capable of building these optical computers, but so far the process has been held up because we need devices that are able to control light beams around a computer without losing their energy. And now scientists from the University of Texas El Paso and the University of Central Florida in the US have created one.

 

The device, which looks sort of like a tiny plastic honeycomb, can turn beams of photons into optical wires that can bend around the tiny structures inside a computer.
 

 

"Computer chips and circuit boards have metal wire connections within them that transport data signals,One of challenges when using light is figuring out a way to make tight bends so we can replace the metal wiring more effectively." - Raymond Rumpf

 

 

Right now, conventional light waveguides, such as optical fibres, can steer light across great distances and gradual turns, but if a curve is too steep, the light beams escape and energy is lost. 

But the new device, which you can see above, is like a miniature lattice, and when light is beamed through it, the researchers found they could control beams through 90 degree bends that are twice as tight as has previously been possible - and all without any loss of the light beam's integrity or intensity.
 

The results, which have been published in Optics Express, set a new record in optics, but the team is now looking to double the tightness of the turns once again, so that they can beam light across tiny electronic circuit boards without any loss of energy or information.
 

Although there's still a lot of work to be done, if they can achieve that, we'll be on the verge of creating computers that are thousands of times faster than the ones we currently use. Just imagine the potential.



Source : http://today.ucf.edu/scientists-invent-new-way-control-light-critical-next-gen-super-computing/

[Pugs501]

[Guest Kloaked]

inb4 Intel/IBM patents

[ThisIsMyWorkAccount]

But can it run Crysis?

[Rohith_Kumar_Sp]

But can it run Crysis?

It's not a Graphics card 

[ThisIsMyWorkAccount]

It's not a Graphics card 

Okay?

[soaringchicken]

So glad I was born in this timeslot

[Doobeedoo]

Yey, light beam computers!

[tedjani]

[The_Strict_Nein]

For those interested, 20 micro meters is 20000 nano meters. Considering current transistor sizes are 14 nano meters, we've a long way to go before this is anywhere near a desktop

[spartaman64]

and costs thousands of times more

[Tonylu1595]

and costs thousands of times more

All new technologies cost more. If we didn't want to develop technologies just because they costs hundreds or thousands of times more, there would be no progress.

 

One example would be the computer.

[Sidiox]

For those interested, 20 micro meters is 20000 nano meters. Considering current transistor sizes are 14 nano meters, we've a long way to go before this is anywhere near a desktop

If I'm not mistaken light in this kind of a processor is faster than the electrical signals used to day, so light based cpus could have much larger cores with higher clockspeeds. But even the we are indeed still a long way off

[Maslofski]

for some reason i expected to see the 49 USB hub again

[sgloux3470]

I don't see this sort of technology ever reaching consumers.

 

It's simply too expensive to manufacture.  Unless we get some crazy advanced 3D printers (that can do optics) with economic yields... on top of the R&D for the hardware itself.

 

This is the sort of thing that will make its way into supercomputers within the next 40-50 years.

 

 

On the plus side, this makes me really excited for the possibility of significantly more miniaturized and accurate analytical instruments in my profession.  One of these days it'd be cool to have a tricorder. 

[LukaP]

 

This is the sort of thing that will make its way into supercomputers within the next 40-50 years.

 

more like 10 years for first working chips, 20 years for commercial use id wager

 

If I'm not mistaken light in this kind of a processor is faster than the electrical signals used to day, so light based cpus could have much larger cores with higher clockspeeds. But even the we are indeed still a long way off

yes, as well as being able to carry more than one bit with a light pulse, by modifying wavelenghts in it

 

For those interested, 20 micro meters is 20000 nano meters. Considering current transistor sizes are 14 nano meters, we've a long way to go before this is anywhere near a desktop

We are, but not because of the size. more because its such a new tech we dont actually have it researched enough to start mass production any time soon.

 

FYI, 10um class chips were a thing in the 70s, so with a little more RND, this could be made in a viable size soon, or as i said before, 10 years till first working chips, 20 for commercial products

[Makeishura/Taurus Silver]

"Right now, conventional light waveguides, such as optical fibres, can steer light across great distances and gradual turns, but if a curve is too steep, the light beams escape and energy is lost. "

 

 

You mean total internal reflection. PLEASE use proper physics terminology.

[Thorium19]

Now this is interesting, but as previously said, it's going to take a while before any chips are made using it due to this only just creating the lattice that allows it to turn 90^o, let alone making millions of them on a small enough scale close enough together for them to be able to operate in a system.

[mr moose]

Does this mean the optronic coupler will become a thing?

[LukaP]

 

"Right now, conventional light waveguides, such as optical fibres, can steer light across great distances and gradual turns, but if a curve is too steep, the light beams escape and energy is lost. "

 

 

You mean total internal reflection. PLEASE use proper physics terminology.

 

His sentence is still completly legitimate, even without using the term. he says optic fibres guide light through great distances and gradual turns. which is completly true. he only left out technical details as to why this can happen.

 

Here is the physics behind it.

 

and a video

[Makeishura/Taurus Silver]

His sentence is still completly legitimate, even without using the term. he says optic fibres guide light through great distances and gradual turns. which is completly true. he only left out technical details as to why this can happen.

 

I know. I am a Japanese Physics Graduate. Replying on phone does not help to write out extended responses, and being on a subway away from my PC doesn't help.

[FakezZ]

If I'm not mistaken light in this kind of a processor is faster than the electrical signals used to day, so light based cpus could have much larger cores with higher clockspeeds. But even the we are indeed still a long way off

Actually afaik electrical signals travel at the speed of light too, however you can store much more data on a single photon, which is what makes fiber so damn fast(plus the lack of noise etc).