Optical computers: Would theoretical red, green, blue "bits" = 1 Byte?

[MadRat]

000 = Black = 0
100 = Red = 1
010 = Green = 2
110 = Orange = 3
001 = Blue = 4
101 = Purple = 5
011 = Cyan = 6
111 = White = 7

I'm thinking that if an optical computer maintained three energy states (red, green, blue) relative to colors then every pathway could contain up to a full Byte of information. Is my logic flawed in this respect?

 

[Killbat]

They multiplex data on fiber optic networks now. I don't think it's like you describe, though. They run several, independant bitstreams on different wavelengths. This way, they can create a type of optical bus.

 

[MadRat]

True that the different wavelengths would not be colours, but I am only theorizing here.

Technically I am all wrong; it would be easier to use white light and have color-blockers to represent an "1"-bit rather than to make a colored light turn on each time. But cut me some slack, I'm just trying to reason out optical computing.

 

[Killbat]

If I let my mind drift, optical computing begins to take an interesting turn.
I picture computers that operate entirely on light, efficient enough to run on ambient infrared. Houses of the future will have invisible light emitters in every room to power electronics. Maybe copper power wiring will be replaced by fiberoptic lines, and each house will have a central "light pump".
I want one of those.

 

[MadRat]

A central "white light" pump is kind of how I'm seeing optical computing taking shape.

Since different frequencies of light follow curves in fiber differently its easy enough to sort out frequencies with a simply turn in the fiber. I haven't really thought out the whole process. Right now I just want to understand if storing a single "Byte" inside of a 3-colors is feasible or logical. Perhaps someone else knows a better way to do it?

 

[HaVoC]

I work at a fiber optic network company and my friends surmised that we will first see fiber optics within the computer for system/interconnect busses. The data rates you can push with light are simply insane, especially if you consider the short to medium range distances involved. The optical bus would provide a super-fast serial data path that will allow for interconnects of next generation super-CPUs. Eventually processors will go to optical technology but that is quite a ways away, though I couldn't tell you how long.

100 Terabits/sec * 1024 Gigabit/Terabit / 8 bits/Byte = 12.8 TeraBytes/sec!

 

[Mday]

efficient?

d00d, latest hardware uses lasers as the light for fiber.

--

also, optical gates =O

 

[ST4RCUTTER]

I think LED's would be a more efficient way to go with a computer. Lasers are better with longhaul optic cables because they're not apt to spread over long distances like LED's will. With such a short distance though, the LED would be the ticket. The real problem here is an optic transistor. How will logic states, such as on and off, be stored? Need an optic transistor. Then you need one small enough to place thousands on a chip. Then you need tiny LED's...you get the picture. We got a long way to go yet.

 

[SpideyCU]

ST4RCUTTER, I'd wager to say it's not as bad as you think. The folks over and OakRidge National Laboratory are doing some amazing work with optical components that size. Yeah, there's still a lot amount of work to be done, but the foundation for it is there, and progress is being made.

I just love the prospect of it all when the reports say that optical transistors can oust the speed of traditional ones by a hundred-fold.

EDIT: Removed my first comment because I deemed it to be too silly.

 

[Moohooya]

The transfer of data via a fiber optic is not slow, and so there is no desire to start using different wavelengths for different bits as this would then introduce more complex and costly lasers.

However, given an optical computer, you could double your computing power by adding a new laser with a different wavelength. When yet again you need more power, add a new laser. Rather ingenious really.

Actually I love my idea of a fast sequential access optical storage. You grow a specific crystal/build a box, than when light enters is will bounce around millions of times with total internal reflection until finally is escapes via a pre determined point. It would be designed so the internal path of the light was approx 100 miles. So now when you pump data in, it will come out approx 0.5msec later. If optical fiber will be puming as fast as 100 tera bits per second, then you'll have 500 gigabits inside, giving you access to 50+GB of storage will sub millisecond access. Need more storage, add a different coloured laser.

Granted sucks when the power goes out, there goes all you data. So you'd either have to have backup disks that maintain a copy, or else stick a battery on this and make it 100% reliable. But imagine all those huge databases. Wouldn't they love to be able to return results in under a millisecond.

 

[MadRat]

The optical transistor would be akin to electrical transistors. The difference being that multiple energy states could be set by using different colors of blocking mechanisms against the same white light "stream". The colors in light represent different wavelengths of light, just as different wavelengths of electricity operate in electronics equipment. Is this not how broadband works over power lines? Is this not how DDR and QDR work? Its just a matter of timing.

There are substances out there that can hold a photonic charge for making a crude form of optical RAM, too. You have to think outside the box a little, but I don't see optical computing being much different than how electronic computing is handled.