Optical Processing

Posted:
in General Discussion edited January 2014
Wow. Check this sucker out: CNN Linky



8TFlops in one processor... that's amazing:

Quote:

The processor performs 8 trillion operations per second, equivalent to a super-computer and 1,000 times faster than standard processors, with 256 lasers performing computations at light speed.



I'm really curious how this would work... Are there optical logic gates? ICs? Wouldn't things still need to be converted to standard electric signal at some point?

Comments

  • Reply 1 of 6
    shetlineshetline Posts: 4,695member
    Quote:

    Originally posted by bauman

    8TFlops in one processor... that's amazing



    To pick a nit: the article simply says 8 trillion operations per second, not 8 trillion floating point operations per second. Still damned fast even if that's all just integer operations.



    I'd like to know just exactly what this processor can do, which the article doesn't make clear. Is it a full general purpose CPU, or more limited than that? Is there some kind of optical memory to go along with this processor that can remotely keep up to pace?
  • Reply 2 of 6
    placeboplacebo Posts: 5,767member
    I personally don't see why an optical processor would be any faster. I mean, why should the energy medium make computing accelerated? It's all the same to the software.



    (I still don't understand why Toslink is faster than electrical based standards)
  • Reply 3 of 6
    correct me if im wrong, but wouldnt an optical based processor not have to deal wiht the problems of heat that an electric processor would? and dosent light travel faster than electricity? or are they the same speed.
  • Reply 4 of 6
    bungebunge Posts: 7,329member
    I remember 10 years ago when they were talking about gates that could flip from reflective to transparent. Although I haven't clickyed the linky I think this is probably what they're talking about.



    Light without resistance travels faster than electricity with resistance. The bottleneck would be the speed of the gates. No, not Bill.
  • Reply 5 of 6
    shetlineshetline Posts: 4,695member
    Quote:

    Originally posted by ThunderPoit

    correct me if im wrong, but wouldnt an optical based processor not have to deal wiht the problems of heat that an electric processor would? and dosent light travel faster than electricity? or are they the same speed.



    Light, and all electromagnetic energy, has a constant speed, in a vacuum. The speed of light and other electromagnetic signals, when not traveling through a vacuum, varies depending on the frequency of the signal and the medium through which the signal travels. Off the top of my head: in ordinary copper wire, low frequency electrical signals travel at about 0.3c, but high frequency signals travel through coaxial cable at something like 0.9c.



    Much more relevant to the speed of computational processing power is how fast a switching element can react and change state. There must be something about the optical technology here that allows very fast switching.
  • Reply 6 of 6
    shetlineshetline Posts: 4,695member
    Quote:

    Originally posted by Placebo

    (I still don't understand why Toslink is faster than electrical based standards)



    You should be clear about what you mean by "faster". Switching speed, latency, propagation of the signal down the cable? Of course, even if you are clear about what you mean, the answer is: faster or not, it doesn't make a difference.



    Switching speed can vary depending on implemenation, and a lot of cheap Toslink modules aren't blindingly fast, but if they're fast enough to deal with the bandwidth of the audio data they are carrying -- and they're more than up to the challenge, switching speed is irrelevant.



    Any latency, related to switching speeed, through either a Toslink or a coax link is miniscule compared to other processing steps that a digital audio signal goes through. Does music sound any different if it starts playing at 10:37:00.000002 instead of 10:37:00.000001?



    Propagation, that is the time it takes for light to travel through a meter or two of fiber optic cable, or an electrical signal through a similar length of coax, is trivially short, trivially different, and irrelevant to the use of something like a CD or DVD player.



    Crazy golden-eared audiophiles will, of course, wax endlessly over sound quality differences they imagine to hear being caused by the color of the plastic jacket on the cable or feng shuiness of your cable routing, but given properly designed playback equipment (nothing fancy or expensive needed either) Toslink vs. coax should be absolutely irrelevant to audio performance. "Faster" doesn't enter into it.
Sign In or Register to comment.