Slashdot Mirror

← Back to Stories (view on slashdot.org)

Intel Announces Laser Breakthrough

Posted by ryuzaki0 on from the speed-of-light dept.

AdmiralWeirdbeard writes "Intel has just announced a breakthrough in laser technology allowing a continuous laser wave on a silicon chip. Apparently they devised a method to sap the interfering field of electrons previously generated in silicon by the lasers. Intel says that hardware exploiting the advance might begin appearing at the end of the decade."

2 of 185 comments (clear)

  1. Power supply... by jim_v2000 · · Score: 3, Interesting

    The article didn't mention this, or I didn't see it, but wouldn't using lasers instead of wires really use a lot of power? Epecially when you start using a lot of them. But then again, maybe these are really low powered lasers and don't take much power at all. Anyone have any ideas or know anything about wires vs lasers?

    --
    Don't take life so seriously. No one makes it out alive.
  2. Re:Am I the only one that doesn't get it? by thpr · · Score: 4, Interesting
    There are probably two major uses. The first is in an optical switch. Traditionally, switches were OEO (optical-electrical-optical) until the all-optical craze hit in 2000. OOO (all optical) are (in theory) able to switch the light faster, which reduces latency, power usage, and lots of other things in the optical core of the network. However, if you eliminate the separate optics devices and can run the optics directly onto the semiconductor, OEO may be a lot more competitive (meaning cheaper). Go search LightReading for "OEO" or "OOO" to follow that debate (of whether there is benefit to all-optical and the current state of the art. [Infinera is a rather interesting startup driving OEO into the future to compete with OOO]

    The second major use would be chip-to-chip interconnect. However, this becomes a challenge, as you try to keep a ribbon of fiber-optics (think 200-2000 strands) perfectly lined up with the lasers on the die. I'm not saying it can't be done, but it is one of the hurdles to face before it could be used that way in mass-produced systems like a PC. The theory goes that at about 1 foot per second, electrical propagation between chips is causing us lots of headaches. HyperTransport and other technologies make some advances to get around the plain limits, but there are still major problems with sending high-speed signals on circuit boards. Even if this can't help speed up absolute memory access time, it could help to improve throughput between memory and the processor, helping to avoid some of the single-threaded bottlenecks that led IBM and its partners to develop Cell