NJ NY Affordable Web Design Personal and Business Website SEO Computer Repair Virus Removal Bogota Bergen County NJ
  • Home
  • Prices and Services
  • Make a Payment
  • Contact David
  • Blog
MIT almost produces an optoelectronic computer chip by Sebastian Anthony 02/16/2012
0 Comments
 
Two researchers at MIT, working with an IBM foundry, have almost produced a
monolithic optoelectronic computer chip — a chip that integrates both silicon
components and optical interconnects.


This is primarily significant because of MIT’s use of a monolithic process; a
process where all of a chip’s features are fashioned on a single die. Every
modern chip is fabricated in this way — but as it stands, optical equipment is
never produced like this. Currently, optoelectronic solutions involve a silicon
chip and bulky off-chip optical devices, such as lasers, detectors, and
modulators. These devices consume large amounts of electricity and are much too
large to include in a laptop or desktop.


As a result, the telecommunications industry is the only real sector that
utilizes optical networks. MIT’s thinking is that if monolithic optoelectronic
chips can be produced — where lasers, waveguides, photodetectors, and modulators
are all on the same piece of silicon — companies like Intel and TSMC will be
much more likely to pursue optoelectronic solutions. So far, MIT has produced a
monolithic optoelectronic chip with on-die waveguides, ring resonators, and
waveguides — but they haven’t succeded at etching channels under the waveguides,
which is necessary to prevent light leakage.


Vladimir Stojanovic, one of the researchers working on the project, admits
that existing processes would have to be changed a little for monolithic
optoelectronic chips to become a reality. He also says, however, that it could
be easier to add optical components to chips built from the bottom up, which
covers most “3D” designs, including Intel’s 22nm FinFET chips.


In the next few years, then, we are likely to see the first generation of
optoelectronic chips. These early chips would have off-chip lasers, meaning they
would likely be used for off-die interconnects, such as between your CPU and
RAM. On-chip lasers are a little further away, but when they finally roll around
we’ll be looking at multi-core chips that use light instead of electricity to
communicate.


The main advantage of using light is a huge reduction in power, and
potentially a vast increase in bandwidth. In general, increasing bandwidth over
an electrical connection requires more power. Computers already constitute a
large percentage of humanity’s power needs, and optical interconnects could go a
long way to keeping power consumption down in the long term. The same
multiplexing techniques that are used to carry terabits per second over fiber optic networks
would also apply to computer chips, too.


Finally, MIT speculates that it should be possible to create chips where a
single core can communicate with every other core at the same time using lasers.
Basically, imagine a 100-core chip where each core has its own laser. Instead of
shuffling data to and from memory/cache, each chip could simply fire data
directly to other cores. Not only would this save power, but it would be
incredibly fast.

 


Comments


Your comment will be posted after it is approved.


Leave a Reply

    Click on the Title of Blog to Post or View Comments.

    Archives

    February 2012
    August 2010
    July 2010
    February 2010

    Click to set custom HTML
    WARNING: Google Buzz Has A Huge Privacy Flaw   1/11       Nicholas Carlson | Feb. 10, 2010, 4:49 PM  Update 3: Here's the story of how panicking just enough may have saved Google's answer to Facebook and Twitter.

    Update 2: Google has made further changes that address most -- but not all -- of our concerns.

    Update 1: Google released some privacy fixes for Buzz. They're a nice start, but we don't think they go far enough. Read more here: Sorry, Google's Improvements To Buzz Don't Fix Privacy Flaw

    Earlier: There is a huge privacy flaw in Google's new Twitter/Facebook competitor, Google Buzz.

    When you first go into Google Buzz, it automatically sets you up with followers and people to follow.

    A Google spokesperson tells us these people are chosen based on whom the users emails and chats with most using Gmail.

    That's fine.

    The problem is that -- by default -- the people you follow and the people that follow you are made public to anyone who looks at your profile.

    In other words, before you change any settings in Google Buzz, someone could go into your profile and see the people you email and chat with most.

    A Google spokesperson asked us to phrase this claim differently. Like this: "In other words, after you create your profile in Buzz, if you don't edit any of the default settings, someone could visit your profile and see the people you email and chat with most (provided you didn't edit this list during profile creation)."

    (Freaking out already? Here's how to IMMEDIATELY make these list private and then edit them >)

    When you first post to Google Buzz, there is a dialogue box that reads "Before participating in Buzz, you need a public profile with your name and photo." It also says -- albeit in tiny gray letters against a white background, "Your profile includes your name, photo, people you follow, and people who follow you."

    But it does not say that these publicly viewable follower lists are made up of people you most frequently email and chat with.

    Even if it did say that, we doubt most users bother to read the text in the dialogue box before clicking "save profile and continue."

    (This is why it's always safest for Web services providers to make it so sharing information is always an "opt-in," rather than "opt-out," setting. Just ask Facebook, which still remembers Beacon.)

    There is also a "Welcome To Buzz" panel that shows who you are following and who is following you. In a long bit of unbolded text, it says "Buzz is a new way to share updates, photos, videos and more, and start conversations about the things you find interesting. You're already set up to follow the people you email and chat with the most."

     If a user notices the box, it might help users "catch" that they might be following people they don't want the world to know they're following. But you don't have to close the box to use Buzz. Closing the box does not trigger a warning or anything else that alerts the user they've agreed to publish a list of the people they email and chat with most.

    It looks like this:

    The whole point is: Google should just ask users: "Do you want to follow these people we've suggested you follow based on the fact that you email and chat with them? Warning: This will expose to the public who you email and chat with most."  Google should not let users proceed to using Buzz until they click, "Yes, publish these lists."

    In my profession -- where anonymous sourcing is a crucial tool -- the implications of this flaw are terrifying.

    But it's bad for others too. Two obvious scenarios come to mind:

    • Imagine if a wife discovering that her husband emails and chats with an old girlfriend a ton.
    • Imagine a boss discovers a subordinate emails with executives at a competitor.
    A Google spokesperson tells us the followers lists are public by default so that people can quickly find new people to follow. Obviously, that's a good thing for Google, which is hoping to get as many people using Google Buzz as soon as possible. It's also meant to be helpful for users. And for those who are unconcerned with telling the world who they email most, it is. But for everyone else, it's terrible.

    It gets to a deeper problem with Google Buzz: It's built on email, which is a very different Internet application than a social network.

    The good news for Google is that this is a very easy problem to fix. Google must either shut off auto-following, or it must make follower lists private by default as soon as possible.

    RSS Feed