Welcome to Laser Pointer Forums - discuss green laser pointers, blue laser pointers, and all types of lasers

Buy Site Supporter Role (remove some ads) | LPF Donations

Links below open in new window

FrozenGate by Avery

Silicon Photonics

Joined
Dec 26, 2007
Messages
6,129
Points
0
read from the inquirer, thought you would like this :)

SUN MICROSYSTEMS has purportedly received a $44 million contract from the Pentagon to explore replacing the wires between computer chips with laser beams. The technology, called silicon photonics, connects chips in such a way that they are able to communicate with each other at incredibly high speeds. If the technology works, it could signal the dawn a new generation of faster, more compact and energy efficient computers.

Semiconductor companies generally produce processor and memory chips by breaking silicon wafers, with thousands of tiny identical circuits etched into them, into pieces. This is supposed to improve manufacturing yields and ensure that a small defect doesn[ch8217]t wreck all the chips.

But now, according to the New York Times, Sun[ch8217]s boffins may have discovered a technique that would be able to do away with the bottleneck data traffic that occurs when information has to be moved around quickly through wires to solve highly complex problems.

[ch8220]Wafer scale integration[ch8221], which involves building bigger chips on a single wafer of silicon, is something that computer scientists have long been suggesting as a technique for developing quicker and faster computers. Silicon Photonics, too, is becoming an increasingly popular field in the Semiconductor firms[ch8217] efforts to replace electrical wiring in processors with optical networking directly built in.

Sun[ch8217]s project will apparently run for five years, in conjunction with Stanford, the University of California, San Diego, and two silicon photonics firms, Luxtera and Kotura. The project builds on previous research done by Sun which attempted to connect chips electrically by stacking them edge to edge. The new method will be based on accurately aligning chips with high precision in order to be able to transmit beams of light across their surface via very narrow channels called wave guides. Basically, every chip would be able to communicate directly with every other chip through a beam of laser light carrying tens billions of bits of data a second. Sun boffins call the new system the [ch8220]macrochip[ch8221], but admit that it has a high risk of failure and is definitely a risk for the company.

A researcher at Sun Laboratories leading the project, Ron Ho, reckons [ch8220]We expect a 50 per cent chance of failure, but if we win we can have as much as a thousand times increase in performance[ch8221]. Researchers also claim that if their idea works, they could finally break Moore[ch8217]s law which states that the number of transistors on a computer chip has to double roughly every two years.

Sun reckons that if the research bears fruit, the project could lead to the manufacturing of smaller computers that were up to a thousand times faster than the ones available today.


yay

i hope they can create it.

how? dont know.. but if it will make our cpu's better, then go ahead.
 





It wont make CPUs better but it will make intercomponent communication exponentially faster, which is better anyway since it would virtually eliminate bottlenecks.
 
And it certainly won't help us make lasers, considering the scale we're talking about. I could be wrong though.
 
this is the future of computers...

man 1: "yea man, i got myself a new 99mw today"
man 2: "what do ya mean, a laser?"
man 1: "nahh, im talking about my computer" ;D
 
anyone here from the philippines? im from philippines making DIY lasers. Chido and Gazoo knows me well.... ;)
 
adgmeijin said:
anyone here from the philippines? im from philippines making DIY lasers. Chido and Gazoo knows me well.... ;)
wtf has that got to do with anything??? :P
 
adgmeijin said:
anyone here from the philippines? im from philippines making DIY lasers. Chido and Gazoo knows me well.... ;)
A better question would be who cares.
 
GD!! What the hell is going on here?

Adgmeijin, Perhaps you should:

See this video:
http://www.laserpointerforums.com/forums/YaBB.pl?num=1206361389


Then explain yourself here:
http://www.laserpointerforums.com/forums/YaBB.pl?num=1206323969


Anyway, back to the topic, seems like technology is going for a leap of faith! Willing to invest $44mill! I’m not sure how the material efficiency would be improved over the cheap common silicon chips. Plus, (not that it would be too hard), wouldn’t heat still need to be dispersed? And you couldn’t vent it because dust will screw it up. Or maybe fans will become a thing of the past! - Just my thoughts.
 
heat will always be a problem (until some revolutionary new thing proves me wrong)

so lasers will generate heat...but they will replace some interconnections...

what i dont know is HOW they're gonna include lots of diodes with SUPER high precision into one TDP package (is it called TDP or am i wrong? im trying to talk about the microprocessor's square package)
wouldnt a simple hit move the diode, even if it is 2 micrometers?
 
bob1122 said:
And it certainly won't help us make lasers, considering the scale we're talking about. I could be wrong though.

Actually, it very easily could. Lasers are expensive because they're hard to make, because we make them out of III-V semiconductors like the GaN system and on expensive/not ideal substrates. If you can make lasers on silicon (even by putting other materials, like a III-V, on the silicon), they become extremely easy to make relative to the III-V systems, and therefore much cheaper.

These applications will never make high-powered lasers. But, it could spurn a technology revolution that would allow the manufacture of lasers on silicon substrates, allowing for cheaper manufacture and for other people to work on making these cheaper lasers with higher powers suitable for us.
 
wait....so building a laser onto a silicon wafer would be possible? i mean, NOT using a diode? (for dpss, i know this just isnt correct :P)....diode pump without a diode..cmon


or is it that the driver and circuit will be on the silicon chip and the diode on the end of it, thus making the lasers relatively small except for the diode?
 
I think you misunderstand what a "diode" looks like or is. The diode in the cans that we use is a small rectangular chip cut out from a wafer and placed in a cylindrical can. But in its most basic form, the diode is simply thin films placed upon a crystal wafer. The whole diode is the system of thin films within the can with the 2 wires attached, the can is just a means for mounting it to or in things. So yes, the idea is to build the diode on a silicon wafer instead of on a sapphire or silicon carbide or gallium nitride substrate crystal, making it cheaper and making it possible to use light for data transmission directly on a silicon computing chip.
 
so the "diode" we are talking about is the micro sized chip that goes ONTO the bigger diode, what we see, which acts as a dissipator and a base for the chip?
i never understood that, i am getting confused :P
 
nikokapo said:
so the "diode" we are talking about is the micro sized chip that goes ONTO the bigger diode, what we see, which acts as a dissipator and a base for the chip?
i never understood that, i am getting confused :P

Take a look at this photo, this is an open can diode, but the closed cans have the same thing inside, it's just totally surrounded. You can see the wires connected to the diode, and the rectangle that is the actual diode. The diode, the rectangle shown, could be constructed very small on a wafer, and used to transmit data. What you see as a diode is the whole can containing the diode (which is technically only the semiconductor chip/die), its connections, a mounting block, the "mirrors" that reflect light back and forth making it a laser diode, and heatsinking material.
 

Attachments

  • diode.GIF
    diode.GIF
    39.5 KB · Views: 112
Another, better picture showing the diode die a little better. Thanks m.carroll for the nice macro.
 

Attachments

  • diode2.PNG
    diode2.PNG
    930.6 KB · Views: 85
pullbangdead said:
[quote author=bob1122 link=1206394336/0#2 date=1206409616]And it certainly won't help us make lasers, considering the scale we're talking about. I could be wrong though.

Actually, it very easily could. Lasers are expensive because they're hard to make, because we make them out of III-V semiconductors like the GaN system and on expensive/not ideal substrates. If you can make lasers on silicon (even by putting other materials, like a III-V, on the silicon), they become extremely easy to make relative to the III-V systems, and therefore much cheaper.

These applications will never make high-powered lasers. But, it could spurn a technology revolution that would allow the manufacture of lasers on silicon substrates, allowing for cheaper manufacture and for other people to work on making these cheaper lasers with higher powers suitable for us.[/quote]


Hehe, absent minded physicist eh?

What I meant was that it won't help US light matches or burn black tape ;D
 





Back
Top