(Page 2 of 2)
Intel has already made a name for itself in silicon photonics. In 2005, the company announced a silicon laser, an engineering feat that many thought was impossible due to the physical properties of silicon. And within a couple of years, the company's engineers demonstrated other silicon-based optical devices, such as high-performing modulators for encoding data onto light, and high-performing detectors for capturing the light-encoded data. Other researchers and companies, including those at University of California at Santa Barbara, University of Southern California, and MIT joined the fledgling field of silicon photonics. In 2007, optical chipmaker Luxtera announced an optical cable called Blazar that contained some silicon-based chips and was designed to connect servers in data centers.
The first generation of Light Peak cables will use the same sort of $75 optical chips found in telecommunications devices. But Intel has employed some tricks to drive down cost by more than a factor of 10, says Victor Krutul, director of Intel's optical I/O team. For one, the chips don't need to transmit data over the distances of telecom devices. For another, they don't need to last as long or withstand harsh conditions. Because telecom chips in consumer cables won't need to last for decades or withstand heat and humidity, manufacturing standards can be relaxed and allow the chips to be made more inexpensively.
While cables may not seem a cutting-edge technology, says Alan Willner, professor of electrical engineering at the University of Southern California, it is the ideal early application for silicon photonics because the market is potentially huge. "Silicon-based cables provide high-bandwidth connections, and frankly, nowadays everything is high-bandwidth." Willner adds that the devices inside the chips, such as lasers and detectors, may not be the highest-performing, but they don't have to be. "What they do have to be is robust and cheap and manufacturable," he says. "From a user's point of view, all they see is a lighter, cheaper, and faster cable than what used to be. That's a great thing."
"We're launching an optical technology for a mainstream platform," says Mario Paniccia, director of Intel's photonics technology lab. "We're going to start at 10 gigabits per second and scale to 100 gigabits per second." Paniccia was reluctant to forecast a timeframe for silicon photonics to be used in Light Peak, but in order to achieve the higher bandwidths, he says, silicon photonics will need to come into play.
Intel says it will be able to make Light Peak cables up to 100 meters long. And because new connective technologies, be they wireless or wired, require standards and industry collaboration, Intel is working to form partnerships with various companies. At the developer forum, Perlmutter announced that Sony is supportive of the technology, with more announcements planned.
Replacing metal wiring with fiber optics could change everything from supercomputers to laptops.
Chips that let errors happen, then correct them, use less power overall.
An energy-efficient silicon device compresses light to make ultrafast signals.
Silacon proposed to Franklin Signal four years ago using fiber optic wires to transmit signals within circuit boards. Franklin demonstrated short distance application of the technique.
What I failing to comprehend is why people are turning to a very fragile product to replace copper. True, while copper is slower, it can at least BEND before it breaks. Optical fibers just break, as they are made of glass.
Optical fiber is great for long distances, as you don't need repeater sites, but if you run them over with a HMMWV, they will snap in half. Ethernet on the other hand, being made of copper, can get run over, and still function, as it is a malleable material. I congratulate Intel on their ingenuity and courage to bring something new to the table, but optics and copper are probably better off the way they are utilized now.
EDIT: By the way, khairulsyahir, I see where you are coming from. I have into the same problems with multiplexers. It doesn't matter how fast your pathway is (for example, a CV-FOM that runs 8192 kb/s), if the original device cannot transmit that much information that fast (an FCC-100, which runs at 2048 kb/s). I've seen it, and it is vastly redundant, unless you plan to upgrade the system, or run multiple systems through it (multiplexing multiplexers, what a laugh).
A Silicon Valley start up, Silicon Pipe, demonstrated copper capable of 10 to 20 Gbps data rates over distances up to 1 meter at relatively low power through a differential pair by making connections direct from chip to chip or chip to connector. They also described a path to high speed memory. There was a multi-company paper presented on the technology at DesignCon a few years ago.
Since copper is robust, as was mentioned, and there will be need for copper for power, ground and lower speed circuits anyway, it is uncertain of what the advantage might be. (at this time, that is...)
Re: Copper shown good to 10G also
Copper introduces a delay in the signal relative to glass in fiber optics as the signal propagation is slower at the same data rate.
Re: Copper shown good to 10G also
The electric field travels through copper at about 75% the speed of light in free space. Light in the fiber travels at 70% the speed of light in free space. In reality, the propogation delay of fiber and copper are both in the range of 1 to 1.5 ns per foot.
If there is a USB connector on both ends, then it shouldn't really matter what's inside the cable itself.
A USB 3.0 (480Mbit)cable with copper conductors for the signal, OR a USB 10Gbit cable with electronics embedded in the connector at both ends, but linked by optical fibers in between.
Years back Sony had plans on marketing a Hard Disk Drive using a sealed Optical media instead of the traditional Magnetic platter. As long as they supplied a standard ATA-IDE interface, what was inside the drive package, was none of your business!
Manufacturing in the United States is in trouble. That's bad news not just for the country's economy but for the future of innovation.
This document is part of the “How-To Guide for Most Common Measurements” centralized resource portal. This tutorial provides a detailed guide for measurement and device considerations to take temperature measurements using thermocouples. Get an introduction to thermocouples, which are inexpensive sensing devices widely used with PC-based data acquisition systems. Also review some specific thermocouple examples and learn how thermocouples work and ways to integrate them into a data acquisition measurement system.
View full PDF >
Listen to story > 
On Twitter
Become a Fan on Facebook
Subscribe to the Feed
khairulsyahir
3 Comments
Speed bottleneck
I am wondering if the Light Peak will give any significant advantage for current and near-future devices, seeing that the storage and memory components in electronic devices nowadays have far less writing and reading speed? The USB2.0 boasts 400+ MB/s transfer speed, but we barely reach 10% of that speed in actual operation of an external hard drive, for example.
Sure, 10GB/s sounds impressive, but will there be any useful application of it now or let's say in 3 - 5 years' time?
Reply
Guest (michel.jansens@ulb.ac.be)
Re: Speed bottleneck
Well, actually USB2 has a raw capacity of 480Mbits/second.
Remove the protocol overhead and you are sensibly below that number.
Most usual 1TB SATA2 disks can saturate this connection in burst read.
SATA2 in itself provides up to 3Gbits/sec
That's why USB3 is coming soon.
But length of cable will never beat fiber.
Reply
khairulsyahir
3 Comments
Re: Speed bottleneck
True, but still, the bottleneck remains inside the hard disk itself. How much use can we make of a 3Gbit/s connection if the read/write head inside the hard disk can only work as fast as 20 - 40MBytes/s?
Reply
justahick
19 Comments
Re: Speed bottleneck
True, but where hard disks are replaced by SSDs, such as in many racks of servers, I could see how this would increase throughput.
Also, it should help with keeping temperature lower inside the server, which should help with reducing power and cooling costs, although the article does not mention the power requirements, which seems to be an odd omission.
Down the road further, I'd like to see multi-wavelength photonic interconnects both between the cores in a CPU, and the memory they use.
Reply
rudnric
4 Comments
Re: Speed bottleneck
Speed Bottleneck doesnt have a clue about USB or HDDs.
"The USB2.0 boasts 400+ MB/s transfer speed, but we barely reach 10% of that speed in actual operation of an external hard drive, for example."
Youve got it completely wrong. USB 2.0 is 480 megabits per second (Mbps), not megabytes(MB/s).
Mechanical HDD's have surpassed the bandwidth of USB years ago. Even old HDD's can can hit physical read rates from the media, over 50MB/s. Effective transfer rate of USB 2.0 is only 20-40MB/s tops due to data overhead built into USB protocols, not 60MB/s as the raw data rate of USB 2.0 would suggest. And CPU utilization is also quite high because of that protocol when at maximum data rate.
Reply