Sections
Front Page
Business
Arts & Leisure
Travel
Technology
Health & Science
Sports
Editorials & Opinion
Money Report
Special Reports
Article Index
Regions
Africa & Middle East
Americas
Asia - Pacific
Europe
Publishing Partners
The Daily Star, Frankfurter Allgemeine Zeitung, Ha'aretz, Italy Daily, JoongAng Ilbo and Kathimerini
Services
Currency Converter
Weather
Sponsored Sections
Classifieds
Help
Search

Advanced Search
Articles by Day
Monday
Tuesday
Wednesday
Thursday
Friday
Weekend
 
As Chips Reach Speed Limit, Makers Tap Into 'Clockless' Logic
Brad Spurgeon International Herald Tribune
Monday, December 17, 2001
PARIS Those cold, calculating machines called computers have, it turns out, a heart. But, unlike a human heart, a computer's heart - its microprocessor - always beats in the same rhythm, sending not blood but data through its system at a steady rate, regardless of the task at hand.

Lately, however, as chips grow ever more crowded with transistors and clock speeds increase to as much as 2 gigahertz - or 2 billion clock ticks a second - the system has begun to reach its limits.

Challenging as that may be to the chip industry, it is good news for the small, yet worldwide community of private and academic researchers who are perfecting a kind of lateral-thinking, anarchic method of chipmaking based on asynchronous logic, which does away with the clock altogether.

"The nirvana for the 'clockless' people like us is that clocked design is just getting more and more complex as we're able to put more and more transistors together running at higher and higher clock speeds," said David Lamb, president of Theseus Logic Inc., an asynchronous chip company in Florida. "That whole paradigm is just getting more difficult every day, and the complexity is becoming intractable."

Furthermore, it turns out that clockless chips, in addition to being more energy efficient, can also work faster, more quietly and more securely than synchronous chips. All of which makes them perfect for applications such as computer networks, mobile phones, smart cards and embedded medical devices.

"I'd be surprised to see your next laptop be fully asynchronous in the processing department, but I would be less surprised to see that of your next mobile phone," said Steve Furber, a professor at the University of Manchester and founder of Self-Timed Solutions, an asynchronous chip company connected with the university.

The personal computer industry has lived for so long with clocked chips that it will take years to develop the infrastructure necessary to mass-produce asynchronous systems for computers. But the smaller devices have begun to enter the market.

Philips Electronics NV of the Netherlands began developing tools for asynchronous chips in the late 1980s. It introduced its Myna pager in 1998 with an asynchronous chip because it produced less radio interference and extended battery life.

Sharp Corp. of Japan built an asynchronous media player in 1997 for computers, audiovisual equipment and mobile devices to process video, graphics and audio files .

Intel Corp. has researched asynchronous design, and its latest processor, the Pentium 4, incorporates some aspects of asynchronous logic. But it falls far short of the kind of future envisioned by Ivan Sutherland, head of asynchronous research at Sun Microsystems Inc., who told the computer industry last March that asynchronous chips could revolutionize the microprocessor.

Although the asynchronous approach has existed for decades, the industry grew up around synchronous chips, which were simpler and more reliable. They were also easier for consumers to understand. The processor wars between such companies as Intel Corp. and Advacned Micro Devices Inc. are still based on the idea that the faster the clock, the faster the computer.

In fact, where a synchronized processor will operate only as quickly as the slowest of its parts - some 700 megahertz processors work more quickly than other 1 gigahertz processors - asynchronous processors operate at the average speed of their components.

Where the synchronous processor waits for a clock cycle to finish before starting the next task, an asynchronous one can do multiple tasks at different speeds.

"It's very much more the way that people normally work," said Jim Garside, a senior lecturer in Mr. Furber's group at the University of Manchester. "You don't do one thing per hour before going on to the next thing. You don't say, 'That task only took five minutes, now I'll have to wait for the clock to go around again before doing the next job.'"

Data-crunching also gets done more quietly in an asynchronous chip, in which lower radio signal noise causing less interference with other parts of the system. Mr. Garside compared it to the noise made by a group of marching soldiers versus the sound made by the random footsteps of the general public.

Asynchronous chips use 10 percent to 50 percent less energy than synchronous chips, in which the clocks are constantly drawing power. That makes them ideal for mobile communications applications - which usually need low power sources - and the chips' quiet nature also makes them more secure, as typical hacking techniques involve listening to clock ticks. That is what attracted companies like Theseus and Self-Timed Solutions to the technology for use in smart cards.

Despite the advantages, asynchronous technology faces a battle in the mainstream, not only in infrastructure but also in image.

Carl Howe, principal analyst at Forrester Research Inc., voiced the popular industry view of asynchronous chips as being mostly "an intellectual curiosity," while adding that it "requires a bigger chip to do the same function as a similar synchronous chip, and that means that asynchronous chips cost more to make."

Clockless-chip supporters counter that while asynchronous chips do require more real estate, synchronous chips use one-third of their space for the clock.

Asynchronous chips, they say, are more expensive to build because most of the computer-assisted design tools were made for synchronous chips, not asynchronous ones, and few technicians are trained in the field.

Kees van Berkel, a researcher at Philips, said he had run into more business-linked problems than technological ones, specifically, finding investors "willing to take the risk of being the first to apply such an exotic technology."

To penetrate the mainstream, Mr. Furber, who is attending Britain's semiannual meeting for asynchronous research in Cambridge this week, said products might have to use a mixture of technologies.

"Although my group has spent 10 years building fully asynchronous parts, I'm not really convinced this is the right way to take on the world," he said. "I feel the more promising strategy may be to meet the world halfway and try to build sensible mixes of clocks and asynchronous stuff on the same chip."


Printer Version
 

For Related Topics See:
Technology


 

 
Subscriptions E-mail Alerts About the IHT : Privacy & Cookies : Contact the IHT  

Copyright © 2001 the International Herald Tribune All Rights Reserved
Terms of Use | Contributor Policy