Today’s bottomless pit of information creates an endless need for computers with greater storage, higher performance, and more energy-efficient systems. Computing can generate so much heat that servers must be positioned near water for cooling. And the power they require is beginning to exceed the capabilities of local power plants. So, is it time to rethink the computer?
IEEE Senior Member Elie Track thinks so. He is president of the IEEE Council on Superconductivity and cochair of the IEEE Rebooting Computing Working Group. A senior partner at Hypres, a superconducting electronics company in Elmsford, N.Y., Track chairs the group with IEEE Fellow Tom Conte, a professor of computer systems and software at Georgia Tech.
IEEE began its Rebooting Computing initiative last year to rethink the design and function of computers, hoping to attain greater performance on less energy. The working group members have a broad range of expertise, including computer architecture, multicore approaches, and high speed devices.
They will meet once a month by phone and through a website. The group plans to hold a meeting with thought leaders in December following the IEEE International Electron Devices Meeting in Washington, D.C., to finalize the redesign recommendations they will share with the community.
The idea is for the experts to weigh in on the pros and cons—and dispel concerns— about their particular arenas. “We’re envisioning an interdisciplinary exchange across the board, so experts who usually work in their own silos can now work together,” Track says.
His expertise is in superconductivity. Superconductive ICs utilize the intrinsic properties of superconductors, which include zero electrical resistance. That means “lightning fast speed” can be achieved, according to Track.
So far, superconductor ICs have been able to reach 10 times the speed of semiconductors, with potential for 100 times the speed, Track estimates. The technology has been applied in wideband satellite communications and other wireless transceiver applications.
“The need for cooling the superconducting ICs to cryogenic temperatures is greeted with fear and hesitation,” he says. “The Rebooting Computing group is the first step toward changing these attitudes. The reality is that cryocoolers are very reliable today, and failures often result from electronic rather than mechanical components.”
Superconductivity is a necessary component to realize the initiative’s goals, because computing at lower temperatures will save more energy than it takes to cool the systems, Track says.
With the exception of a two-year teaching sabbatical, Track has spent his 25-year career at Hypres, focusing on the R&D of superconducting microelectronics in communication applications. In the 1990s he applied the technology to help develop a self-contained primary voltage standard, to calibrate devices such as voltmeters, as well as secondary standards.
Superconductivity is indispensable to realizing a primary voltage standard—essentially defining the volt by international agreement—whereby the unit of voltage is derived from the fundamental units of frequency, the electron charge, and Planck’s constant. The relationship among these units is realized physically by a superconducting IC. Track is now applying superconductivity to improve the speed and capacity of transceivers for wireless communications and to improve the image quality at lower fields in magnetic resonance imaging.
Currently, wireless signals travel the airways as modulated analog signals. Communications systems reduce the signal to a lower frequency for digitization, which causes a data loss. Track is working on a method for near-instantaneous digitization that produces superior signal fidelity. The key to realizing the digitization at wide bandwidth is a superconducting analog-to-digital converter uniquely capable of producing very high linearity and dynamic range at high frequencies.
Superconductivity has experienced ebbs and flows of interest over the years. Now, Track says, high-performance and energy-efficiency needs are driving a revival. But the high amount of energy that it takes to bring circuits to super-cooled, superconducting levels makes computers built entirely of superconductors impractical. “It’s not a solution in itself but must work in tandem with other solutions such as multicore processing,” he says. “We must take a holistic look at the whole computing system.”
Track, born in Lebanon, graduated from the American University of Beirut in 1979 with a bachelor’s degree in physics. He was drawn to applied physics after taking a liking to the electronics used in physics experiments. He went on to earn a master’s degree and a Ph.D. in physics from Yale University in 1982 and 1988, with a focus on applied physics.
He joined Hypres as a staff scientist and eventually worked his way up to CEO, a position he held until 2000. When the company, with his blessing, brought in a new CEO to take the firm in a more commercial direction, he became a consultant and partner, exploring new applications for Hypres’s technology.
Throughout his professional life, he has kept in touch with student communities, which has honed his ability to explain his work in layman’s terms, an invaluable skill in describing his research to technical and nontechnical audiences.
Since 1999, Track has been a member of the Yale Science and Engineering Association (including a stint from 2007 to 2010 as president), raising awareness of Yale’s engineering program and raising funds for academic engineering scholarships.
From 2003 to 2005, he indulged a desire to teach by becoming a visiting physics professor at Fairfield University, in Connecticut, where he developed a course in wireless communications for non-science majors. And he helped mentor mechanical engineering students at the University of North Carolina at Charlotte, where he served as an adjunct professor of physics from 2008 to 2011.
a generation away
For now, the working group plans to apply its redesign recommendations to enhance the performance and energy efficiency of large systems. “The temperatures required by superconducting electronics make them more suited to large systems, such as servers and data centers, than for handheld or small consumer products,” Track says. “The performance of large systems would then enable ordinary gadgets to be more powerful and efficient by their connectivity through the cloud, as in cloud computing.”
After that, fundamental innovations in compact, portable, efficient cryocoolers are needed to create portable products using superconductors—a challenge for the next generation of engineers.