ASU engineer aims to help create a more energy-efficient supercomputer
We have progressed far beyond the age of the computer into the age of the supercomputer center, says Arizona State University electrical engineer Nathan Newman.
Governments, economies, health care services, power and transportation systems, and national security operations now depend increasingly on reliable access to the computing technology that enables information gathering, storing and analysis on massive scales that can be provided only by large interconnected clusters of high-powered computers.
With that growing dependence comes growing demand for ever-faster and expanding computing capacity, says Newman, a professor in the School for the Engineering of Matter, Transport and Energy, one of ASU’s Ira A. Fulton Schools of Engineering.
The big challenge in meeting the demand is the great amount of electrical power it would take to increase supercomputing capabilities using current semiconductor technologies and operating systems.
“It has gotten to the point that supercomputer centers cannot do much more than they are capable of doing right now without a paradigm change in how they operate,” Newman explains.
That situation will also make it difficult for popular providers of search engines, web portals and other Internet services, such as Google and Yahoo, to significantly improve and expand upon what they can offer customers, he adds.
Much of his research lab’s effort is now focused on breaking through current technological limits by developing computer logic and memory devices that can enable massive improvements in the speed and capacity of supercomputing facilities.
A team that includes Newman’s research group at ASU, the Northrop Grumman corporation and Michigan State University researchers has recently been selected by the Intelligence Advanced Research Projects Activity (IARPA) – an agency under the office of the Director of National Intelligence – to collaborate on a multimillion-dollar project to make that advance.
The larger supercomputing facilities, such as those run by Yahoo, Google and the National Security Agency, now use an amount of energy equivalent to about 10 percent of the energy that a nuclear power plant can generate – or enough energy to meet the power demands of about 100,000 homes. Increasing the amount of computing by a factor of 10 is prohibitive, due to the tremendous cost of this much additional energy alone, Newman says.
Newman’s research team is optimizing the use of new superconducting materials to develop a new kind of digital circuit. The circuit could potentially enable supercomputer systems to require much less energy to operate effectively. The new digital circuits can be developed using current fabrications facilities, Newman says.
His group’s work with digital-circuit technology has recently led to the development of a computer memory device that operates more rapidly and more energy-efficiently. It could enable supercomputing systems to operate as much as 50 times faster and perform 50 more times the number of operations while using 50 times less energy, Newman says.
The advances spring in large part from work began decades ago by physicist John Rowell and his colleagues at Bell Laboratories, which achieved several breakthroughs that paved the way for today’s advanced electronics and computing technology.
Rowell, a member of the National Academy of Engineering, the Royal Society and the National Academy of Sciences, has joined ASU as a research professor in the School for Engineering of Matter, Transport and Energy, and teamed with Newman’s research group.
Rowell was a pioneer in revealing the basic physics involved in the workings of superconducting devices and discovering what materials enabled the devices to perform most effectively, Newman says.
Newman, the Lawrence Professor of Solid State Science at ASU completed a two-year term in January as chair of the U.S. Committee on Superconductor Electronics. He has authored or co-authored more than 200 technical papers published in science and engineering research journals, and been awarded 12 U.S. patents.
Newman is a Fellow of the American Physical Society and the Institute of Electrical and Electronics Engineers (IEEE), and editor for materials science and engineering subject matter for the journal IEEE Transactions of Applied Superconductivity. He is a winner of the IEEE Van Duzer Prize, awarded annually for the best research paper published in that journal.