MSU collaborative research leads to new supercomputing technologies
As you’re updating your cover photo on Facebook or plotting your next trip using Google Maps, probably the last thing on your mind is how much computer memory and energy you’re using.
But it’s uppermost on the minds of scientists and researchers who work in the energy field.
In a research project funded through the Intelligence Advanced Research Projects Activity organization, several Michigan State University researchers, as well as scientists from Northrop Grumman Systems Corp., developed a superconducting magnetic memory element that has greatly reduced heat generation and power consumption compared to conventional alternatives.
The study is published in the journal Nature Physics.
“Supercomputers consume enormous amounts of power in exchange for their computational abilities,” said Norman Birge, a professor in the MSU Department of Physics and Astronomy and co-author of the paper.
According to the Natural Resources Defense Council, data centers are one of the largest and fastest-growing consumers of electricity in the United States. In 2013, U.S. data centers consumed an estimated 91 billion kilowatt-hours of electricity and are on track to reach 140 billion kilowatt-hours by 2020.
When IARPA projected what larger-scale computers might look like in the future, it seemed they would become almost energy prohibitive. It became evident that new computer technology would need to display much more energy efficiency than current technologies in order to reduce world electricity use.
“One of the major impediments to superconducting computing technology had been the lack of an efficient memory that took advantage of the superconducting state,” said Eric Gingrich, a physicist at Northrop Grumman who was the lead student on this project while he was a student at MSU. “This project has demonstrated experimentally a solution to that problem, and brings the technology one step closer to being realized."
IARPA hopes the superconducting circuits developed based on this new magnetic memory element will use 10,000 times less power than conventional semiconductor circuits, so that after the cooling is taken into account, the final efficiency will be 100 times less power for the superconducting supercomputer.
Additional MSU researchers contributing to this project are Bethany Niedzielski, Joseph Glick, Yixing Wang, Reza Loloee and William Pratt Jr. Don Miller, a scientist at Northrop Grumman, also contributed.