Google has announced its first-ever round of Google Focused Research Awards, and four EECS professors have received two grants totalling $425,000 for their research into energy-efficient computing.
Professor Thomas Wenisch has received a 3-year grant for a project he will pursue in conjunction with colleagues at University of Virginia, University of California – Santa Barbara, and Rutgers University, entitled “Data-Centric Approach to Energy Proportionality.”
The research funded by Google involves redesigning servers and data centers to improve their energy efficiency. Current data centers waste an enormous amount of power because they are designed to handle unusual spikes in user demand; for example, when everyone tries to activate their new cell phone on Christmas day. As a result, data centers are routinely overprovisioned; that is, they provide more computing power than is required under typical use. Under current system designs, underused systems still consume almost as much power as systems running under full usage.
In addition, the overall architectures of today’s servers – from the chips that power them to the software that makes them tick – are optimized for performance, and not efficiency. Through the design of new memory and processor architectures and techniques, the researchers seek to develop high-performance servers that are also energy efficient.