SAN JOSE, Calif. New modeling tools to create more energy efficient data centers, more powerful lasers to create faster optical interconnects and online agents that are paid for making accurate predictions. These are some of the 41 academic projects from 34 institutions HP Labs will back as part of its first annual Innovation Research Awards announced Thursday (Aug. 14).
HP Labs researchers waded through a flood of more than 450 proposals it received from 200 universities in 28 countries wanting to be part of the new program. It will provide funding of up to $100,000 for one year to each academic institution, renewable for a total of three years.
"I figured we might get only have that many submissions because we only had six weeks" to get responses to a call for proposals, said Rich Friedrich, director of the HP Labs group that selected the winners. "It wasn't just the quantity but the quality--it was hard selecting just 41," he added.
The program is one of several new initiatives at the storied R&D group under Prith Banerjee, hired a year ago to help revitalize the HP Labs.
"Deepening HP Labs' strategic collaboration with those in academia, government and the commercial sector ensures HP's research endeavors result in high-impact research that meets the scientific and business objectives of HP and its partners," said Banerjee, director of HP Labs speaking in a press statement.
About four of the projects focus on quantum devices seen as one of the possible successors to semiconductors. In one of those projects, John Bowers of the University of California at Santa Barbara will use CMOS to design a hybrid silicon ring resonator laser that could power a tiny, powerful laser for next generation optical interconnects.
Many of the projects will address the needs of tomorrow's data centers.
A researcher from UC Berkeley is developing a tool to model energy efficiency of data centers. Separate projects at Georgia Tech and the University of Michigan will explore automated tools to manage tomorrow's super-sized data centers.
"In the not-too-distant future, we could have 20,000 servers in a data center each one running multiple virtual machines," said Friedrich. "You could wind up with two million software objects to manage, and a human operator can't comprehend all that information," he said.
An MIT researcher will take automated Internet tools one step further, exploring the possibility of "prediction economies," networks of people and computers paid for accurate predictions of future events.
The program taps a wide range of international researchers spanning academics from nearby Stanford as well as more far flung locales including the Russian Academy of Sciences, Indian Institute of Technology and China's Tsinghua University.