What's beyond dynamic switching on and off of separate idle server components for efficient energy use?

A team of scientists from several US universities has proposed to redesign a server CPU to give a separate power feed to the chip's memory controller. Theoretically, this would enable the rest of the chip to be powered down when idle, while only the controller stays on to make the unit quickly accessible if needed.

The idea is one of several proposals for a new Google-funded research project aiming to develop new "low-power modes" in servers, whereby the maximum possible amount of server components gets powered down when the machine is idling.

Earlier this week, the search giant awarded $1 million to the project, a collaboration of computer scientists from University of California Santa Barbara, Rutgers University, University of Michigan and University of Virginia, according to a Rutgers statement. The two-year grant may be extended by one more year and an additional $500,000.

The team of researchers will work with UCSB's Greenscale Center for Energy-Efficient Computing.

"Greenscale will provide critical infrastructure to the project with the planned construction of the Greenscale Experimental Datacenter, a state-of-the-art miniature data center where systems researchers can conduct radical experiments not possible in production data centers," Fred Chong, professor of computer science at UCSB and the center's director, said in a statement.

Emily Wood, a Google spokesperson, said the project was "one of six looking into energy efficiency and it was the largest one." The company issued grants to a total of 12 research projects this week. While Google has been giving out research grants in the past, at a rate of about 150 grants per year, the grants it awarded this week were much larger than the ones issued previously.

The projects are led by 31 professors at 10 universities.

This week's awards, representing the first round of the company's new Google Focused Research Awards, totaled $5.7 million. Grants were awarded to scientists working in one of four areas: machine learning, the use of mobile phones as data collection devices for public health and environment monitoring, energy efficiency in computing, and privacy.