University of Washington researchers have come up with a way to reduce energy consumption in computers and mobile devices by 50 percent or more by segmenting software code into areas that require high levels of accuracy — and therefore high levels of energy — and those that don’t.
The project, dubbed Enerj, has the potential to reduce energy bills, improve battery life and decrease the size of devices by requiring smaller batteries. The project, from the UW Computer Science & Engineering Department is slated to be presented next week in San Jose, Calif., at the Programming Language Design and Implementation annual meeting.
“Our long-term goal would be 10 times improvement in battery life,” says Luis Ceze, a UW assistant professor of computer science and engineering, in a UW article on the research. “I don’t think it is totally out of the question to have an order of magnitude reduction if we continue squeezing unnecessary accuracy.”
The key is to create a clear and solid barrier between the high- and low-energy processes, avoiding any intermingling. Examples of processes that could require less accuracy or double-checking could include streaming audio and video. Engineering writer Hannah Hickey explains in the UW article that the approach “would work like a dimmer switch, letting some transistors run at a lower voltage. Approximate tasks could run on the dimmer regions of the chip.”
Much more on the technology in this paper by the researchers: PDF, 11 pages.