The prototypes cut energy use by allowing processing components – like hardware for adding and multiplying numbers – to make a few mistakes.
By managing the probability of errors and limiting which calculations produce errors, the designers – from Rice University, Singapore’s Nanyang Technological University, Switzerland’s Center for Electronics and Microtechnology (CESM) and the University of California, Berkeley – say they can simultaneously cut energy demands and boost performance.
One example of the inexact design approach is “pruning,” or trimming away some of the rarely used portions of digital circuits on a microchip. Another innovation, “confined voltage scaling,” takes advantage of improvements in processing speed to further cut energy demands.
In their initial simulated tests in 2011, the researchers showed that pruning some sections of traditionally designed microchips could boost performance in three ways: the pruned chips were twice as fast, used half as much energy and were half the size. In the new study, which earned best-paper honors at the ACM International Conference on Computing Frontiers in Cagliari, Italy this month, the team implemented their ideas in the processing elements on a prototype silicon chip.
Study co-author Avinash Lingamneni, a Rice graduate student, says the latest tests show pruning could cut energy demands 3.5 times with chips that allow for errors .25 percent of the time. After factoring in size and speed gains, these chips were 7.5 times more efficient than regular chips, Lingamneni said. Chips that got wrong answers with a larger deviation—about 8 percent—were up to 15 times more efficient, he said.
The human eye has a built-in mechanism for error correction, CSEM’s Christian Enz, a project co-investigator, said. Researchers found relative errors up to .54 percent were almost indiscernible, and relative errors as high as 7.5 percent still produced discernible images.
Project leader Krishna Palem, who also serves as director of the Rice-NTU Institute for Sustainable and Applied Infodynamics (ISAID), said likely initial applications for the pruning technology will be in application-specific processors, such as special-purpose “embedded” microchips, like those used in hearing aids, cameras and other electronic devices.
The inexact hardware is also a key component of ISAID’s I-slate educational tablet, designed for Indian classrooms with no electricity and too few teachers. Officials in India’s Mahabubnagar District announced plans in March to adopt 50,000 I-slates into middle and high school classrooms over the next three years, according to Rice.