‘Computational Sprinting’ could give mobile devices big bursts of speed


The smartphone you have in your pocket or purse is literally hundreds of times more powerful than the room-sized computers that landed Neil Armstrong and Buzz Aldrin on the moon. Hardware design has accelerated exponentially since then, roughly doubling every two years the number of circuits that can fit on a computer chip.

Yet as powerful as they are, these tiny computer chips are still bound by the laws of physics. The more electricity that passes through these circuits, the more heat that is generated, and the more likely they are to melt or burn. To deal with the problem, hardware designers have matched hotter chips with more elaborate cooling systems—powerful fans and even liquid nitrogen—but mobile devices don’t have the extra space these remedies require.

Computer scientists and engineers worry about this issue, dubbed “dark silicon,” after the larger and larger portions of mobile chips that must remain dormant at one time to prevent overheating.

But a research team led by Milo Martin of the School of Engineering and Applied Science has devised a solution. The team calls it “computational sprinting,” reasoning that mobile device chips only really need to run at top speed for short, intense bursts.

“What if we designed a chip to run at 16 times the sustainable rate, but only for half a second?  Can we do it without burning out the chip?” asks Martin. “We did the calculations and simulations, and found that it is indeed possible to engineer such a system.”

Martin, his students, and research partners at the University of Michigan propose that mobile devices could run on a single processor most of the time, and then engage up to 15 more processors when they need to do computationally intense tasks, such as speech recognition. This “sprint” would be able to return a fast response and then allow the device to cool down and recover.

To enable this recovery, the researchers suggest pairing the processers with an encased material—something similar to candle wax—that would absorb heat by melting during the sprint then slowly dissipate the heat by hardening while the device is waiting for the user’s next request.    

By matching hardware design with how these mobile devices are actually used, computational sprinting could keep smartphones racing toward the future.

Originally published on March 8, 2012