Could pruning lesser used parts away from microchips make them faster, more energy efficient, and smaller? (Well, smaller is a given.) Computing experts from Switzerland, Singapore, and the United States believe so, and have estimated that this breakthrough idea could lead to a doubling of their overall efficiency.
Krishna Palem, the Ken and Audrey Kennedy Professor of Computing at Rice University in Houston, who also holds a joint appointment at Nanyang Technological University (NTU) in Singapore says: “I believe this is the first time someone has taken an integrated circuit and said, ‘Let’s get rid of the part that we don’t need.’ What we’ve shown is that we can boost performance and cut energy use simultaneously if we prune the unnecessary portions of the digital application-specific integrated circuits that are typically used in hearing aids, cameras, and other multimedia devices.”
The process takes into account errors that the stripped down microchips will inevitably make, and limits the calculations that create such errors. Palem and his team from Switzerland’s Center for Electronics and Microtechnology (CSEM) will demonstrate “probabilistic pruning” as a unique and possibly industry innovating concept at the DATE11 conference (March 14 – 18) in Grenoble, France. This exercise in “inexact hardware” is expected to cater to a market that’s developing toward a greener mindset. “Based on what we already know, we believe probabilistic computing can produce application-specific integrated circuits for hearing aids that can run four to five times longer on a set of batteries than current hearing aids,” Palem said. Christian Enz, from the CSEM arm of the collaboration (and co-author of the DATE study), says: “The cost for these gains is an 8 percent error magnitude, and to put that into context, we know that many perceptive types of tasks found in vision or hearing applications can easily tolerate error magnitudes of up to 10 percent.”