Illustrations by Phil Couzens
We demand ever more from our gadgets. Phones are now expected to record and playback high-resolution videos, play music, surf the Internet and keep us reminded of what’s happening to our friends. The cars we drive every day carry, in a much-quoted (and little-understood) statistic, more computing power than the Apollo spacecraft carried to the moon, as they constantly monitor and correct driving and road conditions, and entertain us on the move.
All of these demands produce constraints on technology, and on the developers who bring this technology to fruition. For a long time, the makers of CPUs—the chips that form the “brains” of our devices—engaged in “clock wars,” whereby the speed of the processors constantly kept increasing, enabling them to do more work in a shorter amount of time.
But this leads to problems. Power consumption goes up, and this is no trivial matter in a battery-powered device. Heat also increases, and while it may be acceptable for an automotive computer located under the hood to run hot, it’s a completely different matter for a cellphone in your pocket or even a video player in your living room.
Two Heads Are Better Than One
Twenty-five years ago, Hironori Kasahara, now a professor at Waseda University, and Director of the Advanced Multicore Processor Research Institute, became interested in the technique of harnessing CPUs together. He was not alone in this, of course, with researchers around the world developing similar techniques (reportedly early versions of the Tomahawk cruise missile used a “cube” of 27 interconnected CPUs for terrain contour matching), but this pioneering work has borne fruit for Kasahara with the advent of multi-core chips.
