On cars, old, new and future; science & technology; vintage airplanes, computer flight simulation of them; Sherlockiana; our English language; travel; and other stuff
MY FELLOW tweeters (see @simanaitis) recently received one derived from my reading Turing’s Cathedral, George Dyson’s wonderful book about the post-WWII emergence of the digital age. What’s more, the October 5 issue of Science magazine, published by the American Association for the Advancement of Science, has a related news item that brings matters up to date.
To wit, how potent should a computer be?
The Science article discusses Japan’s K computer. Back in June 2011, it made headlines as the world’s fastest. And, about a year ago, it became the first computer to top 10 petaflops, that is, 10 quadrillion calculations per second. For those really into zeros, this is 10,000,000,000,000,000 of them.
The Bosch Automotive Handbook provides an interesting perspective to this. Its 8th English Edition, published in 2011, lists System International prefixes up through P, peta, 1015, and E, exa, 1018. By contrast, its 1st English Edition, published in 1978, stopped its SI prefixes at “only” T, tera, 1012.
Back to Japan’s K: According to Science, by the time it was available for general use in late September of this year, poor K had already lost its bragging rights.
In June 2012, the IBM Sequoia hit a top speed of 16.32 petaflops. The Sequoia supercomputer is built for the National Nuclear Security Administration at Lawrence Livermore National Laboratory in Livermore, California.
Then, only a few days ago, it was announced that the Titan supercomputer, a Cray XK7, raised the bar even more. This machine (actually, like the others, a linkage of hundreds of thousands of processors) has operated at a rate of 17.59 petaflops, with a theoretical speed of 20 petaflops.
The Titan is at the Oak Ridge National Laboratory, Oak Ridge, Tennessee, known historically for its atomic bomb research during World War II.
Here’s a not unrelated tidbit from Turing’s Cathedral: Particularly in the 1950s, one of five important missions for the era’s supercomputer, the MANIAC, was simulation of nuclear explosions.
The other four, in increasing longer time frames, were shock and blast waves, meteorology, biological evolution and stellar evolution. Together (and recalling SI prefixes), the five involved durations of time extending from 10-6 seconds (microseconds) to 1017 seconds (not quite, but almost exaseconds).
Back again to Japan’s K: According to Science, some critics say it was too narrowly designed for speed on a single benchmark problem, not for varied applications. It has been suggested darkly the project was primarily to benefit Japanese computer manufacturers, not to promote science.
Other scientists are pleased with K. For instance, researchers studying drug-protein binding would like all the speed they can get. Another group is simulating the formation and evolution of dark-matter halos, structures that envelope galaxies. There are already 227 applications for K’s first 62 time slots.
Last, that Turing’s Cathedral tweet? In 1953, MANIAC accomplished its work with a memory of 5 kilobytes.
Author George Dyson observes that 5 kilobytes of memory would support about a 1/2-second of today’s MP3-compressed audio. ds
© Dennis Simanaitis, SimanaitisSays.com, 2012