And the Race is On: China’s Tianhe-1A Supercomputer Knocks the Cray XT5 Jaguar Out of 1st Place
First there was the nuclear arms race of the 1950s. The space race followed quickly in the 1960s. While these nationalistic competitions continue today in 2010, the real race of the 21st Century may be the raw computing power to produce intelligent systems.
About a year ago H+ reported on how IBM’s supercomputer nicknamed “Roadrunner” was knocked off the top perch by the Cray XT5 Jaguar. The Chinese took the number five spot with their Tianhe-1 (“River in Sky”) system installed at the National Super Computer Center in Tianjin, China. Today’s announcement of the Tianhe-1A successor to the Tianhe-1, however, is causing ripples throughout the web: “China Wrests Supercomputer Title From U.S.” reports the New York Times. “China’s Tianhe-1A Now World’s FASTEST Supercomputer, Trumps U.S. Machine” shouts the Huffington Post headline. China’s Tianhe-1A is now number one.
Ironically, the Tianhe-1A relies on a massively parallel architecture powered by Intel Xeon processors and AMD GPUs used as accelerators, but “… the secret sauce behind the system – and the technological achievement – is the interconnect, or networking technology, developed by Chinese researchers that shuttles data back and forth across the smaller computers at breakneck rates,” states the NYT article quoting University of Tennessee computer scientist Jack Dongarra. The Tianhe-1A was benchmarked by a U.S. national laboratory at 2.5 x 1015 floating point operations per second (petaflops), roughly 1.4 times the speed of the Jaguar.
The race to develop faster and faster supercomputers is on. Not that there isn’t a continuing arms race – the Arms Control Association continues to watch the developments in North Korea and Iran – and the space race is still alive and well, thank you. But, don’t imagine for a minute that the U.S. Pentagon isn’t concerned about the implications of the Chinese leading the development of fast supercomputers. Wall Street – along with your 401k and IRA – is powered by increasingly intelligent supercomputers. And cybersecurity and cyberwarfare are undeniably part of daily presidential briefings.
At the beginning of the year AI researcher Dr. Ben Goertzel raised the question in a fascinating H+ article: “… will the Singularity be launched in China?”
Goertzel quotes his colleague and friend Dr. Hugo de Garis, “China has a population of 1.3 billion. The U.S. has a population of 0.3 billion. China has averaged an economic growth rate of about 10% over the past 3 decades. The U.S. has averaged 3%. The Chinese government is strongly committed to heavy investment into high tech. From the above premises, one can virtually prove, as in a mathematical theorem, that China in a decade or so will be in a superior position to offer top salaries (in the rich Southeastern cities) to creative, brilliant Westerners to come to China to build artificial brains…”
Goertzel appeared recently along with Ray Kurzweil in an episode of Dr. Michio Kaku’s Science Channel Sci Fi Science series called “A.I. Uprising.” During the program, Kaku considers Goertzel’s proposal for “a nanny artificial intelligence” that will keep track of a future Terminator-like rogue machine with a grudge against humanity. Kaku, however, doesn’t raise the question of whether the first artificial brains will be Chinese.
Dr. de Garis is well known to H+ readers as the author of The Artilect War: Cosmists Vs. Terrans: A Bitter Controversy Concerning Whether Humanity Should Build Godlike Massively Intelligent Machines. H+ editor-in-chief R.U. Sirius recently interviewed de Garis. In a nutshell, de Garis’ book argues that 21st century global politics will be dominated by question of whether to build “artilects” (artificial intellects, artificial intelligences, massively intelligent machines) that might dwarf human intelligence levels by a factor of trillions – in essence, a new species.
Supercomputers are currently being used as computing substrates to reverse engineer and perhaps go beyond the human brain – witness IBM’s cat brain and Dr. Henry Markram’s Blue Brain projects. The interesting question is whether it means anything for one nation – say the U.S. or China – to be the first to develop such a substrate. Does the whole notion of nationalistic “strategic advantage” disappear when a superior, networked successor “artilect” species suddenly appears?
The NYT’s announcement of the Tianhe-1A Supercomputer benchmark and the press reaction to it show that the race is on to build ever more powerful computing substrates. But whether the concept of a “nation” exists in another 50 years is another story…