It takes a coordinated team, in fact, a very large team of talented, schooled, adequately funded and dedicated people to develop a supercomputer to the pinnacle of ‘fastest in the world.’ This is the point that should not be overlooked in the headlines about China’s Tianhe-1A supercomputer – soon to be anointed as the top computer in the world.

That China would develop a top supercomputer is not a surprise, although it looked like another machine, the Nebulae supercomputer based in Shenzhen would be the first to grab the honors. Tianhe-1A, developed by the China’s National University of Defense Technology and located in the National Supercomputing Center in Tianjin (port city for Beijing) made the leap from sixth on the Top 500 list (compiled by Jack Dongarra at the University of Tennessee, USA) over the top U.S. entry, the Cray Jaguar system. (The Top 500 list is issued twice a year, the next due on November 1, 2010.)

By the numbers:
Sustained performance: 2.507 petaflops (thousand trillion floating point calculations per second), compared to the U.S. Jaguar at 1.75 petaflops.
Processors: 7,168 Nvidia (USA) Tesla M2050 graphics processing units and 14,336 Intel (USA) six-core Xeon CPUs.
Compared power: Equivalent to 175,000 laptops.
Theoretical performance: 4.669 petaflops (at full graphics capability).

At one level, development of the world’s fastest supercomputer is a race between the United States, China, and Japan although Europe, India and other countries also participate. The race does have a point: The faster the supercomputer, the more intensive the computing problems it can solve. This capability can be used for a wide range of real-world applications including meteorology, national defense, nanotechnology, high-finance, scientific modeling, oil and gas exploration, and product development. At another level, the capability to produce a top supercomputer is an indication of the depth, wealth, and coordination of a large scientific endeavor – in short, a very exclusive club. National pride is involved.

The ‘race’ to develop the world’s fastest supercomputer is remindful of the ‘space race’ of not too many years ago – with just a few more players. It was once the almost sole domain of the United States, but in a way the advent of the microprocessor changed the design of supercomputers completely, and made it possible for worldwide competition.

Once the top level of supercomputing is achieved, the issue becomes one of utilization. Speed alone does not make the supercomputer useful. The skills that combine hardware and especially software to achieve super high speeds are not the same as interfacing those capabilities with real-world applications. However, one follows the other. Computer scientists who have learned how to make software integrated thousands of processors come from the same schools and basic training as the applications programmers. That is the depth that a country’s computer science infrastructure must provide, if the moniker of ‘world’s fastest computer’ is to represent more than a temporary acclaim.