October 29, 2010:
China has taken the lead in the supercomputer (the fastest computers on the planet) race, having built a 155 ton system using 7,168 GPUs (Graphic Processing Units, from high end graphic cards) and 14,336 CPUs to achieve peak performance of 2.507 petaflops (a petaflop is one million billion floating point operations per second, otherwise known as FLOPS). Sustained speed is 563 teraflops (one thousand billion). The Chinese Tianhe-1A supercomputer cost $88 million, requires 4.04 magawatts of power and occupies 1,000 square meters (10,764 square feet). Given that China manufactures a growing share of electronic items, and Chinese students have long dominated electronic engineering departments in American universities, this development should come as no surprise. Tianhe-1A will be used for civilian (oil exploration) and military (aircraft and missile design) applications.
Meanwhile, DARPA, the U.S. Department of Defense advanced research organization, has asked computer hardware developers to come up with a very powerful supercomputer (speed of one petaflop), that is small enough to fit into a cabinet 61x198x102 cm (24x78x40 inches) and require no more than 57 kilowatts to operate (including cooling). This ExtremeScale supercomputer would be flown out to combat zones, run off generator power and perform analysis of images and other data, to determine where the enemy is and what they are up to. This sort of predictive analysis has become a major weapon in the last decade, and it needs more computer power to be even more useful. There are currently portable PC cards that will goose a PC up to 20 teraflops (a thousand billion FLOPS). Currently, the most powerful PC can do 50 gigaflops (billion FLOPS). That, in turn, is faster than the fastest supercomputers of the early 1990s. In this area, progress isn't marching on, it's sprinting.
If the ExtremeScale supercomputer existed right now, only five supercomputers on the planet would be faster. But DARPA does not expect the ExtremeScale supercomputer to be ready for action for eight years. By then, one petaflop would be lucky to be in the top hundred supercomputers.
The military has long been a major user of supercomputers. Actually, these machines were first developed, as were the first computers, for military applications. These ultra-powerful computers are used for code breaking, and to help design weapons (including nukes) and equipment (especially electronics). The military is also needs lots of computing power for data mining (pulling useful information, about the enemy, from ever larger masses of information.)
Because there's never enough money to buy all the super-computers (which are super expensive) needed, military researchers have come up with ways to do it cheaper. A decade ago, it was military researchers who figured out how to use GPUs for non-graphic computing. GPUs do something similar to what supercomputers do (lots of math calculations of a fairly simple type), and eventually the manufacturers of GPUs realized that there was a commercial (not just military) demand for GPUs serving as supercomputers.
Three years ago, the Tesla supercomputer add-on for PCs appeared on the market. This was basically an Nvidia graphics board tweaked to act like a supercomputer, rather than a device that put 3-D, photo-realistic game graphics on your computer screen. The latest version of this system will give you a PC with four teraflops of computing power for under $10,000. The Tesla now has competition, as demand for these PC supercomputers has been huge since they were first introduced. The Chinese used Nvidia GPUs for Tianhe-1A.
DARPA is encouraging supercomputer development, with cash and the promise of military sales, and the industry is responding. Several firms believe they can have the ExtremeScale supercomputer in service way before 2018. In addition, lesser (.5 petaflop or betters) supercomputers will probably be available in a few years. And they will be cheap, costing a few hundred thousand dollars each. The current generation of senior commanders grew up with PCs. They understand what computers can do, they know that whoever has the most petaflops, will more likely prevail in future conflicts.