IBM tops list of world's fastest supercomputers

For the fourth year in a row, the Blue Gene/L system at California's Livermore National Laboratory has been named the world's fastest supercomputer.

For the fourth year in a row,a Blue Gene/L system at California's Livermore National Laboratory has been named the world's fastest supercomputer.

Blue Gene/L systemsclaimed four of the top 10 spots in the rankings released Wednesday at the International Supercomputer Conference in Dresden, Germany. Two of the top 10 are at Livermore.

The rankingswere releaseda day after IBM introduced its new generation supercomputer, the Blue Gene/P. The company said it breaks the "petaflop" speed barrier, making it three times faster than today's top-ranked system. It is not yet in operation.

Floating point operations per second, or "flops,"is a measure of computer performance using floating pointnumbers instead of integers. Floating point numbers are numbers where the decimal point position is a separate piece of data within the number. This allows for a wider range of data to be stored, but also means computers take longer to perform calculations than they would with simple integers.

A petaflop equals one-quadrillion floating point operations per second, or 100,000 times faster than a home computer. Sun Micro Systems also announced that its Constellation system has exceeded a petaflop.

For Wednesday's rankings, the fastestLivermore computer performed at 280.6 teraflops (trillions of floating operations per second). Thiscompares toits nearest rivals —the Cray XT4/XT3 at the Department of Energy's Oak Ridge National Laboratory in Tennessee at 101.7 TFlop/s and New Mexico's Sandia National Laboratory's Cray Red Storm system at 101.4 TFlop/s.

A new Blue Gene/P supercomputer. The system will be capable of up to three thousand trillion calculations per second, IBM says. ((IBM))

TheTop500 survey authorshighlighted how quickly the supercomputer field is advancing in speed.

"The system ranked No. 500 on the current list would have held position No. 216 only six months ago when the last list was produced," the announcement said. "This is the largest turnover rate between lists in the Top500 project's 15-year history."

Ten of the Top500 supercomputerswere in Canada. The University of Sherbrooke'sPowerEdge SC1425 3.6 GHz- Infiniband, made by Dell, claimed the top Canadian spot, 177th place.

The U.S. remains the leading consumer of high performance computer systems with 281 of the 500 systems ranked. The European share is growing (127 systems, up from 95) and remains larger than the Asian share (72, down from 79).

World's Fastest Supercomputers


1. Livermore National LabU.S.Blue Gene/L (IBM)


2. Oak Ridge National LabU.S.Jaguar - Cray XT4/XT3 (Cray)101.7
3. Sandia National LabU.S.Red Storm - Sandia/ Cray Red Storm, Opteron 2.4 GHz dual core (Cray)101.4
4. IBM Thomas J. Watson Research CenterU.S.
BGW - eServer Blue Gene Solution (IBM)
5. Stony Brook/BNL, New York Center for Computational SciencesU.S.New York Blue - eServer Blue Gene Solution (IBM)82.1
6. Livermore National LabU.S.ASC Purple - eServer pSeries p5 575 1.9 GHz (IBM)75.7
7. Rensselaer Polytechnic InstituteU.S.eServer Blue Gene Solution (IBM)73.0
8. National Center for Supercomputing ApplicationsU.S.Abe - PowerEdge 1955, 2.33 GHz, Infiniband (Dell)62.6
9. Barcelona Supercomputing CenterSpainMareNostrum - BladeCenter JS21 Cluster, PPC 970, 2.3 GHz, Myrinet (IBM)62.6
10. Leibniz RechenzentrumGermany
GermanyHLRB-II - Altix 4700 1.6 GHz (SGI)
Source: Top500

Supercomputers are used to quickly process huge amounts of data for diverse projects such as space research, weather forecasting, nuclear simulations and molecular modelling.

The Top500 list, which isclosely watchedin the industry, is compiled by Hans Meuer of the University of Mannheim, Germany,Erich Strohmaier and Horst Simon of NERSC/Lawrence Berkeley National Laboratory and Jack Dongarra of the University of Tennessee, Knoxville.

The project measures the speed of computers using the Linpack benchmark as a measure of a computer's floating-point rate of execution. It is determined by running a computer program that solves a dense system of linear equations. There are three benchmarks included in the Linpack report.