IBM leads Top 500 supercomputer ranking, Linux and Intel dominate list

Mannheim (Germany) - Among the world’s top 500 fastest performing supercomputers, an IBM PowerPC 440 model dubbed "BlueGene/L" operated by the US Department of Energy’s Lawrence Livermore Laboratories claimed the top spot for the second straight year, the University of Mannheim and its partners revealed today.

IBM PowerPC processors claimed 25 spots on the university’s 25th annual Top500 list, including ranks 1, 2 (another BlueGene model operated by IBM’s Thomas J. Watson Research Center), 5, 6, 8, and 9. However, Intel dominated the list, claiming 333 of the 500 spots, the university’s database shows. Among these, 254 are driven by Xeon processors, the highest performing model in the 20th spot, dubbed "Tungsten," a cluster of 2500 dual-processor Dell PowerEdge 1750 systems running Red Hat Linux. 79 of the Intel spots were claimed by Itanium systems, the highest ranking being ranked in seventh place, dubbed "Thunder," a 1024 node cluster of Tiger4 1.4 GHz Itanium2 servers, also run by Livermore Labs. Thunder slipped two spots from its fifth position last year.

A single conventional Intel supercomputer, a 1999 model Paragon dubbed "ASCI Red," operated by Sandia National Laboratories, claimed 139th place.

The performance rankings, published annually by the University of Mannheim in association with Berkeley National Laboratory and the University of Tennessee, Knoxville, is obtained by dividing the maximal observed performance of each computer, by the theoretical peak performance of that computer - essentially, how well it does perform divided by how well it can perform - to obtain a relative "site efficiency" percentage.

Today’s supercomputers are prestigious not only for how fast they can do what they’re programmed to do, but for what they can accomplish. A few years ago, the most prestigious supercomputers played a darn good game of chess against Gerry Kasparov. Today, IBM’s BlueGene series is devoted to processes like projecting a three-dimensional model of the chemical constituency and shape of proteins, as it is believed that a protein’s shape may be a key determinant of its function in living organisms. Three-dimensional protein models such as the type projected by BlueGene computers are useful today in simulations that anticipate drug and chemical interactions with proteins, for the production of safer and more effective pharmaceuticals.

It’s worth noting that a database search through the Top500 list only showed Windows in association with rank 326, a PowerEdge 2650 Xeon P4 cluster operated by Cornell University. All the other top performers mentioned here except for ASCI Red were running Linux.