Sunday, February 5, 2012

The Top 10 Supercomputers

The twice-a-year list of the Top 500 supercomputers documents the most powerful systems on the planet. Many of these supercomputers are striking not just for their processing power, but for their design and appearance as well. Here’s a look at the top finishers in the latest Top 500 list
K SUPERCOMPUTER, RIKEN Advanced Institute for Computational Science (AICS), Japan
The “K” supercomputer, a joint project by Fujitsu and the RIKEN center, has overtaken China’s Tianhe-1A system to gain the top spot on the updated list. The K computer system, currently in the configuration stage, features 68,544 Sparc 64 CPUs, each with eight cores, for a total of 548,352 cores—almost twice as many as any other system in the TOP500. The K Computer is also more powerful than the next five systems on the list combined. The system is deployed across 672 computer racks and has a power consumption of 9.89 megawatts. The K Computer’s name draws upon the Japanese word “Kei” for 10^16 (ten quadrillions), representing the system’s performance goal of 10 petaflops.
TIANHE-1A, National Supercomputing Center, Tianjin, China

The Tianhe-1A, the most powerful supercomputer in the world as of November 2010.
The Chinese Tianhe-1A system at the National Supercomputer Center in Tianjin slipped back into second in the latest survey, achieving a performance level of 2.6 petaflop/s (quadrillions of calculations per second). Tianhe-1A is one of the 19 systems in the Top 500 that use NVIDIA GPUs (graphics processing units) to accelerate computation, up from 17 in the November survey.
JAGUAR, Oak Ridge National Laboratory
Jaguar was the top performer in the June 2010 survey, but fell back to second in November 2010 and now slips to third overall with a 1.75 petaflop performance speed running the Linpack benchmark. Jaguar is a Cray XT5 system located at the Department of Energy’s Oak Ridge Leadership Computing Facility in Tennessee. While Jaguar fell a few flops short this year, it retains a leadership position in supercomputing style with its striking Jaguar motif.

Nebulae, which is located at the National Supercomputing Centre in Shenzhen, China, achieved 1.271 PFlop/s to rank fourth overall, down one position from the November survey. Nebulae was built from a Dawning TC3600 Blade system with Intel X5650 processors and NVIDIA Tesla C2050 GPU.

TSUBAME 2.0, GSIC Center, Tokyo Institute of Technology

Like the top-ranked Tianhe system, Tsubame 2.0 is a successor system that builds upon the design of a previously-ranked system. Tsubame 2.0 was developed by the Tokyo Institute of Technology in collaboration with NEC and HP, and is powered by more than 1,400 nodes using both HP Proliant servers and NVIDIA Tesla GPUs. It is Japan’s highest-ranked supercomputer. Plans are being developed for Tsubame 3.0.

CIELO, Los Alamos National Labs
The new supercomputer named Cielo, the Spanish word for sky, will support all three national laboratories at the National Nuclear Security Administration (NNSA), including Los Alamos, Sandia, and Livermore. Cielo is the next generation capability class platform for the Advanced Simulation and Computing (ASC) program. Cielo’s architecture is based on Cray’s next generation “Baker” architecture with AMD’s new Magny-Cours processor, Cray’s “Gemini” high-speed interconnect and Compute Node Linux operating system.
PLEIADES, NASA Ames Research Center
One of the big movers n the latest survey is Pleiades, a supercomputer at the NASA Ames Reseach Center in Mountain Vew, Calif. It achieved a LINPACK benchmark of 1.09 petaflop/s), Since last June, NASA has implemented a series of expansions to the system’s performance capabilities, adding 14 new SGI Altix ICE 8400 systems so that Pleiades now contains 23,296 Intel Xeon quad- and hex-core processors (111,104 cores in 182 racks). Pleiades is used to meet the computing needs on NASA’s most demanding modeling and simulation projects in aeronautics; Earth and space science; exploration systems and technologies; and future space operations.
HOPPER, NERSC at Lawrence Berkeley National Labs
Hopper is named for American computer scientist Grace Hopper, and now powers science research at the National Energy Research Scientific Computing Center’s (NERSC) at Lawrence Berkeley National Labs. Hopper is powered by the Cray XE6 system. A pioneer in the field of software development and programming languages, Hopper created the first compiler. The Hopper system clocked in at 1.05 petaflop/s.
TERA-100, CEA, France

Tera-100 is now Europe’s most powerful supercomputer. The system resides at the Commissariat a l’Energie Atomique (CEA), where it supports the French nuclear weapons simulation program. era 100 consists of 4,300 bullx S Series servers, which were introduced by Bull in April 2010. It features 140,000 Intel Xeon 7500 processing cores, 300TB of central memory and a total storage capacity of over 20PB.
ROADRUNNER, Los Alamos National Laboratory
When the Roadrunner system at Los Alamos first appeared at the top of the June 2008 TOP500 list, it was the world’s first supercomputer to achieve a top performance of more than 1 petaflop/s (1015 floating point operations per second). It has now slipped to seventh place in the latest survey.