51,200-core system boosts NASA to No 3
NASA's Pleiades supercomputer is at No 3 in the latest Top500 list, thanks in part to a 51,200-core SGI Altix ICE 8200EX system from Silicon Graphics (SGI). With its sights set on colonizing the moon and eventually sending astronauts to Mars, NASA is calling on researchers to solve some of the most complex science and engineering problems in history.
Key to that effort is Pleiades, the world's third fastest supercomputer, installed at the NASA Advanced Supercomputing facility at NASA Ames Research Center in Mountain View, California. The Altix system is capable of generating a theoretical peak of 609 trillion operations per second (Teraflops), and demonstrated 487 Teraflop performance on the Top500 LINPACK benchmark. The results make Pleiades the world's most powerful general-purpose supercomputer.
Pleiades offers researchers unprecedented resources for a range of projects in support of all of NASA's mission directorates, though most of the work, at least initially, is expected to support the development of NASA's next-generation space fleet. Known as Project Constellation, the manned space exploration effort will involve years of sophisticated, high-fidelity scientific and engineering studies, from virtually testing re-entry vehicle options to designing safety systems.
Researchers will even use Pleiades to simulate catastrophic failures — specifically so they can design systems and procedures to prevent problems that might threaten the safety and survival of astronauts.
Pleiades supplements Columbia, the 14,336-core SGI Altix system that debuted in 2004 as the world's second-fastest computer. Columbia helped NASA successfully resume its Space Shuttle programme, while saving millions of hours of research time on many other projects.
'With Pleiades, we can do six times the work that we could on Columbia,' said Rupak Biswas, acting chief of the NASA Advanced Supercomputing (NAS) Division. 'Now our researchers are making their projects as large and complex as they need, without having to compromise simulation completeness or fidelity to make room or time for other projects. We're already seeing real productivity benefits that will help keep Project Constellation and other NASA research initiatives on schedule.'