11/18/2010

Lab is at the center of global supercomputing competition

Donald B Johnston, LLNL, (925) 423-4902, johnston19@llnl.gov



David Cooper and Mike McCoy at the Supercomputing 2010 (SC10) in New Orleans this week.

The release of the new Top500 List of the world's most powerful supercomputers puts Lawrence Livermore and other DOE/NNSA national laboratories in the thick of a global high performance computing competition.

The 36th edition of the Top500 List, the industry standard for high performance computing (HPC), was released Tuesday at Supercomputing 2010 (SC10) in New Orleans. The Top500 list appears twice a year in June and at the annual supercomputing conference in November.

China has emerged as a supercomputing power with HPC systems taking the No. 1 and No. 3  spots on the list with, respectively, the Tianhe-1A and Nebulae machines. The Tianhe-1A, built by China's National University of Defense Technology, achieved a performance of 2.5 petaFLOP/s (quadrillion floating operations per second) on the Linpack industry-standard benchmark.

"The Tiahne-1A is a remarkable technological achievement and we congratulate China," said Dona Crawford, associate director for Computation. "The new Top500 list shows that high performance computing is now truly a global enterprise."

Further underscoring the globalization of HPC was the appearance of supercomputers in Japan, France and Germany in the Top500's top 10. The Tokyo (Japan) Institute of Technology's Tsubame 2.0 is ranked No. 4 with a 1.19 petaFLOP/s performance. European and Asian supercomputers account for five of the top 10 HPC systems.

Europe's most powerful system is a Bull machine named Tera 100 built for France's Commissariat a l'Energie Atomique (CEA). Ranked No. 6, the system is dedicated to the French equivalent of the U.S. stockpile stewardship program.  Germany's JUGENE system BlueGene/P solution ranks ninth.

Despite the dramatic changes in the Top500 ranking, DOE/NNSA labs remain well represented. DOE/NNSA has 14 systems in the top 50 of the Top500. Oak Ridge National Laboratory's (ORNL) Cray XT5 Jaguar, dropped from No. 1 to No. 2 on the list with a performance of 1.75 petaFLOP/s. Kraken, a Cray XT5-HE system, owned by the University of Tennessee but located at ORNL is ranked No. 8. Lawrence Berkeley Laboratory's Cray XE6 Hopper system made its first appearance on the list at No. 5.

The IBM Roadrunner system at Los Alamos National Laboratory is now seventh on the list and the joint LANL/Sandia National Laboratory Cray XE6 Cielo system in New Mexico is ranked No.  10. For the first time in almost a decade, the Laboratory had no HPC systems in the top 10 of the new list. LLNL's BlueGene/L, which held the top ranking for seven lists between 2005 and 2008, is now No. 12.

However, LLNL is working with IBM to field Sequoia, a 20-petaFLOP/s system, scheduled for delivery in late 2011 and to go into production for NNSA's Advanced Simulation and Computing program in 2012.

"The globalization of high performance computing underscores the importance of the leadership role DOE/NNSA labs, in general, and Livermore in particular, play in advancing the state of the art in supercomputing," Crawford said. "What's important here is not just leadership in computing, but the science and technology leadership that computing power enables. We need to bring to bear our years of experience in HPC to help ensure our continued leadership in an ever larger, more competitive environment."

For more on the Top500, see the Web


Founded in 1952, Lawrence Livermore National Laboratory is a national security laboratory, with a mission to ensure national security and apply science and technology to the important issues of our time. Lawrence Livermore National Laboratory is managed by Lawrence Livermore National Security, LLC for the U.S. Department of Energy's National Nuclear Security Administration.