ARCHIVE

The BlueGene/L supercomputer at Lawrence Livermore National Laboratory topped the list of the world’s fastest computers for a record sixth straight time, according to the new Top500 list released Wednesday, June 27 at the International Supercomputing Conference in Dresden, Germany. Built by IBM, BlueGene/L (BG/L) clocks in at 280.6 teraFLOPS (trillion floating operations per second) on the LINPACK, the industry standard for supercomputer performance. BGL is a workhorse machine for the U.S. Department of Energy’s National Nuclear Security Administration’s (NNSA) effort to ensure the safety, security and reliability of the nation’s nuclear deterrent without underground nuclear testing, known as the Stockpile Stewardship Program. NNSA’s Advanced Simulation and Computing (ASC) program’s 100 teraFLOP Purple system, another IBM machine at Livermore lab, dropped from fourth to sixth on the Top500 list. “Since BG/L went into production in early 2006 it has performed beyond our expectations and delivered for the ASC program. BG/L’s architecture has proven suitable for a much broader range of applications than originally envisioned,” said Dona Crawford, associate director for Computation. “Likewise, ASC Purple also has demonstrated the system’s ability to deliver weapons simulations of unprecedented spatial resolution for the Stockpile Stewardship Program.” The detailed computer simulations of nuclear weapons performance produced by the ASC program using BG/L, ASC Purple and other supercomputers at the three nuclear weapons labs are a cornerstone of stockpile stewardship. ASC is a tri-lab program uniting the high performance computing expertise of NNSA’s Los Alamos, Lawrence Livermore and Sandia national labs. BG/L’s three year reign as the world’s fastest supercomputer has seen significant progress in code development and the achievement of numerous milestones for NNSA’s stockpile stewardship program. For example, simulations on BGL helped answer critical questions about plutonium aging – a key to understanding the life expectancy of nuclear weapons systems. Breakthrough calculations/simulations run on the machine have over the last two years garnered three Gordon Bell Prizes, widely regarded in the computing community as the Oscars of high performance computing. Simulations on BG/L of high explosives, super-ionic water, and graphite to diamond experiments have provided scientific insights and/or confirmed results of earlier physical experiments. Livermore scientists used BG/L to perform the first instability simulation with a Reynolds number large enough to determine the nature of turbulence beyond the mixing transition, a feat which made the cover of Nature Physics magazine in August 2006. In June 2006, BGL set a new world mark for a scientific application with a sustained performance of 207.3 teraFLOPS on the “Qbox” computer code for conducting materials science simulations. This represents a leap forward in scientists’ capability to perform predictive simulations of large, complex high-Z metals relevant to stockpile science and was awarded the 2006 Gordon Bell Prize for peak performance. ASC Purple played an important role in the development of the Livermore team’s Reliable Replacement Warhead (RRW) design which was selected by NNSA in a competition to design a replacement warhead for a portion of the country’s nuclear weapons stockpile. The modern design of RRW is needed to ensure the long-term reliability and increase the security and safety of the stockpile without underground nuclear testing. In January and February of 2006, a joint team of scientists from Lawrence Livermore and Los Alamos national labs performed a series of weapon simulations at unprecedented spatial resolution, utilizing the most advanced ASC simulation software. These simulations point to phenomena not seen at lower spatial resolutions and give new insight into weapons physics. Such detailed modeling was not practical on any previous computer architecture due to both time and memory constraint. But Purple, and its successors, will enable significant enhancements in both understanding and weapon simulation capability. Department of Energy (DOE) systems also hold the number two and three rankings on the list: the upgraded Cray XT4/XT3 at DOE’s Oak Ridge National Laboratory, ranked No. 2 with a benchmark performance of 101.7 TFlop/s; and Sandia National Laboratory’s Cray Red Storm system, which ranked third at 101.4 TFlop/s. DOE and NNSA laboratories house the four world’s fastest supercomputers. Appearing for the first time at No.19 on the Top500 list with 36.53 teraFLOP/s on Linpack is another Livermore machine dubbed Atlas; a 44.2 teraFLOP/s peak system dedicated to unclassified scientific computing through LLNL’s Multi-programmatic and Institutional Computing program. Atlas is doing the heavy lifting for “Grand Challenge” science collaborations in fields ranging from climate and earthquake simulations to astrophysics, chemistry, materials and bioscience. Grand Challenge allocations of time on Atlas are made available to science projects vital to DOE, NNSA and LLNL missions. Atlas is a Linux cluster built by the Milpitas-based company Appro. The Top500 list is updated and released twice a year; at the June International Supercomputing Conference (ISC) in Germany and at the November Supercomputing (SC) held at different locations in the US. For more information about supercomputer rankings and trends, check the Top500: www.top500.org/. Information about NNSA’s ASC program is available at: www.llnl.gov/asc/.