BlueGene/L applications: Parallelism On a Massive Scale

• Published in: The international journal of high performance computing applications 2008-02, Vol.22 (1), p.33-51
• Main Authors: de Supinski, Bronis R., Schulz, Martin, Bulatov, Vasily V., Cabot, William, Chan, Bor, Cook, Andrew W., Draeger, Erik W., Glosli, James N., Greenough, Jeffrey A., Henderson, Keith, Kubota, Alison, Louis, Steve, Miller, Brian J., Patel, Mehul V., Spelce, Thomas E., Streitz, Frederick H., Williams, Peter L., Yates, Robert K., Yoo, Andy, Almasi, George, Bhanot, Gyan, Gara, Alan, Gunnels, John A., Gupta, Manish, Moreira, Jose, Sexton, James, Walkup, Bob, Archer, Charles, Gygi, Francois, Germann, Timothy C., Kadau, Kai, Lomdahl, Peter S., Rendleman, Charles, Welcome, Michael L., McLendon, William, Hendrickson, Bruce, Franchetti, Franz, Kral, Stefan, Lorenz, Jürgen, Überhuber, Christoph W., Chow, Edmond, Çatalyürek, Ümit
• Format: Article
• Language: English
• Subjects: Analysis; Computer architecture; Computer simulation; Dislocations; Lists; Mathematical optimization; Microprocessors; Molecular dynamics; Optimization techniques; Processors; Scaling; Turbulence
• ISSN: 1094-3420; 1741-2846
• DOI: 10.1177/1094342007085025

BlueGene/L (BG/L), developed through a partnership between IBM and Lawrence Livermore National Laboratory (LLNL), is currently the world's largest system both in terms of scale, with 131,072 processors, and absolute performance, with a peak rate of 367 Tflop/s. BG/L has led the last four Top500 lists with a Linpack rate of 280.6 Tflop/s for the full machine installed at LLNL and is expected to remain the fastest computer in the next few editions. However, the real value of a machine such as BG/L derives from the scientific breakthroughs that real applications can produce by successfully using its unprecedented scale and computational power. In this paper, we describe our experiences with eight large scale applications on BG/ L from several application domains, ranging from molecular dynamics to dislocation dynamics and turbulence simulations to searches in semantic graphs. We also discuss the challenges we faced when scaling these codes and present several successful optimization techniques. All applications show excellent scaling behavior, even at very large processor counts, with one code even achieving a sustained performance of more than 100 Tflop/s, clearly demonstrating the real success of the BG/L design.