The International Conference for High Performance Computing, Networking, Storage and Analysis
High Performance Computations Enable Scientific Discoveries: Adapting Computational Physics Codes to New Hardware.
Authors: James F. Amundson (Fermi National Accelerator Laboratory), Philippe G. Canal (Fermi National Accelerator Laboratory), Donald J. Holmgren (Fermi National Accelerator Laboratory), Jim B. Kowalkowski (Fermi National Accelerator Laboratory), Qiming Lu (Fermi National Accelerator Laboratory), Marc Paterno (Fermi National Accelerator Laboratory), Saba Sehrish (Fermi National Accelerator Laboratory)
Abstract: Physics research as we know it today would not be possible without computers. Evaluating fundamental physical theories at the smallest and largest physical scales requires the production of enormous quantities of numerical data. Designing, building, and commissioning highly complex accelerators and detectors requires the simulation of enormous event samples similar to those collected for an experiment. Breakthroughs in physics research depend on the use of leading-edge computing technologies. The Fermilab Scientific Computing Division together with university and laboratory collaborators is advancing the use of state-of-the-art computing platforms that include GP-GPUs and Blue Gene to enable scientific discoveries by evolving the software of four application toolkits. Here we show the purpose, goals, and progress of this work within accelerator modeling (Synergia), computational cosmology (ART - Adaptive Refinement Tree), detector simulation (Geant4), and LQCD.