DOE Awards Record Supercomputing Time to UC San Diego, SDSC Researchers
Source Newsroom: University of California, San Diego
Newswise — Scientists from the San Diego Supercomputer Center (SDSC) and other areas of the University of California, San Diego, conducting research in physics, computer science, earth science, and engineering, together were awarded an all-time high of more than a quarter billion hours in supercomputing processor time by the U.S. Department of Energy (DOE) as part of the agency’s 2012 Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program.
SDSC and UC San Diego researchers involved in six separate projects were allocated a combined 261 million processor hours on DOE computer resources, up from 99 million hours for 2011. The awards are part of the DOE’s allocation of almost 1.7 billion processor hours for 60 research projects to be conducted in 2012 to address scientific and engineering challenges of national and global importance.
New and renewed projects involving SDSC and UC San Diego researchers who were awarded processing time under the latest INCITE awards include:
Astrophysics: “How High Redshift Galaxies Reionized the Universe.”
Principal Investigator: Michael Norman, Director, SDSC, UC San Diego.
Co-Investigators: Robert Harkness, SDSC, UC San Diego; and Daniel Reynolds, Southern Methodist University.
INCITE Allocation: 32,000,000 processor hours on Cray XK6 at Oak Ridge National Laboratory.
Research Summary: The goal of this project is to critically test whether the population of high redshift galaxies discovered by the Hubble Space Telescope is capable of reionizing the universe, consistent with observational constraints using self-consistent cosmological radiation hydrodynamical simulations. These will be the first fully self-consistent simulations of reionization at high enough resolution and in large enough volumes, and to engage multiple observations to critically test the dwarf galaxy reionizer hypothesis.
Earth Sciences: “CyberShake 3.0: Physics-Based Probabilistic Seismic Hazard Analysis.”
Principal Investigator: Thomas Jordan, Southern California Earthquake Center and University of Southern California.
Co-Investigators: Philip Maechling, University of Southern California; Jacobo Bielak, Carnegie Mellon University; Yifeng Cui, SDSC; Geoffrey Ely, University of Southern California Kim Olsen, San Diego State University; Ricardo Taborda, Carnegie Mellon University.
INCITE Allocation: 40,000,000 processor hours (32,000,000 on Cray XK6 at Oak Ridge National Laboratory, 2,000,000 on IBM Blue Gene/P at Argonne National Laboratory
Research Summary: This project will use INCITE resources to calculate an improved probabilistic seismic hazard forecast for California, following recent destructive earthquakes including those in New Zealand and Japan last year. The SCEC CyberShake 3.0 hazard model calculation will be based on a new USGS Unified California Earthquake Rupture Forecast 3.0 (UCERF3.0) scheduled for release by the USGS in June 2012. The project will also demonstrate to the seismic hazard community how HPC resources can improve official, broad-impact, USGS-regulated seismic hazard forecast data products.
Plasma Physics: “Unraveling the Physics of Magnetic Reconnection with 3D Kinetic Simulations.”
Principal Investigator: William Daughton, Los Alamos National Laboratory.
Co-Investigators: Vadim Roytershteyn, Los Alamos National Laboratory; Homayoun Karimabadi, UC San Diego.
INCITE Allocation: 37,000,000 processor hours on Cray XK6 at Oak Ridge National Laboratory.
Research Summary: Magnetic reconnection is a fundamental plasma physics process that converts magnetic energy into particle energy and plays a critical role in a variety of physical environments such as planetary magnetospheres, solar flares, laboratory fusion experiments, and astrophysical plasmas. With the advent of petascale computing, researchers can now for the first time conduct a systematic study to address some of the key issues in 3D kinetic reconnection, and perform simulations ~100x larger than was previously considered state-of-the-art. The results are expected to not only lead to a major advance in theoretical understanding of magnetic reconnection, but also have an impact in a variety of fields including space physics, solar physics, laboratory plasmas, and astrophysics.
Computer Science: “Performance Evaluation and Analysis Consortium End Station.”
Principal Investigator: Patrick Worley, Oak Ridge National Laboratory.
Co-Investigators: David H. Bailey, Lawrence Berkeley National Laboratory; Jack J. Dongarra, University of Tennessee; William D. Gropp, University of Illinois at Urbana-Champaign; Jeffrey K. Hollingsworth, University of Maryland; Robert F. Lucas, University of Southern California; Allen D. Malony, University of Oregon; John Mellor-Crummey, Rice University; Barton P. Miller, University of Wisconsin at Madison;
Leonid Oliker, Lawrence Berkeley National Laboratory; Allan Snavely, SDSC, UC San Diego; Jeffrey S. Vetter, Oak Ridge National Laboratory; Katherine A. Yelick, University of California at Berkeley; Bronis R. de Supinski, Lawrence Livermore National Laboratory.
INCITE Allocation: 28,000,000 processor hours (18,000,000 on Cray XK6 at Oak Ridge National Laboratory; 10,000,000 on IBM Blue Gene/P at Argonne National Laboratory.)
Research Summary: To maximize the utility of DOE leadership class systems, researchers must understand how to use each system most efficiently. To provide further understanding of these high-end systems, this proposal focuses on four primary goals: (1) to update and extend performance evaluation of all systems using suites of both standard and custom micro, kernel, and application benchmarks; (2) to continue to port performance tools and performance middleware to the leadership class systems; (3) to validate the effectiveness of performance prediction technologies, modifying them as necessary to improve their utility for predicting resource requirements for production runs on the leadership-class systems; and (4) to analyze and help optimize current or leadership class application codes.
Particle Physics: “Lattice QCD”
Principal Investigator: Paul B. Mackenzie, Fermi National Accelerator Laboratory
Co-Investigators: Richard C. Brower, Boston University; Norman H. Christ, Columbia University; Frithjof Karsch, Brookhaven National Laboratory; Julius Kuti, UC San Diego; John W. Negele, Massachusetts Institute of Technology; David G. Richards, Jefferson Laboratory; Stephen R. Sharpe, University of Washington; Robert Sugar, University of California, Santa Barbara.
INCITE Allocation: 96,000,000 processor hours (50,000,000 on IBM Blue Gene/P at Argonne National Laboratory; 46,000,000 on Cray XK6 at Oak Ridge National Laboratory.
Research Summary: This project will deepen scientists’ understanding of the interactions of quarks and gluons, the basic components of 99% of the visible matter in the universe, and play a key role in ongoing efforts to develop a unified theory of the four fundamental forces of nature. These fundamental questions in high energy and nuclear physics are directly related to major experimental programs and milestones set out by the Department of Energy’s Office of Science.
Engineering - Fluids and Turbulence: “Mixing in Incompressible and Compressible Turbulence.”
Principal Investigator: Pui-kuen Yeung, Georgia Institute of Technology.
Co-Investigators: Diego Donzis, Texas A&M University; Toshiyuki Gotoh, Nagoya Institute of Technology; Dmitry Pekurovsky, SDSC, UC San Diego; Katepalli Sreenivansan, New York University.
INCITE Allocation: 28,000,000 processor hours on Cray XK6 at Oak Ridge National Laboratory.
Research Summary: Turbulent mixing often plays a rate-limiting role in technological applications such as heat and mass transfer in industrial processes, pollutant dispersion, and the mixing of fuel with air in aircraft engines. This project is focused on producing detailed simulations to help answer long unresolved fundamental questions related to turbulence theory, while exploring further performance improvements in our already highly scalable codes.
Now in its ninth year, INCITE supports computationally intensive scientific investigations, allowing researchers at national laboratories, universities, and throughout industry to explore a wide range of scientific challenges. Many of the new and continuing INCITE projects aim to further renewable energy solutions and understand of the environmental impacts of energy use. The program, open to all scientists, is supported by the Department's Office of Science and managed by the DOE Leadership Computing Facilities at the Department's Argonne and Oak Ridge National Laboratories. For more information about the DOE Office of Science, visit: www.science.doe.gov.