Computing Sciences Demonstrates Latest Scientific Computing Tools at High-Performance Computing and Networking Conference

When SC98, the annual conference on high-performance computing and networking, convenes Nov. 7 in Orlando, computational scientists from the Department of Energy's Lawrence Berkeley National Laboratory will be out in force.

During the week-long conference of the nation's leaders in computing and networking, members of Berkeley Lab's Computing Sciences organization will be presenting leading-edge computer simulations, demonstrating the latest tools for enhancing scientific research, and sharing their expertise with hundreds of other attendees by leading tutorial sessions. The Laboratory is home to the National Energy Research Scientific Computing Center (NERSC), which serves researchers at national labs, universities and industry, and the Information and Computing Sciences Division, which conducts computing and networking research.

"With the combination of these two successful programs, Berkeley Lab has truly emerged as one of the nation's leading centers for computational science," said C. William McCurdy, associate laboratory director for Computing Sciences. "And our leadership in the field is reflected in the program for SC— we're leading tutorials, demonstrating the latest technologies, highlighting our collaborations with other research organizations, and bringing home one—if not two—of the top computing honors for the year."

Phillip Colella, a mathematician and leader of the Applied Numerical Algorithms Group at NERSC, has been named as the recipient of the IEEE Computer Society's 1998 Sidney Fernbach Award, given each year to one person who has made "an outstanding contribution in the application of high performance computers using innovative approaches." He will receive the award during the conference.

NERSC physicist Andrew Canning is a member of a team of scientists from Oak Ridge National Laboratory and NERSC which is a finalist for the Gordon Bell Prize, given for best accomplishment in high-performance computing. The team, which also includes collaborators at the Pittsburgh Supercomputing Center and the University of Bristol (UK), performed a 1,024-atom first-principles simulation of metallic magnetism in iron which ran at 657 Gigaflops (billions of calculations per second) on a 1024-processor Cray/SGI T3E supercomputer.

Representatives from Berkeley Lab will be leading in-depth tutorials in the emerging fields of large-scale, data-intensive computing, an introduction to Message Passing Interface (the standard for writing scalable applications), and parallel programming of industrial applications. Additionally, four of the 48 technical papers accepted for the conference are by authors who either work or are affiliated with Berkeley Lab.

Also, Berkeley Lab researchers will be organizing and staffing a display for DOE2000, a Department of Energy program to develop computer applications allowing scientists at DOE sites across the country to remotely use scientific facilities and collaborate on-line to conduct experiments and research.

Among the demonstrations scheduled by Berkeley Lab are: Semilocal strings: a computer model, in striking detail, of a possible state of the universe only a hundred billionth of a trillionth of a trillionth of a second (10-35 second) after the Big Bang. In 3-D computer, objects called "semilocal strings" condense out of interacting quantum fields to form writhing tubes of energy.

Climate modeling: A partnership between climate modeling experts at the National Center for Atmospheric Research in Colorado and NERSC has resulted in more comprehensive models running 28 times faster than before. The work, funded by Department of Energy and the National Science Foundation, is aimed at investigating the effect of greenhouse gas increases and sulfate aerosols on global warming.

The Human Genome and Computational Biology: As the Human Genome Project continues to decipher the secrets of our genetic makeup, scientists around the world are gaining new insight into our understanding of the biology of health and of disease. Berkeley Lab scientists are developing new ways to predict protein folding, a key step in designing new drugs and understanding genetic diseases.

Radiance: A computer application which allows users to model different lighting conditions in buildings—and see how the light would change as they moved through the room. Radiance is part of the Virtual Building Laboratory, a project which aims to use energy simulations and 3-D visualization technology to allow scientists to test building performance and to examine and test materials and designs in virtual real-time, bypassing reducing the need for expensive laboratory experiments and avoiding design mistakes that later might require costly building retrofits.

NetLogger: NetLogger is a tool for diagnosing problems in networks and in distributed systems code and is being adopted by researchers at other national laboratories. The toolkit allows users to monitor exactly what is happening inside a distributed application - from the time a request for data is sent and received to the time the data are starting to be read, the point at which the read is completed, and the time when processing begins and ends.

The Energy Sciences Network, or ESnet: ESnet, is a high-speed network serving thousands of Department of Energy researchers and collaborators worldwide. Managed and operated by the ESnet staff at Lawrence Berkeley National Laboratory, ESnet provides direct connections to more than 30 DOE sites at speeds up to 622 megabits per second.

Turbulent flow in three dimensions: By its very nature, turbulence is a constantly changing problem involving complex flows. Although fluid flow and resulting turbulence are central to many scientific and engineering problems, such as building more efficient and less polluting internal combustion engines, it's a field that is difficult to understand and even harder to accurately simulate using a computer. In a parallel simulation of a three-dimensional turbulent fluid jet, the Center for Computational Sciences and Engineering at Lawrence Berkeley National Laboratory demonstrates one method of solving a hyperbolic computational physics problem.

Subsurface flow modeling: From groundwater contamination to increasing the flow from oil and natural gas fields, understanding the movement of liquids and gases in the subsurface is essential - and computer simulations give scientists insight into otherwise inaccessible regions. In the Earth Sciences Division at Lawrence Berkeley National Laboratory, earth scientists have developed a code for simulating multiphase flow and transport processes in fractured-porous media. Called TOUGH2, the code can model one-, two- and three dimensional flows of multiple phases, such as gas, aqueous liquids and oil, and multiple components, such as water, air, organics and radionuclides.

Finally, Saul Perlmutter, leader of the international Supernova Cosmology Project and a member of the Center for Particle Astrophysics based at Berkeley Lab, will be an invited speaker at SC98. In his talk, "Supercomputing and the Fate of the Universe," Perlmutter will discuss how he used computer simulations to confirm conclusions from observing supernovae that the universe will continue to expand, rather than eventually collapse, as some scientists have predicted.