 |
By Karen Green
August marked the first time usage of a National Science Foundation
high-performance computer topped 1 million normalized CPU hours in one month.
According to figures from Quantum Research, which measures computer usage at
NSF-supported sites, NCSA's 1,536-processor SGI Origin2000 supercomputer provided
1,136,676 normalized CPU hours to 736 national users in August, tripling the usage
from August 1998.
NCSA's Origin is one of a large spectrum of computing resources available to the
national user community through the NSF's Partnerships for Advanced Computational
Infrastructure (PACI) program. The NSF PACI program defines a normalized CPU hour
as a Cray X-MP processor hour equivalent, based on standard benchmarks.
"The country is crossing a threshold in capability computing," said Larry Smarr,
director of NCSA and the National Computational Science Alliance, one of the two
PACI partnerships. "With the Origin, we are able to give unprecedented amounts of
dedicated time to 'super-users', while continuing to provide broad access to a
large community."
"By having uninterrupted access to the NCSA Origin 256-processor system for over a
week, we were able to carry out the most highly resolved cosmological simulation
ever done," said Jeremiah Ostriker, provost of Princeton University and head of the
Alliance Cosmology team. "With this level of computational resolution, new
cosmological tests of competing models of the universe against the observations are
now possible. Over a billion particles were followed as they moved under the
influence of gravity in a representative piece of the universe, the ratio of the
box size to resolution detail being approximately 75,000."
Researchers in cosmology used the most time on NCSA's systems in August, logging
nearly 400,000 normalized CPU hours. Chemists were also big users followed by
molecular biologists, materials researchers, physicists, and atmospheric
scientists. With the rapid increase in the size of the NCSA Origin, the 400,000
hours logged by cosmology researchers in August is larger than NCSA's total
monthly usage for the months before February 1999.
Among the top Origin users for August were the following.
Jeremiah P. Ostriker,
Princeton University (173,163 normalized CPU hours)
Ostriker, head of the Alliance Cosmology team, does numerical simulations of the formation and evolution of cosmic structure. His team looks at the consequences of various theories of cosmic structure formation across epochs, including formation of the first cosmic structures, recombination, reionization, and formation of galaxies, clusters, gravitational lenses, and large-scale structures. For more on his research,
see http://access.ncsa.uiuc.edu/CoverStories/StarLight/.
Michael L. Norman, University of Illinois at Urbana-Champaign (112,018 normalized CPU hours)
Another member of the Alliance Cosmology team, Norman conducts large-scale 3D simulations of the dynamics of molecular clouds, which are the sites of star formation in our galaxy. By using a technique called adaptive mesh refinement, Norman's team has been able to zoom in on developing molecular clouds that existed in the early universe and track their collapse into tight knots that gave birth to the first stars. A story on Norman's work will appear in the Fall/Winter edition of Access Magazine and in an upcoming issue of Access Online.
David Ceperley, University
of Illinois at Urbana-Champaign (90,521 normalized CPU
hours)
Ceperley is working to develop computational methods to accurately
predict the physical properties of quantum many-body systems from fundamental
equations. A member of the Alliance Nanomaterials team, he is applying these
methods to a wide range of problems in materials science, low-temperature physics,
quantum chemistry, surface science, astrophysics, and experimental physics.
Klaus J. Schulten, University
of Illinois at Urbana-Champaign (85,532 normalized CPU
hours)
Schulten carries out molecular dynamics simulations of supramolecular biological
systems such as light-harvesting proteins, protein-DNA complexes, and high-density
lipoproteins. His team looks at molecular photodynamics, or how these structures
react to light, and uses steered molecular dynamics methodologies to study these
large protein structures.
Wai-Mo Suen, Washington
University, St. Louis (77,187 normalized CPU
hours)
Suen and his co-researchers study 3D numerical relativity and relativistic
astrophysics. Using Einstein's general theory of relativity, the team simulates
gravitational phenomena that occur when black holes or neutron stars experience
head-on or grazing collisions. For more on Suen's research, see
http://access.ncsa.uiuc.edu/Features/o2krun/O2KRecord.html.
Peter A. Kollman, University
of California, San Francisco (43,438 normalized CPU
hours)
A professor of pharmaceutical chemistry, Kollman develops simulations that provide
detailed models, insight into molecular mechanisms, and better predictions of
reactions in biomolecular experiments. His team's overall goal is to develop and
apply new methodologies that will enable accurate computer simulations of complex
molecular structures with many hundreds or thousands of atoms.
Kelvin Droegemeier, Center for
Analysis and Prediction of Storms, University of Oklahoma
(37,192 normalized CPU hours)
Using a numerical storm prediction system called Advanced Regional Prediction
System (ARPS), Droegemeier investigates the fundamental characteristics of large
storms and tornadoes while evaluating ARPS's ability to perform realtime weather
forecasting. For more on Droegemeier's research, see
http://access.ncsa.uiuc.edu/CoverStories/StormPrediction/.
Margaret Geller, Harvard
University (33,039 normalized CPU hours)
Geller investigates the evolution of galaxies in clusters, developing simulated
cluster environments that look at how cluster evolution is tied to evolution in the
cluster's galaxy. Her team studies the properties of giant elliptical galaxies in
the center of clusters as part of their effort to understand how galaxies merge to
form these ellipticals.
Marvin L. Cohen, University of
California, Berkeley (23,462 normalized CPU
hours)
The research done by Cohen's group covers broad topics in materials science,
including semiconductors, metals, surfaces and interfaces, defects,
superconductivity, materials under pressure, fullerene-based materials, nanotubes,
and many-body effects in solids. The main purpose of this research is to explain
and predict the properties of materials at the microscopic level through quantum
mechanical calculations.
Lars Hernquist, University of
California, Santa Cruz (20,469 normalized CPU
hours)
Hernquist, a member of the Alliance Cosmology team, conducts research in galactic
dynamics and cosmology. His current work involves the study of galaxy formation and
the associated large-scale structure of galaxies.
The National Computational Science Alliance (Alliance) is a partnership to
prototype an advanced computational infrastructure for the 21st century and
includes more than 50 academic, government and industry research partners from
across the United States. The Alliance is one of two partnerships funded by the
National Science Foundation's Partnerships for Advanced Computational
Infrastructure (PACI) program, and receives cost-sharing at partner institutions.
NSF also supports the National Partnership for Advanced Computational
Infrastructure (NPACI), led by the San Diego Supercomputer Center.
The National Center for Supercomputing Applications is the leading-edge site for
the Alliance. NCSA is a leader in the development and deployment of cutting-edge
high-performance computing, networking, and information technologies. The National
Science Foundation, the state of Illinois, the University of Illinois, industrial
partners, and other federal agencies fund NCSA.
Access Online | Posted 10-5-1999
|
