Highlight Archive 2000
New World Record for Data Transfer Set in Probe
In collaborative testing involving Genroco staff and Dan Million
of ORNL's Probe storage and networking testbed, beta tests of hardware
and software running on Probe's RS/6000 Model S80 server and GSN equipment
achieved a transfer rate of 154 megabytes per second between the S80 and
ORNL's Origin 2000. This rate exceeded by over 20% a record announced by
Genroco only a month earlier. Genroco, Inc., are developers and
manufacturers of Gigabyte System Network
(GSN) hardware and software.
High-Tech Cars May be Risky to DriversDr. Phil Spelt of CSM has released early results from the federal government's first attempt to measure how drivers deal with a potential information overload from an array of high-tech features now being installed in automobiles. While still crunching numbers, Spelt said the overall outcome already is obvious: "People who got bombarded with three or four devices all at once had more trouble dealing with the whole situation than people where we spread them out." more
New Mathematical Method Will Benefit Semiconductor Research
ORNL researchers studying growth of cystalline silicon from amorphous phase are the first to couple Boundary Integral methods for stress evaluation with the Level Set method for front tracking. Simulations done using this new method yield a very good match with experiments. Moreover, they provide conclusive proof that interface instabilities observed result from effects of stress on interface atoms mobility. Growth instabilities due to stress are important in semiconductor device fabrication.
Computational Sciences to get new FacilityOn September 11, 2000 Secretary of Energy Bill Richardson announced a $100 million modernization plan for the Laboratory. The plan includes a new building for computational science. details
Zacharia named Deputy Associate Lab Director for High Performance Computing
On August 4, 2000 Dr. Thomas Zacharia was appointed to the new position of Deputy Associate Laboratory Director for High Performance Computing. In this position, Thomas will lead the development and implementation of ORNL's high performance computing strategy. This strategy includes the development of topical user centers for high performance computing, and continued leadership in terascale computational applications, algorithms, and hardware. details
ORNL Dedicates Terascale Computing Facility
On June 20, 2000, the newly expanded IBM RS/6000 SP supercomputer, Eagle, and the recently acquired Compaq AlphaServer SC system, Falcon, were dedicated in a ceremony that included remarks by Under Secretary of Energy Ernest Moniz, and ORNL Director Bill Madia. After a virtual ribbon cutting, several applications were launched on the new machines, which have a combined peak computational power of 1.5 teraflops.
R&D 100 awardDennis Wolf of CSM was part of an ORNL team receiving an R&D 100 award for the Block II Chemical Biological Mass Spectrometer.
ORNL hosts Networking Research workshop
ORNL hosted a DOE-sponsored workshop to facilitate establishing a regional collaboration in high performance networking research. The small group of invited attendees represented nine universities, two supercomputing centers, three telecommunications organizations, and the Department of Energy.
ARM team recognized by DOE
The ARM team -- staff in ESD, CPED, CSM, and CIND -- received a plaque from Wanda Ferrel, DOE-HQ on May 9th. Coincidentally, on May 9th, HPSS storage (where the ARM data is housed) topped 25 TB of stored data.
ORNL Materials application enshrined in Smithsonian
A team, led by ORNL's Malcolm Stocks and including CSM's William Shelton, has been nominated for the 2000 Computerworld Smithsonian award in honor of their achievement of TeraFLOP performance with a materials code in November of 1998 (for which they won the Gordon Bell award). As a Laureate, this project will be described in the Smithsonian's National Museum of American History's Permanent Research Collection and in the book Faces of Innovation. Now in its twelfth year, the Computerworld Smithsonian Program (CWSP) is considered the most prestigious awards program in the information technology industry.
TWO new supercomputers in the same week!
The new Compaq Sierra system, "Falcon", has arrived at ORNL and is being installed for use. A 64-node Compaq AlphaServer SC with four 667 MHz Alpha EV67 processors per node, Falcon is an early evaluation machine, like Colt, the 16-node, 64 GFLOPS system delivered earlier. Together the Compaqs' estimated computational power is 428 GFLOPs. photos
Can you say "TFLOPS"?
B. Radhakrishnan invited to speak at Gordon conference
B. Radhakrishnan has been invited to speak at the 2000 Gordon Conference: Physical Metallurgy. Gordon Research Conferences provide an international forum for the presentation and discussion of frontier research in the biological, chemical, and physical sciences, and their related technologies. "Rad" will present Multi Length-scale Modeling of Thermomechanical Processing.
New ORNL algorithm shatters record
Knowing vibrational modes of molecules provides important insight into their behavior. A new algorithm for Normal Coordinate Analysis (NCA) has shattered the record for largest modeled system (previous record: 3,000 atoms, ORNL: 24,000 atoms). more info
New method for crack-tip analysis
ORNL researchers have significantly improved the accuracy of computed crack stress intensity factors by incorporating the ORNL-developed "modified quarter-point" element (MQP) into boundary integral fracture codes. This improvement in accuracy makes it feasible to reliably model real problems, particularly those where cracks follow curved paths. This new method is currently being applied to failure in thermal barrier coatings.
Computing with quantum dot arrays
An LDRD team led by Jacob Barhen of ORNL's Center for Engineering Science Advanced Research (CESAR) and the Computer Science and Mathematics division (CSM) is seeking to achieve the world's first practical computation on a fabricated nanoscale device. See ORNL Review article Quantum Computing by Connecting the Dots or Emerging Technologies exhibit for more about this project.
Compaq system added for evaluation
"Colt" is a 16-node Compaq AlphaServer SC being used for early evaluation. With four 500 MHz Alpha EV6 processors and 2GB of memory per node and 365 GB of Fiber Channel disk attached, Colt's estimated computational power is 64 GFLOPs. For more information see Colt page.
"Eagle" is soaring
With the recent upgrade from two processor nodes to four processor Power3+ nodes, Eagle - the IBM RS/6000 SP system - is now rated at 400 GFLOPS. For a look at Eagle and other ORNL High Performance Computing systems, see CCS page or Eagle page. To take a look at how this powerful machine is being used, see Eagle projects and Eagle highlights.
CSM in the newsThe Eagle upgrade was mentioned in the February 7 Frank Munger's In the Lab column in the Knoxville News-Sentinel. For this and other articles, see our PR page.
Jack Dongarra elected as IEEE Fellow
Recognizing the achievements of its members is an important part of the mission of the IEEE. Each year, following a rigorous evaluation procedure, the IEEE Fellow Committee recommends a select group of recipients for one of the Institute's most prestigious honors, election to IEEE Fellow. Less than one in one thousand members of the IEEE will receive this honor in 2000. At its meeting on 14 November 1999, the IEEE Board of Directors elected Jack Dongarra as an IEEE Fellow, effective 1 January 2000, with the following citation: "For contributions and leadership in the field of computational mathematics."
ORNL adds more computing power!
Those boxes being delivered to the computer room are for the IBM RS/6000
SP system. The IBM system, initially configured to perform 100 GigaOP/S,
will be upgraded to 400 GigaOP/S during December. By about the middle
of 2000, we expect the unit to be upgraded to a 1 TeraOP/S capability.