NERSC Delivers One Million T3E CPU-Hours to Scientific Users
On Thursday, Oct. 23, NERSC passed a milestone of delivering 1,000,000 parallel computing CPU-hours to the scientific community. The records of all parallel computation performed on the prototype Cray T3E-600 from late January to early September, and on the full configuration 512-processor T3E-900 from late July to October 23 indicate that Grand Challenge users received 425,000 parallel computing hours, while the remaining 575,000 hours were used by more than 100 other research groups. (And thanks to NERSC's Frank Hale for monitoring T3E use and noting the million-hour milestone.)
"Not only does this demonstrate NERSC's value to the Energy Research community, it also demonstrates our leadership in moving MPP technology into a production environment," said NERSC Division Director Horst Simon. "As a matter or course, we're now delivering about 50,000 CPU-hours a week to our research community."
Next NERSC Division Seminar to Spotlight Data Intensive Computing
Brian Tierney, head of the new Data Intensive Computing Group, will give an overview of the Distributed-Parallel Storage System (DPSS), a scalable, high-performance, data storage system developed by the Data Intensive Distributed Computing Group as part of the DARPA-funded MAGIC Testbed. Brian will give his talk from 2-3:30 p.m. Tuesday, Nov. 4, in Bldg. 50B, room 4205. All interested employees are invited to attend. The DPSS is a collection of wide-area distributed disk servers which operate in parallel to provide logical block level access to large data sets. Operated primarily as a network-based cache, the architecture supports cooperation among independently owned resources to provide fast, large-scale, on-demand storage to support data handling, simulation, and computation in a wide-area inter-networked environment. Brian will discuss the architecture and implementation of the DPSS, and some sample DPSS applications that require high data rate as well as future directions, including HPSS integration, survivability, fault tolerance, and dynamic reconfiguration, using an agent-based architecture. The seminar is the second in a monthly series in which people from the different NERSC groups will describe their work and have an opportunity to learn about each other's projects.
More Recipients of CS Outstanding Performance Awards
In the latest round of OPAs, 30 employees were recognized for their contributions to the Computing Sciences organization. Over the past few weeks, we've been presenting the recipients in InTheLoop. Here's the third and final group: Eric Beals, Hua-Pei Chen, Sandra Chinchilla, Norma Early, Eric Hibbard, Bill Kramer, Marsha Mathews, Emmy Randol, Mark Rosenberg, Ted Sopher and Dave F. Stevens were all honored for their work in the LBNL Computing Environment Design Team, which analyzed the state of the Lab's computing infrastructure and came up with recommended improvements. To gather information, the team met with Lab employees and organizations, conducted a survey and held two town meetings. In addition to coming up with recommendations, the team also analyzed the costs associated with various proposals. The efforts resulted in the creation of the Computing Infrastructure Support Department within ICSD and a revamped system for computer hardware and software support at the Lab.
Jane Colman Races to Fastest Age-Group Finish in Recent Runaround
Congratulations to Jane Colman of the UNIX Support Group, who placed first in the women's 50-59 age group in the 20th annual Lab Runaround held earlier this month. Although the results published in Currents didn't include Colman's name, she did receive a gold-colored medal attesting to her strong finish. She completed the 1.86-mile course in 16 minutes, 57 seconds. She's run up quite a track record in the Runaround, missing only three races since she started competing in the annual event in 1979 and winning her age group three times. Colman, who runs nearly every day and runs competitively on many weekends, says she prefers to run longer, slower races, such as marathons and half-marathons.
DOE's Tom Kitchens, Bob Aiken to Visit Computing Sciences
Among the visitors scheduled for this week are Tom Kitchens and Bob Aiken of the Division of Mathematical, Information, and Computational Sciences. Tom is program manager for NERSC, while Bob is program manager for network research. Both will be here for a day of meetings and discussions.
SC97 Scheduled for Nov.15-21 in San Jose
There's still time to register for SC97, the annual high-performance computing and networking conference. This year's meeting will be held in San Jose and highlights include technical papers, tutorials, education sessions, research exhibits, poster exhibits and exhibits from industry. Computing Sciences will be represented in two research exhibits, one on the Lab's various programs and one showcasing DOE2000 technologies. If you'd like to help with either exhibit, call Roberta Boucher at 7580. For more info on the conference or to register, see http://www.supercomp.org/sc97
Questions, Comments About InTheLoop?
Send your questions, comments, suggestions or news items to JBashor@lbl.gov or call editor Jon Bashor at 5849.