November 13, 2003
NSF Cyberinfrastructure Previews at SC2003 Merge Fast Networks, High-End Computing and Big Data
Organizations and researchers supported by the National Science Foundation (NSF) will showcase the emerging capabilities of the nation's cyberinfrastructure at SC2003, the annual conference of the high-end information technology community, during the meeting exhibition November 17-20, in Phoenix, Ariz. For more information on the projects described here, contact David Hart at 703-292-7737 or firstname.lastname@example.org.
For more on SC2003, see http://www.sc-conference.org/sc2003/.
LESSONS FROM THE TERAGRID—The NSF-supported Extensible Terascale Facility, better known as the TeraGrid, will be featured in the exhibits of all nine TeraGrid partner sites. The partners will demonstrate TeraGrid capabilities in bioinformatics applications, atmospheric forecasting, large-scale and long-distance data movement and grid application development. Several TeraGrid demonstrations have been entered in the SC2003 High-Performance Bandwidth Challenge. In addition, experts from the National Center for Supercomputing Applications (NCSA), San Diego Supercomputer Center (SDSC), Argonne National Laboratory, Pittsburgh Supercomputing Center and Center for Advanced Computing Research at Caltech will conduct a TeraGrid tutorial as part of the SC2003 technical program.
Top of Page
TUNGSTEN AND DATASTAR LIGHT UP PACI SITES—In the coming year, NCSA and SDSC, the two leading-edge sites for NSF's Partnerships for Advanced Computational Infrastructure (PACI), will be operating new supercomputers that will likely rank among the world's most powerful computers. NCSA's Tungsten cluster, which boasts 2,900 Intel Xeon processors and a peak performance of more than 17 teraflops, is expected to be available for production use in December. SDSC will preview technology from DataStar, ultimately a 10-teraflops system slated to enter initial production in spring 2004, in an SC2003 demonstration of scientific applications and high-performance database systems. These new systems will expand the high-end computing capability NSF support makes available to researchers investigating many of science's most important questions.
UPDATE 11/17/03: Tungsten ranked #4 on the Top500 list released Nov. 16.
PACI Resources: www.paci.org
Top of Page
NLANR AMPS UP SC2003 BANDWIDTH CHALLENGE—The SDSC-based Measurement and Network Analysis group of the NSF-supported National Laboratory for Applied Network Research (NLANR) will deploy its new 10-gigabit (OC-192) passive monitoring capability to capture SC2003 Bandwidth Challenge network traffic for later public analysis, according to NLANR's Ronn Ritke, who will serve as a Bandwidth Challenge judge. The NLANR group will also deploy an Active Measurement Project (AMP) monitor on SCinet, the SC2003 high-performance network, to provide network link performance information from SC2003 to over 150 other sites. The AMP monitors help network researchers and engineers track site-to-site problems and tune network links for maximum end-to-end performance. More than 150 AMP monitors are deployed on high-speed research networks in the United States and abroad as part of cooperative efforts with Australia, Brazil, Canada, China, Finland, Germany, the Netherlands, Hungary, Israel, Japan, Korea, Mexico, New Zealand, Norway, Russia, Taiwan and Thailand.
NLANR Measurement and Network Analysis: http://moat.nlanr.net/
Top of Page
MULTINATIONAL GRID DEMOS BECOME PRAGMA-TIC—At SC2003, members of the Pacific Rim Applications and Grid Middleware Assembly (PRAGMA) will participate in more than a dozen demonstrations that will emphasize the power of international computing grids. The demonstrations include climate simulations that span systems in Japan, Korea, India, Taiwan, Singapore, Hong Kong, the United States and Thailand; bioinformatics calculations running on systems in the United States, Singapore, Japan, Australia, Brazil and the United Kingdom; and global telescience that involve resources and users from the United States, Argentina, Japan, Korea, the Netherlands, Sweden and Taiwan. PRAGMA was founded in 2002 with NSF support, and NSF recently awarded $1.2 million to the University of California, San Diego, to enable PRAGMA to continue and expand its activities. PRAGMA helps forge sustained collaborations among its 18 member institutions and advance the use of grid technologies among a community of investigators at leading institutions around the Pacific Rim.
Top of Page
BIG MAC BREAKS INTO HIGH-END COMPUTING ELITE—A $400,000 NSF grant to build a large-scale cluster "was the catalyst that led us here," according to Srinidhi Varadarajan, the researcher who spearheaded the effort to create the Big Mac cluster at Virginia Tech out of 1,100 of Apple's PowerMac G5 computers. The cluster, built for a total cost of $5.2 million, is expected to rank solidly among the top 10 most powerful computers in the world when the 22nd Top500 list (www.top500.org) is released during SC2003. An NSF CAREER award supported Varadarajan in development of Déjà Vu, a software package that helps address reliability issues for such large clusters.
UPDATE 11/17/03: Virginia Tech's cluster ranked #3 on the Top 500 list released Nov. 16.
Virginia Tech Terascale Cluster: http://computing.vt.edu/research_computing/terascale/
Top of Page
GRIDS TUTORIAL OFFERED TWICE BY POPULAR DEMAND—Participants in the NSF Middleware Initiative (NMI) GRIDS Center will present a number of SC2003 tutorials, technical papers and posters. GRIDS leaders are among the principal architects of the Open Grid Services Architecture, the first major implementation of which—Globus Toolkit 3.0 (GT3)—is the subject of a full-day tutorial. The tutorial was so popular SC2003 organizers scheduled it twice. "How to Build a Grid Service Using GT3" will be offered on Sunday and Monday, November 16 and 17. The GRIDS Center and NMI EDIT Consortium together are collaborating to produce the NMI releases of software and related components.
GRIDS Center at SC2003: http://www.grids-center.org/news/sc03/
Top of Page