SCIENCE

National Science Foundation begins transition to eXtreme Digital generation of cyberinfrastructure

Scientists, engineers and other U.S. researchers may apply until January 15, 2011 (12:00 midnight submitter’s local time) for the next quarterly review of requests for free allocations of high-performance computer time, advanced user support, data, and visualization resources that are available through the National Science Foundation (NSF) Office of Cyberinfrastructure’s (OCI) TeraGrid.To apply for an allocation of any size, please visit TeraGrid’s online submission system: https://pops-submit.teragrid.org/.

Each quarter, a panel of computational experts known as the TeraGrid Resource Allocations Committee (TRAC) evaluates requests primarily on the appropriateness and technical aspects of using TeraGrid resources. Applications received by the January 15 deadline will be considered at the March 2011 TRAC meeting and awards will be available April 1, 2011 through March 31, 2012. Multi-year allocations may be requested with an appropriate justification for use over an extended period of time.

TeraGrid allocates more than 1.5 billion processor hours to meritorious requests each year. Resources currently exceed 2.5 petaflops of combined computing capability and approximately 50 petabytes of online and archival data storage from 11 resource provider sites across the nation. At the December 2010 TRAC, 125 requests for computational time and storage were reviewed, and 290 million service units of computational time were awarded.

For the next cycle, researchers can request time on 24 systems, including eight that were featured on the November 2010 Top500 list. Among the diverse resources available through TeraGrid are:

· Kraken at the National Institute for Computational Sciences (NICS)—TeraGrid’s largest system is a 1030 teraflops Cray XT5, recently upgraded by 144 teraflops with an additional 1,152 nodes for a peak performance of 1174 teraflops. Users may also request time on NICS’ Athena, a 166 teraflops Cray XT4 system, which is co-allocated with Kraken.

· Ranger at the Texas Advanced Computing Center (TACC)—TeraGrid’s second largest system is a 579 teraflops Sun Constellation Cluster.

· Nautilus (NICS) and Longhorn (TACC) remote visualization and data analysis systems are the first resources to be awarded under the eXtreme Digital, or XD phase of NSF-supported cyberinfrastructure.

· Wispy, a production cloud environment is available at Purdue, featuring a cluster running KVM and the Nimbus cloud software. Users can submit disk images to run a virtual machine with up to 4 CPUs and 16GB of memory.

· Ember, a new shared memory supercomputer at the National Center for Supercomputing Applications (NCSA), features a peak performance of 16 teraflops, double the performance of its predecessor, Cobalt.