ARSC gains fast Fish
April 23, 2012
The Arctic Region Supercomputing Center will soon be home to Fish, a new Cray XK6m supercomputer named for Alaska’s fisheries and ocean and water resources. ARSC is one of the first organizations to purchase Cray’s newest model, which uses graphics processing units—like those used by gaming computers—to make scientific calculations run even faster.
ARSC director Greg Newby said Fish will be a valuable tool for researchers working to understand and predict changes in arctic systems, including weather and climate, oceans and ice, permafrost and other materials.
“Supercomputers are used for science and engineering problems that require a lot of number crunching and very large storage,” Newby said. “By coordinating the efforts of hundreds of CPUs, a supercomputer can do in a day what a normal computer would take years to do.”
The Cray XK6m system will be delivered with 1152 AMD Opteron CPU cores and 48 NVIDIA Tesla GPU processors. The total theoretical peak system performance is calculated to be 41.75 teraflops. Fish will be housed in two cabinets that will feature artwork created by UAF art student Yumi Kawaguchi. ARSC’s current supercomputer is housed in 12 cabinets and has a peak performace of about 30 teraflops.
In addition to netting Fish, ARSC is in the process of deploying a new 300TB center-wide Lustre file system and expanding its hierarchical file system. These storage resources allow scientists to process massive datasets and store results. Thanks to a new data portal initiative, ARSC users can also provide data to the public via the Web.
Fish’s purchase was funded primarily through National Science Foundation grants, including PACMAN, the Pacific Area Climate Monitoring and Analysis Network and the Major Research Instrumentation program. Installation is expected to begin in June 2012.
ADDITIONAL CONTACTS: Liam Forbes, ARSC HPC systems analyst, 907-450-8618, loforbes@alaska.edu. Greg Newby, ARSC director, 907-450-8663, gbnewby@alaska.edu.
ON THE WEB: http://www.arsc.edu/arsc/
MG/4-23-12/224-12