The University of Minnesota Supercomputing Institute for Advanced Computational Research has installed a new high-performance computing system from SGI, christened Koronis, that it will use for multi-scale modeling, chemical dynamics, bioinformatics, computational biology, and biomedical imaging.
The university purchased the 1,152-core system with a $3.6 million grant from the National Institutes of Health's National Center for Research Resources. It will support NIH-funded research projects at the university.
Jeff McDonald, assistant director of high-performance computing operations at MSI, told BioInform that the latest purchase is the largest system at MSI and that it was selected because its shared memory capabilities best fit the researchers' needs.
In the grant abstract, the researchers wrote that the new system will help 33 research groups supported by 91 NIH grants "tackle ... the acquisition, analysis and visualization of petascale data from high-performance computing and high-throughput technologies."
McDonald explained that Koronis includes Nvidia graphical processing units to speed up calculations as well as visualization capabilities, such as the ability to generate detailed graphical renditions of their data, that aren’t available in MSI’s other resources.
Furthermore, he said, because Koronis includes both graphical and computational components, it offers a one-stop solution for researchers looking to use both capabilities. Other MSI systems would require the data to be moved from one part of the system to the next, he said.
Koronis, which is named after one of Minnesota’s lakes, is one of four supercomputers housed at MSI and the only resource for life science researchers. It includes a shared-memory system with 3.1 terabytes of memory, 750 terabytes of disk storage, and 1,152 processor cores. It takes up four racks of standard system space covering roughly 32 square feet.
Currently, Koronis is being used to model biochemical interactions with anthrax as well as to study how brain networks are altered in psychiatric disorders.
McDonald said that the biological and medical research activities at U of Minnesota currently generate several terabytes worth of data each month combined.