NCSA 30 | Norman team wins HPC Challenge at SC95
51684
post-template-default,single,single-post,postid-51684,single-format-standard,ajax_updown_fade,page_not_loaded,,tactile-ver-1.8.1,smooth_scroll,wpb-js-composer js-comp-ver-4.11.2.1,vc_responsive

Norman team wins HPC Challenge at SC95

Michael Norman

Norman team wins HPC Challenge at SC95

NCSA not only sponsored the High-Performance Computing Challenge at SC95, the annual supercomputing conference, a team led by those from the center won it in the “Best Integration of Heterogeneous Applications” category. The contest featured 10 groups of researchers seeking to outdo each other in the race for what was at the time the first teraflop-scale computation. Their simulation took advantage of the high-bandwidth, low-latency network called I-WAY assembled especially for the San Diego conference, as well as computers at NCSA, the San Diego Supercomputer Center, the Cornell Theory Center, and the Pittsburgh Supercomputer Center.

NCSA research scientist Mike Norman (now at the University of California at San Diego), graduate student Greg Bryan (now a professor at Columbia University), and programmer John Shalf (now at Lawrence Berkeley National Laboratory) simulated our own Milky Way galaxy colliding with the neighboring Andromeda galaxy as part of the winning project.

“Two supercomputers, one in Illinois and one in San Diego, each simulated the stars in one galaxy, while the I-WAY was used to communicate the information needed for the galaxies to interact. We designed the application so that it could still make some progress using the regular internet if the I-WAY failed for some reason, which was a good thing because during our live presentation the I-WAY did go down,” remembered Bryan.

They worked with others from the National Science Foundation’s Grand Challenge Cosmology Consortium, including long-time NCSA collaborators Lars Hernquist, John Dubinski, and Dennis Gannon.