A special report titled Accelerating Computational Science Symposium 2012 details comparison GPU accelerators with today’s fastest central processing units (CPUs), which were presented earlier this year at Accelerating Computational Science Symposium 2012 in Washington, D.C. The Oak Ridge Leadership Computing Facility, which ORNL operates for the U.S. Department of Energy Office of Science, co-hosted the meeting with the National Center for Supercomputing Applications and the Swiss National Supercomputing Centre. Additional sponsors were Cray Inc., and NVIDIA. The meeting convened nearly 100 experts in science, engineering, and computing from around the world to discuss research advances that are now possible with extreme-scale hybrid supercomputers, which combine traditional CPUs with high-performance, energy-efficient GPUs. Attendees explored how hybrid supercomputers speed discoveries, such as deeper understanding of phenomena from earthquakes to supernovas, and innovations, such as next-generation catalysts, materials, engines, and reactors.
Can scientists and engineers benefit from extreme-scale supercomputers that use application-code accelerators called GPUs (for graphics processing units)? Comparing GPU accelerators with today’s fastest central processing units, early results from diverse areas of research show 1.5- to 3-fold speedups for most codes, which in turn increase the realism of simulations and decrease time to results. With the availability of new, higher-performance GPUs later this year, such as the Kepler GPU chip to be installed in the 20-petaflop Titan supercomputer, application speedups are expected to be even more substantial.
The Accelerating Computational Science Symposium 2012 report is available for download here.