Yale purchases world-class supercomputer

Yale researchers will be able to crunch larger amounts of data even faster, now that the University has acquired a new supercomputer — one rated #146 in the world by the TOP500.org, which tracks and rates supercomputer systems. It is also the top-rated supercomputer among the Ivies.

This new supercomputer brings the total number of shared high-performance computing (HPC) clusters on campus to six.

“High performance computing plays a crucial and expanding role across many disciplines at Yale ranging from genome sequencing, to astro- and geo-physics, combustion engineering and modeling protein interactions,” says Steven Girvin, the Eugene Higgins Professor of Physics and Applied Physics and deputy provost for science & technology. “Our newest cluster which is about to be deployed, dubbed Bulldog Omega, is among the top 100 academic machines in North America.”

David Frioni, manager of the ITS High-Performance Computing Group, notes: “The new 704-node cluster increases our current overall count by approximately 70%. In addition, the new nodes will be faster than some nodes on our current clusters. This will be our largest single cluster, which will allow BulldogO to run much larger jobs than is currently possible.”

One-third of Bulldog Omega — or, more informally, Bulldog O — will be dedicated to atmospheric dynamics studies by a group of recently hired climate scientists in the Department of Geology and Geophysics and the School of Forestry & Environmental Studies. “In order to help faculty and students learn how to make most efficient use of these resources, we will be expanding our staff who support HPC training and software development from three to four research scientists,” notes Girvin.

One Yale researcher who has leveraged the power of Yale’s current HPC resources is Richard Easther, associate professor of physics & astronomy. His research aims to understand the physical processes that took place in the very early universe, exploring both the origin of our universe, and the fundamental forces of nature.

“We now have very large quantities of data about the universe on large scales that were unavailable even 15 years ago,” says Easther. “We can measure the age of the universe to within a few percent, and our theoretical work now has to be accurate to within a few percent or better. Because of that, we have to do our calculations with a lot more precision, which means moving away from pencil and paper and moving towards serious computing.”

Modern research technologies in the physical, biological, engineering, and social sciences are increasingly producing massive datasets that need to be analyzed, according to Andrew Sherman, an associate research scientist in the Department of Computer Science and one of the HPC specialists in the Faculty of Arts & Sciences HPC Center.

“Such datasets may arise from automated experiments, surveys, or simulations,” he says. “For instance, one physicist reports collecting 100 gigabytes of data nightly from a telescope in Chile (that’s roughly equivalent to collecting 100 thousand books per night), while the new West Campus DNA Sequencing Center generates approximately terabytes of data per day from their sequencing machines (think more than one million books per day).”

Yale’s current wave of HPC expansion began in fall of 2008 with the installation of the 128-node “BulldogJ” cluster in the ITS data processing center at 300 George St. This was followed in March of 2009 with the addition of 192 additional nodes in the “BulldogK” cluster. Three additional clusters were installed in 2009-2010 at the West Campus High Performance Computing Center: “BulldogL,” a 128-node cluster for general FAS use; “BulldogM,” a 128-node cluster primarily for astrophysics computations; and “BulldogN,” a 96-node cluster that will be part of the West Campus DNA Sequencing Center.

The latest addition, BulldogO, which was purchased from Hewlett-Packard, will be housed at West Campus and is expected to be operational in May.

Support for HPC is provided free-of-charge to members of the Faculty of Arts and Sciences, as well as affiliated researchers in the life sciences. Faculty members of the Faculty of Arts & Sciences can request an account by sending email to hpc@yale.edu. Life Science researchers can send email to either Nick Carriero (carriero@cs.yale.edu) or Rob Bjornson (robert.bjornson@yale.edu) to request an account.

For more High Performance Computing resources, see http://cmi2.yale.edu/hpc/.

Share this with Facebook Share this with Twitter Share this with LinkedIn Share this with Email Print this