Newswise — BOSTON— Boston University researchers this month will fire up the largest academic GPU (graphics processing unit) computer cluster on the East Coast, thanks to a gift of 160 GPUs from the Hewlett-Packard Company. The GPUs, valued at $320,000, will be installed in a new cluster purchased by BU Information Services & Technology (IS&T) and the Center for Computation Science (CCS).
The new system, called BUDGE (Boston University Distributed GPU Environment), is based on GPU computing, which uses GPUs together with a central processing unit (CPU) to accelerate general-purpose scientific and engineering applications. GPU computing has quickly become a computer industry standard, offering unprecedented application performance by offloading compute-intensive portions of applications to the GPU, while the remainder of the code runs on the CPU. From a user's perspective, GPU computing makes applications run significantly faster.
The GPU-enabled BUDGE cluster offers peak performance in excess of 32 teraflop/s (or 32 million million operations per second) at an unprecedentedly low cost for such number-crunching power. This supercomputer will be used by principal investigators Lorena Barba, assistant professor of mechanical engineering; Richard Bower, professor of electrical and computer engineering; Martin Herbordt, associate professor of electrical and computer engineering; and Claudio Rebbi, professor of physics and former director for the Center for Computational Science (CCS), for research in particle physics, fluid dynamics, and bioinformatics and computational biology. The cluster will also be made available to other researchers for their projects.
BU’s involvement with GPU computing began five years ago, when a graduate student of Professor Rebbi introduced him to CUDA, a programming model invented by GPU manufacturer NVIDIA. Rebbi soon purchased a GPU system for CCS and encouraged his students to write code for the device. Following these early efforts, the CCS received a donation of several more GPUs from NVIDIA at the end of 2008, allowing the center to spearhead a new national effort to develop GPU code. An NSF EAGER grant funded the purchase of BU’s first GPU cluster, named BUNGEE.
With so many university groups now using GPU computing for their research, BU has achieved status as a national leader in GPU research, which was underscored earlier this year when Professor Barba was appointed a CUDA Fellow by NVIDIA.
About Boston University
Founded in 1839, Boston University is an internationally recognized private research university with more than 30,000 students participating in undergraduate, graduate, and professional programs. As Boston University’s largest academic division, the College and Graduate School of Arts & Sciences is the heart of the BU experience with a global reach that enhances the University’s reputation for teaching and research.
# # #