Jere Segrest, M.D., Ph.D., knows his desktop computer is pretty fast. But it’s not that fast.
|From left to right, James Patterson, Martin Jones, Jere Segrest, Tommy Foley and Mike Perez take a look at the new IBM Blue Gene supercomputer recently purchased by UAB. |
“How fast would it rank in megaflops?” he asks James Patterson, Ph.D., his postdoctoral assistant.
“It’s not even a flop,” Patterson says with a wry grin.
This kind of exchange – talking about megaflops and teraflops, measures of a computer’s speed – is happening more and more on campus thanks to the recent purchase of an IBM Blue Gene supercomputer. The IBM Blue Gene is a new family of supercomputers optimized for bandwidth, scalability and the ability to handle large amounts of data while consuming a fraction of the power and floor space required by today’s fastest systems. Blue Gene is the fastest computer in the world.
The purchase of the system now means that the computing power of UAB’s cyber-infrastructure has tripled twice in four years, says David Shealy, Ph.D., chair of the Department of Physics.
“UAB has done a reasonable job the past four years in developing high-performance computer resources, and the acquisition of the Blue Gene is a perfect example,” Shealy says.
UAB has been developing a community of high-performance computing (HPC) machines, or clusters, for quite some time. Five other clusters – named Coosa, Olympus, Cheaha, Everest and Cahaba – have been the resources utilized by researchers such as Segrest, the director of the Center for Computational and Structural Biology.
Those machines, while fast, still are at least five-and-a-half times slower than the Blue Gene for certain applications. A recent simulation run by Segrest and Patterson using Cahaba, which was purchased four years ago, would have taken six months. On the Coosa, purchased two years ago, Patterson estimates it would have taken a month-and-a-half. On the Blue Gene?
“You’re talking a few weeks,” Patterson says. “And that means we can do it again and again and improve the quality. You can test if your results are actually meaningful or if you just got the one case where it’s doing something completely off the wall. That’s really the advantage – the speed.”
Benefits are numerous
The new Blue Gene enhances UAB efforts in a number of areas.
First, it gives the institution an enormous amount of computing power and computing speed. Based on the latest Top 500 rankings, the UAB Blue Gene is the 148th fastest computer in the world, giving the university a fast machine to do top-quality research. According to IBM, UAB’s Blue Gene is the sixth fastest non-governmental computer in the Southern United States.
“That in turn positions us to recruit premium faculty researchers and top-quality graduate students, and to set the stage for a computational biology community here,” says Mike Perez, program manager in the?UAB Center for Computational and Structural Biology. “Plus, the NIH Roadmap is high on team science, collaboration and cross discipline involvement in research. High performance computing is an extremely good collaborative tool.
“In addition to the other things we have, I think the Blue Gene will help us with the team science approach the NIH is focusing on and enable us to receive more funding,” Perez notes.
Shealy is adamant that the university greatly improves the quality of its research and education mission by investing in forefront infrastructure like the Blue Gene.
“You recruit better faculty and students, you win more grants,” Shealy says. “If you don’t have good people and infrastructure to do the work, you don’t win new grants.”
Acquiring the Blue Gene
Richard Marchase, Ph.D., vice president of Research and Economic Development, gave a grant for half of the purchase cost of the Blue Gene through the Research Enhancement Fund. Segrest played the major role in getting the Blue Gene matching funds from the partnering schools and departments and from Southern Research Institute (SRI).
Others securing the funds to match the grant include: Robert Kimberly, M.D., School of Medicine; Linda Lucas, Ph.D., School of Engineering dean; Lowell Wenger, Ph.D., School of Natural Sciences & Mathematics dean; John Secrist, Ph.D., Southern Research Institute; Kirby Bland, M.D., Department of Surgery; David Allison, Ph.D., Department of Biostatistics; John Amos, O.D., School of Optometry dean; Max Michael, M.D., School of Public Health dean; David Chaplin, M.D., Ph.D., Department of Microbiology; Mary J. MacDougall, Ph.D., School of Dentistry.
The Blue Gene is housed in the UAB Shared High Performance Computing Lab in the School of Engineering. The Office of the Vice President for Information Technology provided the infrastructure component, including power and additional cooling for the site.
The Blue Gene completed installation testing and was accepted by UAB Feb. 28. Patterson began using the Blue Gene April 6.
Up to four users have been identified to begin using the machine immediately, and Segrest’s group is in the process of identifying specific areas of research that will benefit from the massively parallel supercomputing power of the Blue Gene.
Perez says he expects that as other researchers identify applications for supercomputing to help in their research, the user community will grow significantly – and the available computing time will be spread over a larger and scientifically diverse community.
“Given that even today’s fastest supercomputers require sometimes long runs, days or weeks for a particular experiment, the UAB Blue Gene is certain to be highly utilized,” Perez says.