Professors and students in the science departments will soon be tracking the evolution of DNA strands and the development of tumors faster than ever before thanks to the recent installment of a new high-performance computer cluster at Information Technology Services (ITS).
The new cluster offers a drastic increase in power over the computer resources formerly available, connecting 14 times faster than high speed Internet. It will be up and running by mid-March, and it is estimated that professors, undergraduates, and graduate researchers will have full access by April.
“The cluster allows you to distribute workload across multiple machines,” said Ganesan Ravishanker, associate vice president for ITS.
The cluster boasts 288 processors from Dell, which work together as a single unit, enabling the calculation of dozens of complicated problems at the same time. The cluster will soon be connected to a ten-terabyte cluster of hard drives (a terabyte is 1,000 gigabytes). The cluster already generates enough heat to warrant the installation of a new 15-ton cooling unit.
The majority of the funding for the project came from a National Science Foundation (NSF) grant totaling $190,000. A group of professors who were all conducting heavily computer-intensive research, including Professor of Science and Mathematics David Beveridge, Assistant Professor of Physics Francis Starr, and Kathryn Johnston, a former Assistant Professor of Astronomy, submitted the proposal.
According to Ravishanker, the funding was approved partly because the NSF was impressed with the grant’s proposal, which stressed the “partnership between ITS and faculty members.”
This partnership will partly manifest itself by having ITS Applications Technology Specialist Hank Meij act as the cluster’s system administrator and help to support its maintenance.
In addition to fostering more collaboration between ITS and the science faculty, another key goal behind the cluster is to centralize high-end computing resources across campus.
“[The goal is to] consolidate them and move them into the data center here [on the fifth floor of the science center],” Ravishanker said.
Formerly, individual faculty members had no centralized computer resource and had to maintain their own clusters, which were prone to breakdowns. Professors emphasized how pleased they were with the amount of free time they could soon spend working with students rather than with their machines.
“Now [faculty] don’t have to use their time being computer administrators,” Starr said. He currently maintains his own computer cluster for his research, but will use the new technology once it is operational.
ITS has agreed to upgrade the cluster over the next few years, with funding support coming from Academic Affairs and new grant funding.
“Rather than build new ones, we’ll expand this,” Ravishanker said.
ITS has already been in contact with professors about the new cluster, such Professors of Chemistry David Beveridge, George Petersson, and Rex Pratt and Assistant Professor of Physics Tsampikos Kottos. Computer science and biology professors may also use the cluster.
Starr estimates that there will be four or five “heavy users” of the machine, but that many other professors will be able to use it for smaller projects.
Beveridge, for example, works in computationally intensive molecular dynamics, studying how DNA and protein complexes evolve structurally. The increase in computing power will allow him to model how atoms interact when surrounded by water. This involves thousand and thousands of calculations, impossible to do on a regular desktop computer.
Starr will use the cluster to support his work in nanotechnology. Because he works on such a small scale, it is often not useful to do real life experiments. Instead, theoretical models can yield better results. Starr uses computers to test how changing variables at the nano-scale results in new materials.
Use will not be limited to professors, though.
“We want to get more and more students involved with it,” Starr said. “It gives many more students a chance to get involved with high end performance computing than before.”
Starr envisions a situation where students trained on the machine in classes or research labs can help other faculty members less versed in the technology. This is made easier by the simplicity of remote connection.
“My hope is that there is no end in what can happen here,” Starr said. “This is a new paradigm for computing at Wesleyan.”
Leave a Reply