Nov. 8, 2007 — Each day, many of us take the capabilities of Internet search engines for granted as we type in queries and receive almost instant answers. Yet developing quicker and more efficient computation to discern relevant information from ever-increasing quantities of data is a constant challenge for computer scientists like Sudhanva Gurumurthi, assistant professor of computer science at the University of Virginia.
Traditionally, information storage and processing has occurred in separate realms. Data resides on a disk and must be accessed and moved in order to be processed. Gurumurthi envisions an entirely different model of information processing — rather than moving the data to where it can be computed, he suggests that speed and efficiency can be gained by moving computation to where the data are residing.
Gurumurthi’s innovative proposal for "adaptive active storage systems" won him a $400,000 award from the National Science Foundation's Faculty Early Career Development program earlier this year. The award is one of the most prestigious grants available to junior faculty members in science and engineering fields. The grant provides resources to faculty who have demonstrated great potential early in their careers.
Gurumurthi and his students are using the NSF funding to explore different architectures that make the unification of storage and processing possible. “There are lots of challenges to be addressed,” says Gurumurthi. “The power consumption of the devices needs to be low and we need to develop the necessary software support for using adaptive active storage architectures. There is also a need to build new infrastructure to study such architectures.”
The NSF's Early Career program requires awardees to find meaningful ways of integrating their research with teaching in order to benefit students with hands-on training. Students working in Gurumurthi’s lab are exploring various aspects of active storage architectures. The simulation infrastructure developed is also being used by students in his graduate course. In addition, a Research Experience for Undergraduates participant assisted last summer with software design aspects of the project.
Potentially, Gurumurthi’s work will hold benefits for such diverse applications as search engines, bioinformatics, data mining, and high-performance computing — enabling speedier and more precise results.
Gurumurthi has built working relationships with industry leaders such as Intel and Hewlett Packard in order to gain feedback on research as well as share information and ideas. “We’re hoping to have an impact soon — our insights from this research could help industry right away,” he says.
• Written by Melissa Maki,