Skip to Navigation Skip to Page Content

MU assistant professor earns back-to-back Best Paper Awards at international symposium

ECE Assistant Professor Michela Becchi accepts the Best Paper Award at the 21st International ACM Symposium on High-Performance Parallel and Distributed Computing (HPDC ’12) in Delft, the Netherlands. Photo by the HPDC

A team led by Michela Becchi, an assistant professor in the University of Missouri College of Engineering’s Department of Electrical and Computer Engineering, received the NVIDIA Corporation-sponsored Best Paper Award at the 21st International ACM Symposium on High-Performance Parallel and Distributed Computing (HPDC ’12).

Becchi’s paper, “A Virtual Memory-Based Runtime to Support Multi-tenancy in Clusters with GPUs,” proposes a software framework that enables graphic processing units (GPUs) to be used in clusters of computers — a network of machines. GPUs are many-core processors that facilitate faster processing within the individual computers. Modern GPUs are primarily used for performing graphical operations in PCs also are found in video game consoles and smart phones.

“GPUs are highly parallel devices. They were first designed for graphics applications, but have been recently used for many scientific problems, including those in bioinformatics, data mining, weather prediction and option pricing,” Becchi said, adding that GPUs have become widespread in the last five or six years.

“GPUs are traditionally used as dedicated accelerators on a single researcher’s computer; we sought to employ them as shared resources, allowing multiple applications to execute concurrently on the same GPU,” she said.

“It’s pretty applicable,” Adam Procter, a computer science doctoral student said, adding that this type of supercomputing could be used in computational biology, graphic animation and more.

Cloud computing is one of the primary examples of how the technology impacts the general public. For example, Becchi said an online web service, such as Amazon Elastic Compute Cloud, could use the software discussed in her paper to find space among its clustered GPUs to run an application sent to that “cloud” by a developer. Such a system allows both faster solution times for the user and more efficient use of the cloud hardware for the service provider.

This is the second year in a row that Becchi has been a Best Paper Award recipient at HPDC. Two years ago, while working at NEC Laboratories America, she began research into how to use GPUs to simultaneously run multiple applications. The research, a collaborative project with NEC Laboratories America and Ohio State University researchers, resulted in a paper titled “Supporting GPU Sharing in Cloud Environments with a Transparent Runtime Consolidation Framework.” Last year, it too garnered an HPDC Best Paper Award. In addition, the same team won a Best Student Paper Award at the 12th Institute of Electrical and Electronics Engineers/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID’12) for a paper titled “Scheduling Concurrent Applications on a Cluster of CPU-GPU Nodes.”

This year’s HPDC paper was co-authored by MU electrical engineering doctoral candidate Kittisak Sajjapongse, computer science doctoral candidates Ian Graves and Procter, Vignesh Ravi, a recent doctoral graduate from Ohio State University, and Srimat Chakradhar of NEC Laboratories America.

Procter said working with Becchi was a good learning experience.

“Dr. Becchi is really great to work with,” he said. “She’s really up and coming.”

The HPDC ’12 symposium was held June 20-22 at Delft University of Technology in Delft, the Netherlands at which Becchi presented the paper and accepted the award. The annual symposium includes an additional two-day workshop session, keynote speakers and a poster competition.

HPDC is the annual conference on the design, implementation, evaluation and use of parallel and distributed systems for high-performance computing. Such advanced computing is increasingly important for problems such as climate-modeling, aeronautical design and nuclear security, among other applications. The Special Interest Group on Computer Architecture of the Association for Computing Machinery (ACM) sponsors HPDC. In 2012, HPDC accepted 23 papers out of 143 submissions, a 16 percent acceptance rate.

ACM is a global organization whose members include more than 100,000 computing educators, researchers, students and professionals and sponsors more than 170 conferences annually.