Asgari Receives $1M NSF Award for Optimizing High-Performance Computing

Bahar Asgari, an assistant professor of computer science with an appointment in the University of Maryland Institute for Advanced Computer Studies, has received funding from the National Science Foundation for innovative research to optimize high-performance computing (HPC) platforms.

The $1 million, five-year award supports Asgari’s efforts to improve communication overheads in HPC applications. These overheads involve “packets” of data transmitted from one node to another, a process that has become hampered by a rapid increase in data used for large-scale simulations in areas like molecular dynamics or computational fluid dynamics.

Asgari’s award is part of a larger grant totaling $5 million from the NSF’s Principles and Practice of Scalable Systems program, which supports interdisciplinary research that spans the entire hardware-software stack, focusing on the scalability and accuracy of modern applications, systems, and toolchains built on heterogeneous architectures.

The larger grant includes four $1 million awards given to faculty from the Georgia Institute of Technology, where Asgari earned her doctoral degree in electrical and computer engineering 2021. The Georgia Tech researchers—Hyesoon Kim, Haesun Park, Richard Vuduc, and Jeffrey Young—have a wealth of experience in areas like energy efficient computer architectures, numerical computing, performance engineering, and high-performance algorithms.

Collectively, the researchers plan to focus their work on three key areas: a) enhancing the scalability of computing for high-performance applications via reductions in data movement; b) developing novel algorithmic and hardware designs for the broader HPC community to leverage emerging technologies combining communication and computation; and c) building an open-source software infrastructure that can be used to facilitate education in parallel computing, HPC, and computer architecture at the graduate and undergraduate levels.

“We expect that our hardware and software advancements will allow for a larger class of applications to take advantage of next-generation, high-performance computing that minimizes communication overheads and data movement,” Asgari says. “Ultimately, we believe this will have an impact on important societal issues that demand efficient, data-centric computing applications to find a best solution.”