Abhinav Bhatele Receives NSF CAREER Award to Optimize Parallel Software and Systems

The funding supports Bhatele’s efforts to use data-driven machine learning models to optimize the efficiency and throughput of extreme-scale parallel systems, and the performance of parallel applications running on them.
Descriptive image for Abhinav Bhatele Receives NSF CAREER Award to Optimize Parallel Software and Systems

A University of Maryland expert in high performance computing (HPC) has received a National Science Foundation award to develop innovative methods for optimizing the performance of parallel applications and runtimes, and the operational efficiency of supercomputers and HPC clusters.

Abhinav Bhatele, an assistant professor of computer science with an appointment in the University of Maryland Institute for Advanced Computer Studies (UMIACS), is principal investigator of an NSF Faculty Early Career Development (CAREER) award, which is expected to total $550,000 over five years. 

The funding supports Bhatele’s efforts to use data-driven machine learning (ML) models to identify performance anomalies and their root causes in parallel systems. The information gleaned from these ML models is then used to optimize performance of these systems and the parallel applications running on them.

“If we can remove—or at minimum, decrease—the human element and associated guesswork in the performance engineering loop, parallel software and systems can become truly self-tuning,” Bhatele says.

The project will use a holistic approach to focus on three main tasks: analyze large volumes of software and system data collected over time; apply machine learning techniques to model application and system behaviors; and use these models to guide application, runtime and system optimization decisions that impact future executions.

Bhatele says that this type of approach can significantly improve the performance and portability of parallel software, as well as the operational efficiency of HPC and data center systems, even as applications and systems evolve.

This can ultimately lead to better performance of individual applications, he adds, which in turn leads to faster science breakthroughs and increased system efficiency.

The NSF funding will also provide educational opportunities to engage high school students, as well as train undergraduate and graduate students in parallel computing. This can address the current shortage of computer and computational scientists who are focused on high performance computing, both in industry and in national laboratories. 
Bhatele has been at the University of Maryland since 2019, where he leads the Parallel Software and Systems Group and is part of a consortium of scientists from across the U.S. who are employing the latest advances in machine learning, artificial intelligence, supercomputing and social science data against epidemic outbreaks.

Prior to coming to UMD, he was a principal computer scientist at Lawrence Livermore National Laboratory.

Bhatele receive his doctoral degree in computer science from the University of Illinois at Urbana-Champaign in 2010.

 

***

Self-tuning Parallel Software and Systems” is supported by NSF grant #2047120 from the NSF’s Division of Computing and Communication Foundation.

PI: Abhinav Bhatele, assistant professor of computer science with an appointment in UMIACS.

About the CAREER award: The Faculty Early Career Development (CAREER) Program is an NSF activity that offers the foundation’s most prestigious awards in support of junior faculty who exemplify the role of teacher-scholars through outstanding research, excellent education and the integration of education and research within the context of the mission of their organization.

 

The Department welcomes comments, suggestions and corrections.  Send email to editor [-at-] cs [dot] umd [dot] edu.