Significant research in the area of computational power and large-scale application efficiency is being conducted by Professor Barbara Chapman, one of the latest computer science faculty members to receive a National Science Foundation (NSF) funding award.
Chapman won the NSF Scalable Parallelism in the Extreme, otherwise known as an SPX award, for her research, Cross-layer Application-Aware Resilience at Extreme Scale (CAARES).
The funded research addresses the challenges imposed by future extreme-scale architectures that will require dynamic programming approaches, where different software layers, potentially developed using different programming paradigms, will have to closely interact with each other.
“I’m extremely excited to represent Stony Brook University by winning this award thanks to the NSF,” Chapman said. “The SPX award is one of the most prestigious in the industry, and it means the world to me to be named a recipient.”
Chapman’s work quantifies from a theoretical standpoint what the possible benefits are of gathering two well-understood resilience mechanisms, along with ABFT numerical libraries, in one representative application. Never proposed before, the target application is representative of computational science domains.
“In this project we will explore an in-staging data management runtime that combines multiple strategies using them appropriately at execution time based on data locality and/or importance,” the proposal states. “For example, spatial/temporal data locality and user provided hints can be used to classify data based on its access pattern and access mode (read only, write only, or both), and this classification can be used to select appropriate data resilience strategies based on desired cost-benefit ratios.”
The three-year project received more than $300K in funding and will conclude in July 2020. In total, three graduate students will be hired to participate in the SPX project, and they will work closely with the project lead, Anthony Curtis.
A faculty member at Stony Brook University since 2015, Chapman is a core part of the Institute for Advanced Computational Science and the Department of Computer Science. Additionally, she is a professor in the Department of Applied Mathematics and Statistics (AMS).
Prior to joining Stony Brook’s College of Engineering and Applied Sciences, she served as the director of the Center for Advanced Computing and Data Systems at the University of Houston. While working in Houston, her research group developed OpenUH, a state-of-the-art open source compiler that is used to explore language, compiler and runtime techniques, with a special focus on multi-threaded programming.
Chapman also leads the Exascale Programming Models Laboratory, which aims to increase programmer productivity, enhance application performance, develop novel implementation technologies that anticipate architectural changes and emerging user needs, and provide reference implementations of our ideas.
— Joseph Wolkin