University of Illinois at Urbana-Champaign
Amanda Bienz is a postdoctoral researcher in the computer science department at the University of Illinois at Urbana-Champaign. Her research is focused on analyzing and reducing communication costs on emerging parallel architectures. Current research topics include analyzing communication on heterogeneous architectures, reducing communication in graph algorithms, and applying node-awareness to MPI communication routines. She received her PhD in Computer Science from the University of Illinois in August 2018, with research focused on reducing communication costs in sparse matrix operations and linear solvers. She was a recipient of the NSF graduate research fellowship from 2012-2017, and was an NSF GROW awardee in 2015.
Advances in parallel architectures yield the potential to solve increasing large and difficult problems efficiently. However, applications relying on linear solvers, graph algorithms, and neural networks contain performance bottlenecks associated with inter-process communication. My research is focused on reducing communication costs and improving performance of parallel applications on emerging architectures. Throughout my dissertation, my research focused on analyzing parallel communication through improved performance models as well as reducing communication costs through both algorithmic and implementation changes. These methods for reducing communication include using node-awareness to reroute MPI messages in an effort to reduce the amount of costly inter-node data. Current research topics include improving performance models to analyze communication costs on heterogeneous architectures, improving the cost functions in graph partitioning, and extending node-awareness in MPI collective algorithms.