Duke CS Wins Best Paper Award at ISPASS 2023

Duke CS Wins Best Paper Award at ISPASS 2023
Xu, Mason, and Wills

Duke Computer Science researchers (L-R in photo) Entropy Xu (3rd year CS PhD), Mason Ma (2nd year CS PhD), and Assistant Professor Lisa Wills have won the Best Paper Award at ISPASS-2023, the 2023 IEEE International Symposium on Performance Analysis of Systems and Software, which took place April 23-25, 2023 in Raleigh, NC.

The paper, entitled “PyTFHE: An End-to-End Compilation and Execution Framework for Fully Homomorphic Encryption,” presents PyTFHE, an ingenious, innovative end-to-end toolchain for developers to create and execute their privacy-preserving, fully homomorphic encryption applications at high performance without hassle. Their PyTFHE toolchain features a neural network compiler that converts a PyTorch model to a Chisel equivalent, an assembler that generates an FHE binary program, and backends that enable efficient execution on GPUs and distributed CPU clusters providing performance, productivity, and security. Xu, Ma, and Wills compared PyTFHE with state-of-the-art frameworks using an MNIST network TFHE program execution and showed that PyTFHE achieves 200x-4000x performance advantage over existing works.

The best paper award was judged on paper quality, potential for future impact, and the presentation by Mason at the ISPASS event, a niche computer architecture conference. Dr. Wills said, "I am very proud of my students and this is Mason’s first paper!"

Prof. Wills received her PhD in Computer Science from Columbia University in 2014, followed by postdoctoral research in  Electrical Engineering and Computer Science at the University of California at Berkeley. Wills joined Duke CS and ECE faculty at the Pratt School of Engineering in 2019. Her research is focused on computer architecture and micro-architecture, hardware-software co-designs, accelerators, and emerging application domains ― especially those in big data analytics such as genomics, graphs, and databases.  She received an NSF CAREER Award in 2021, was recognized by IEEE Micro Top Picks in 2021 for Genesis: A Hardware Acceleration Framework for Genomic Data Analysis, by Google with a Rising Systems Faculty Award in 2020, and also received a Facebook Research Award on Systems for Machine Learning in 2020.