Automatic vectorization for multi-party computation

Authors
Sherman, Benjamin
ORCID
Loading...
Thumbnail Image
Other Contributors
Cutler, Barbara M.
Turner, Wes
Milanova, Ana
Issue Date
2022-08
Keywords
Computer science
Degree
MS
Terms of Use
Attribution-NonCommercial-NoDerivs 3.0 United States
This electronic version is a licensed copy owned by Rensselaer Polytechnic Institute (RPI), Troy, NY. Copyright of original work retained by author.
Full Citation
Abstract
We introduce a new compiler for multi-party computation (MPC) which performs backend-independent optimizations over the MPC Source intermediate representation and outputs C++ code using the MOTION framework for MPC. We showcase a specific optimization: novel automatic vectorization over MPC Source. This optimization is shown to provide significant performance improvement in most benchmarks, and mitigations are suggested to detect and prevent cases where this optimization leads to a performance regression. Both circuit generation and evaluation time improves, and there is a consistent reduction in communication size and amount. We additionally compare results to hand-optimized MOTION code and find that there are only minor differences which can easily be overcome in future work.
Description
August 2022
School of Science
Department
Dept. of Computer Science
Publisher
Rensselaer Polytechnic Institute, Troy, NY
Relationships
Rensselaer Theses and Dissertations Online Collection
Access
CC BY-NC-ND. Users may download and share copies with attribution in accordance with a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 license. No commercial use or derivatives are permitted without the explicit approval of the author.