Precimonious tuning assistant for floating-point precision

• Published in: 2013 SC - International Conference for High Performance Computing, Networking, Storage and Analysis (SC) pp. 1 - 12
• Main Authors: Rubio-González, Cindy, Nguyen, Cuong, Nguyen, Hong Diep, Demmel, James, Kahan, William, Sen, Koushik, Bailey, David H., Iancu, Costin, Hough, David
• Format: Conference Proceeding
• Language: English
• Published: New York, NY, USA ACM 17.11.2013
• Series: ACM Conferences
• Subjects: Accuracy; Algorithm design and analysis; delta-debugging algorithm; dynamic program analysis; floating-point arithmetic; General and reference > Cross-computing tools and techniques > Verification; Heuristic algorithms; Mathematics of computing > Mathematical analysis > Mathematical optimization; mixed precision; Numerical analysis; Partitioning algorithms; Performance analysis; program optimization; Software and its engineering > Software creation and management > Software development process management; Software and its engineering > Software creation and management > Software verification and validation > Formal software verification; Software and its engineering > Software creation and management > Software verification and validation > Software defect analysis > Software testing and debugging; Software and its engineering > Software notations and tools; Software and its engineering > Software organization and properties > Software functional properties > Formal methods > Software verification; Theory of computation > Design and analysis of algorithms > Mathematical optimization; Theory of computation > Semantics and reasoning > Program reasoning > Program verification; Tuning; floating-point arithmetic; dynamic program analysis; mixed precision; delta-debugging algorithm; program optimization
• ISBN: 9781450323789; 1450323782
• ISSN: 2167-4329
• DOI: 10.1145/2503210.2503296

Given the variety of numerical errors that can occur, floating-point programs are difficult to write, test and debug. One common practice employed by developers without an advanced background in numerical analysis is using the highest available precision. While more robust, this can degrade program performance significantly. In this paper we present Precimonious, a dynamic program analysis tool to assist developers in tuning the precision of floating-point programs. Precimonious performs a search on the types of the floating-point program variables trying to lower their precision subject to accuracy constraints and performance goals. Our tool recommends a type instantiation that uses lower precision while producing an accurate enough answer without causing exceptions. We evaluate Precimonious on several widely used functions from the GNU Scientific Library, two NAS Parallel Benchmarks, and three other numerical programs. For most of the programs analyzed, Precimonious reduces precision, which results in performance improvements as high as 41%.