Dynamic floating-point cancellation detection

Authors:
Michael O. Lam;Jeffrey K. Hollingsworth;G. W. Stewart
Affiliations:
Dept. of Computer Science, University of Maryland, A.V. Williams Building, College Park, MD 20742, United States;Dept. of Computer Science, University of Maryland, A.V. Williams Building, College Park, MD 20742, United States;Dept. of Computer Science, University of Maryland, A.V. Williams Building, College Park, MD 20742, United States
Venue:
Parallel Computing
Year:
2013

Citing 7
Cited 2

On Local Roundoff Errors in Floating-Point Arithmetic

Journal of the ACM (JACM)
Pracniques: further remarks on reducing truncation errors

Communications of the ACM
Accuracy and Stability of Numerical Algorithms

Accuracy and Stability of Numerical Algorithms
A Priori Worst Case Error Bounds for Floating-Point Computations

IEEE Transactions on Computers
An API for Runtime Code Patching

International Journal of High Performance Computing Applications
Program transformation for numerical precision

Proceedings of the 2009 ACM SIGPLAN workshop on Partial evaluation and program manipulation
Semantics-based transformation of arithmetic expressions

SAS'07 Proceedings of the 14th international conference on Static Analysis

On-the-fly detection of instability problems in floating-point program execution

Proceedings of the 2013 ACM SIGPLAN international conference on Object oriented programming systems languages & applications
Efficient search for inputs causing high floating-point errors

Proceedings of the 19th ACM SIGPLAN symposium on Principles and practice of parallel programming

Quantified Score

Hi-index	0.00

Visualization

Abstract

As scientific computation continues to scale, it is crucial to use floating-point arithmetic processors as efficiently as possible. Lower precision allows streaming architectures to perform more operations per second and can reduce memory bandwidth pressure on all architectures. However, using a precision that is too low for a given algorithm and data set will result in inaccurate results. Thus, developers must balance speed and accuracy when choosing the floating-point precision of their subroutines and data structures. We are building tools to help developers learn about the runtime floating-point behavior of their programs, and to help them make implementation decisions regarding this behavior. We propose a tool that performs automatic binary instrumentation of floating-point code to detect mathematical cancellations. In particular, we show how our prototype can detect the variation in cancellation patterns for different pivoting strategies in Gaussian elimination, as well as how our prototype can detect a program's sensitivity to ill-conditioned input sets.