Subscribe

Issue No.02 - February (2009 vol.58)

pp: 197-207

Christoph Quirin Lauter , École Normale Supérieure de Lyon, Lyon

Vincent Lefèvre , LIP ENS Lyon INRIA, Lyon

DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/TC.2008.202

ABSTRACT

The correct rounding of the function pow: (x, y) \mapsto x^{y} is currently based on Ziv's iterative approximation process. In order to ensure its termination, cases when x^{y} falls on a rounding-boundary must be filtered out. Such rounding-boundaries are floating-point numbers and midpoints between two consecutive floating-point numbers. Detecting rounding-boundaries for pow is a difficult problem. Previous approaches use repeated square root extraction followed by repeated square and multiply. This paper presents a new rounding-boundary test for pow in double precision, which reduces this to a few comparisons with precomputed constants. These constants are deduced from worst cases for the Table Maker's Dilemma, searched over a small subset of the input domain. This is a novel use of such worst-case bounds. The resulting algorithm has been designed for a fast-on-average correctly rounded implementation of pow, considering the scarcity of rounding-boundary cases. It does not stall average computations for rounding-boundary detection. This paper includes its correctness proof and experimental results.

INDEX TERMS

Floating-point arithmetic, correct rounding, power function.

CITATION

Christoph Quirin Lauter, Vincent Lefèvre, "An Efficient Rounding Boundary Test for {\rm pow}(x, y) in Double Precision",

*IEEE Transactions on Computers*, vol.58, no. 2, pp. 197-207, February 2009, doi:10.1109/TC.2008.202REFERENCES

- [1]
Standard 754-1985 for Binary Floating-Point Arithmetic, ANSI/IEEE, 1985.- [2] A. Ziv, “Fast Evaluation of Elementary Mathematical Functions with Correctly Rounded Last Bit,”
ACM Trans. Math. Software, vol. 17, no. 3, pp. 410-423, Sept. 1991.- [3] J.-M. Muller,
Elementary Functions, Algorithms and Implementation. Birkhäuser, 1997.- [4] L. Fousse, G. Hanrot, V. Lefèvre, P. Pélissier, and P. Zimmermann, “MPFR: A Multiple-Precision Binary Floating-Point Library with Correct Rounding,”
ACM Trans. Math. Software, vol. 33, no. 2, June 2007.- [5] F. de Dinechin, A. Ershov, and N. Gast, “Towards the Post-Ultimate LIBM,”
Proc. 17th IEEE Symp. Computer Arithmetic (ARITH-17 '05), June 2005.- [6] F. de Dinechin, C.Q. Lauter, and J.-M. Muller, “Fast and Correctly Rounded Logarithms in Double-Precision,”
RAIRO, Theoretical Informatics and Applications, vol. 41, pp. 85-102, 2007.- [7] V. Lefèvre and J.-M. Muller, “Worst Cases for Correct Rounding of the Elementary Functions in Double Precision,”
Proc. 15th IEEE Symp. Computer Arithmetic (ARITH-15 '01), pp. 111-118, N.Burgess and L. Ciminiera, eds., http://www.ece.ucdavis. edu/acsel/arithmetic/ arith15/papersARITH15_Lefevre.pdf, 2001.- [8]
CRLibm, a Library of Correctly Rounded Elementary Functions in Double-Precision, http://lipforge.ens-lyon.fr/wwwcrlibm/, 2008.- [9] “Sun Microsystems,”
LIBMCR, a Reference Correctly-Rounded Library of Basic Double-Precision Transcendental Elementary Functions, http://www.sun.com/downloadproducts.xml?id=41797765 , 2008.- [10] D. Stehlé, V. Lefèvre, and P. Zimmermann, “Searching Worst Cases of a One-Variable Function Using Lattice Reduction,”
IEEE Trans. Computers, vol. 54, no. 3, pp. 340-346, http://csdl.computer. org/dl/trans/tc/2005/ 03t0340.pdf, Mar. 2005.- [11] The MPFR Team,
The MPFR Library: Algorithms and Proofs, http://www.mpfr.orggforge.html, July 2007.- [12] G.H. Hardy and E.M. Wright,
An Introduction to the Theory of Numbers. Oxford Univ. Press, 1979.- [13] V. Lefèvre, “New Results on the Distance between a Segment and${\hbox{\rlap{Z}\kern 2.0pt{\hbox{Z}}}}^{2}$ . Application to the Exact Rounding,”
Proc. 17th IEEE Symp. Computer Arithmetic (ARITH-17 '05), pp. 68-75, P. Montuschi and E. Schwarz, eds., http://arith17.polito.it/finalpaper-147.pdf , June 2005.- [14] P. Kornerup, V. Lefèvre, and J.-M. Muller,
Computing Integer Powers in Floating-Point Arithmetic, Laboratoire de l'Informatique du Parallélisme, Lyon, France, Research Report RR2007-23, http://www.vinc17.org/research/papersrr_power.pdf , May 2007.- [15] S. Boldo,
Preuves Formelles en Arithmétiques à Virgule Flottante, PhD dissertation, Ecole Normale Supérieure de Lyon, http://www. ens-lyon.fr/LIP/Pub/Rapports/ PhD/PhD2004PhD2004-05.pdf, Nov. 2004.- [16] A. Feldstein and R. Goodman, “Convergence Estimates for the Distribution of Trailing Digits,”
J. ACM, vol. 23, no. 2, pp. 287-297, http://portal.acm.orgcitation.cfm?id=321948 , Apr. 1976.- [17] C.E. Leiserson, H. Prokop, and K.H. Randall,
Using de Bruijn Sequences to Index a 1 in a Computer Word, http://supertech.csail. mit.edu/papersdebruijn.pdf , 1998. |