# Floating Point Operation

## Abstract

The range of numbers available in a digital computer word as discussed so far is strictly limited. A 32-bit number has a range of about 2^{32} or 10^{10} numbers. If the numbers are regarded as integers, then it is necessary to scale many problems in order to represent fractions. The scale used by astronomers and atomic physicists to represent length and mass, for instance, would be totally different. Even so, some 9 decimal digits are not sufficient for some problems (±10^{9} = 10^{10} numbers). One might consider increasing the number of words to represent each number to get a better range, but this is wasteful of storage space and of computing time, since most of the time one now computes to more digits of accuracy than are required. The solution is to use floating-point notation, sometimes referred to as’ scientific’ notation in connection with pocket calculators.

## Keywords

Digital Computer Decimal Digit Arithmetic Unit Float Point Operation Logic Stage## Preview

Unable to display preview. Download preview PDF.

## References

- Burks, A. W., Goldstine, H. H., and von Neumann, J., ‘Preliminary Discussion of the Logical Design of an Electronic Computing Instrument’, in
*Collected Works of John von Neumann*, vol. 5 (Pergamon, Oxford, 1963) pp. 34–79.Google Scholar - Gosling, J. B., ‘Design of Large High Speed Floating Point Arithmetic Units’,
*Proc. I.E.E.*,**118**(1971) 493–8.CrossRefGoogle Scholar - Kahan, W., ‘Implementation of Algorithms’, University of California, Berkeley, AD-769 124 (1973). Available from Microinfo Ltd, PO Box 3, Hamlet House, Alton, Hants, U.K.Google Scholar
- Kuck, D. J., Parker, D. S., and Sameh, A. H., ‘Analysis of Rounding Methods in Floating Point Arithmetic’,
*I.E.E.E. Trans. Comput.*,**26**(1977) 643–50.MathSciNetCrossRefMATHGoogle Scholar - Metropolis, N., and Ashenhurst, A. L., ‘Basic Operations in Unnormalised Arithmetic System’,
*I.E.E.E. Trans. electronic Comput.*,**12**(1963) 896–904.CrossRefMATHGoogle Scholar - Sterbenz, P.,
*Floating Point Computation*(Prentice-Hall, Englewood Cliffs, N.J., 1974). An excellent book, but limited to IBM and Fortran.Google Scholar - Sweeney, D. W., ‘An Analysis of Floating Point Arithmetic’,
*IBM Syst. J.*,**4**(1965) 31–42. Excellent article.CrossRefGoogle Scholar - Wilkinson, J. H.,
*Rounding Errors in Algebraic Processes*(Prentice-Hall, Englewood Cliffs, N.J., 1963).MATHGoogle Scholar