Fast normal random number generators on vector processors

141. R. P. Brent, Fast normal random number generators on vector processors, Technical Report TR-CS-93-04, Computer Sciences Laboratory, ANU, March 1993, 6 pp. arXiv:1004.3105v2

Abstract: dvi (3K), pdf (63K), ps (27K).

Technical Report: dvi (19K), pdf (135K), ps (56K).


We consider pseudo-random number generators suitable for vector processors. In particular, we describe vectorised implementations of the Box-Muller and Polar methods, and show that they give good performance on the Fujitsu VP2200. We also consider some other popular methods, e.g. the Ratio method of Kinderman and Monahan (1977) (as improved by Leva (1992)), and the method of Von Neumann and Forsythe (see [123]), and show why they are unlikely to be competitive with the Polar method on vector processors.


For a significantly faster method, based on an idea of Wallace, see [170, 185].

Go to next publication

Return to Richard Brent's index page