DATA-PARALLEL EVALUATION OF UNIVARIATE POLYNOMIALS BY THE KNUTH-EVE ALGORITHM

The performance of the Knuth-Eve algorithm for data parallel evaluatio n of univariate polynomials of degree 8, 16 and 32 has been systematic ally compared with that of the classical Newton-Horner algorithm using three vector processors and three processor arrays, Significant perfo rmance improvements have been observed beyond a critical problem size, as predicted by theory. The practical implications of this result for large-scale computations involving function approximation are outline d. (C) 1997 Elsevier Science B.V.