POSIX threads polynomials(PTPol): a scalable implementation of univariate arithmetic operations

In this paper, we describe the design of a C library named PTPol implementing arithmetic operations for univariate polynomials and report on practical experiments showing the relevance of using threads on recent multi-core computers. We show how to use efficiently an API named OpenMP and POSIX Threads to achieve scalability. On multi-core architectures, we obtain a speed-up equivalent to the number of cores on addition and multiplication on univariate polynomials for degrees a completer. Categories and Subject Descriptors G.0 [Mathematics of Computing]: General General Terms Algorithms, application Keywords Univariate polynomial, symbolic-numeric computation, scalability

• Type: Conference
• Year: 2007
• Where: ISSAC
• Authors: Mohab Safey El Din, Philippe Trebuchet
• Keywords: Api Named Openmp, ISSAC 2007, Keywords Univariate Polynomial, Univariate Polynomial