Date Feb 3, 2006
Speaker Ed Stewart (Visual Numerics)
Topic High Performance Computing with the IMSL Libraries using OpenMP and MPI
Abstract Over the past 35 years the computational capabilities of the IMSL Libraries have evolved along with the advances in high performance computing technology. From the "high-powered mainframes" of the late 1970s to the vector processor systems of the 1980s to the shared and distributed memory parallel systems of today, modifications have been incorporated in IMSL to adapt to the architectural changes. The changes are not limited to the improvements in the performance of the mathematical and statistical algorithms of IMSL but also in the usability of the libraries within a parallel computing environment. The usability features cover a wide range of computing tasks from the compilation step to programming techniques that significantly reduce debugging time. The presenter will discuss some of the programming issues that developers face today within parallel processing environments when using OpenMP and MPI protocols and how features in the IMSL Libraries help to alleviate some of these issues. Examples will be presented illustrating the use of these IMSL features.



We thank the generous support of MIT IS&T, CSAIL, and the Department of Mathematics for their support of this series.

MIT Math CSAIL EAPS Lincoln Lab Harvard Astronomy