Scalable High Performance Libraries for Large Scale Scientific Computing

Organizers:
A. Murli, Università di Napoli "Federico II", Italy
P. Messina, California Institute of Technology, Pasadena, USA

Summary: In the last few years, many advances have been made in the development of parallel algorithms for scalable architectures. However, one of the main difficulties for the scientific community in effectively using these architectures is the lack of software libraries (both mathematical software and graphics routines). Presently only few routines, mainly of linear algebra, are available (see the ScaLAPACK project).

The aim of this minisymposium is to give a perspective of the development of Scalable High Performance Libraries for large scale scientific computing. The speakers will describe recent results in the field of mathematical software for advanced architectures in some of the main areas of Computational Mathematics (linear algebra, multidimensional quadrature, ..). The minisymposium is addressed to all scientists interested in solving scientific problems on advanced architectures.

Day: July 5, 1995 Time: 15.30

J. Demmel, University of California at Berkeley, USA:
The scalapack linear algebra library for distributed memory machines

J. Dongarra, The University of Tennessee, USA:
Software tools and templates for developing a software library

S. Hammarling, The Numerical Algorithm Group Ltd, United Kingdom:
Development of a parallel library for distributed memory machines

P.C. Messina, California Institute of Technology, USA:
System functions in support of mathematical software libraries on parallel computing environment

J.N. Lyness, Argonne National Laboratory, USA:
Guidelines for parallel software in numerical quadrature


Last modified: 2 June 1995
Imprint