Summary: Portable Parallel Implementation of BLAS3
appeared in Concurrency Practice and Experience, Vol. 6, No. 5, pp. 411­459, 1994
Dganit Amitai
Amir Averbuch
Ronen Friedman
Eran Gabber
Department of Computer Science
School of Mathematical Sciences
Sackler Faculty of Exact Sciences
Tel­Aviv University
Tel­Aviv 69978, Israel
Abstract
Multiprocessors systems offer fast enough capabilities for handling large numerical
tasks, such as real­life linear algebra problems. Yet, rogramming such systems has
proven itself to be awkward, error prone and architecture specific.
One successful method for alleviating this problem, a method that worked well in the
case of the massively pipelined supercomputers, is to use subprograms libraries. Those
libraries are built to efficiently perform some basic operations, while hiding low­level
system specifics from the programmer. Efficiently porting a library to a new hardware,
be it a vector machine or a shared­memory or message­passing based multiprocessor,

Source: Averbuch, Amir - School of Computer Science, Tel Aviv University

Collections: Computer Technologies and Information Sciences