EFFECT OF MEMORY ON THE PERFORMANCE OF ONE-STEP RECURSION PARALLEL STRASSEN’S ALGORITHM ON DUAL CORE PROCESSOR

Abstract

The main aim of this research work is to compare the speedup obtainable from executing One-
Level Recursion parallel Strassen’s matrix multiplication algorithm on a dual core with 2G and
4G memories relative to parallel conventional algorithm. This is to determine the effect of
additional memory on the performance of the algorithm. The speedup obtainable by Strassen’s
Algorithm is at the expense of additional memory space for intermediate matrices. Even though
additional memory improve performance, there is need to experimentally investigate how
additional memory influence the performance of the new algorithm. The execution time of
parallel implementations of the algorithms on dual core processor with 2G and 4G memories
were measured, analyzed and interpreted. The programs were written in C++/OpenMP. The
result of the research work showed a minor improvement in performance of the parallel One-
Level Recursion Strassen’s algorithm for large matrix size on the 4G memory. It was also
observed that conventional algorithm did not show any significant performance improvement
with increased memory. The result suggested that with large amount of memory One-Level
Recursion, Strassen’s algorithm performs a little better for larger matrices. The result will serve
as a motivation for further research to determine the role played by cache memory in
performance of the new algorithm.