Original scientific paper

Naïve matrix multiplication versus Strassen algorithm in multi-thread environment

Filip Belić ; Tvornica elektro opreme Belišće d.d., Ante Starčevića 1, HR-31551 Belišće, Croatia
Domagoj Ševerdija ; Department of Mathematics, J. J. Strossmayer University of Osijek, Trg Ljudevita Gaja 6, HR-31000 Osijek, Croatia
Željko Hocenski ; Faculty of Electrical Engineering, J. J. Strossmayer University of Osijek, Kneza Trpimira 2B, HR-31000 Osijek, Croatia

Abstract

In last few decades computational power of computers has greatly increased. Highest speeds and power are still reserved for super-computers, but high-speed computers have been available for home and amateur users for some time. Normal user most of the time uses only a small amount of computational resources available; even in cases of high-strain, a good part of these resources stays unused. This is partly a result of poor programming. Most of programmers still use single-threaded programming although platforms for parallel programming have been widely available for long time. This article describes using one such platform (.NET Framework) to decrease time needed for multiplication of matrices. This article tries to present what results can be achieved using common equipment and easily acquirable software.

Keywords

computing resources; multi-core; multi-thread; parallel programming; Strassen's algorithm

Hrčak ID:

71808

URI

https://hrcak.srce.hr/71808

Publication date:

26.9.2011.

Article data in other languages: croatian

Visits: 3.673 *