University of Twente Student Theses


Sparse matrix vector multiplication on a field programmable gate array

Veen, Marcel van der (2007) Sparse matrix vector multiplication on a field programmable gate array.

[img] PDF
Abstract:Sparse Matrix Vector Multiplication (SMVM) has been a subject of research in the computer science field quite some time. The SMVM is an expensive operation used in the Conjugate Gradient algorithm. The Conjugate Gradient is an iterative algorithm used to solve linear equations. The Finite Element Method is used in a lot of engineering areas like structural analysis, fluid dynamics, heat transport and electromagnetism. Within the Finite Element Method, the Conjugate Gradient algorithm is used to solve large sets of linear equations. Within the research field of SMVM there are two directions, one direction tries to get the most out of general purpose processors while the other direction tries to make a fast implementation on a Field Programmable Array. This project introduced a new method for the SMVM on a Field Programmable Gate Array called Small Bandwidth Coverage (SBC).The main advantage of SBC is that the variation in performance caused by differences in the system matrices is only a factor two. Similar solutions have a variation in performance of a factor ten.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science MSc (60300)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page