• norsk
    • English
  • norsk 
    • norsk
    • English
  • Logg inn
Vis innførsel 
  •   Hjem
  • Faculty of Mathematics and Natural Sciences
  • Department of Informatics
  • Department of Informatics
  • Vis innførsel
  •   Hjem
  • Faculty of Mathematics and Natural Sciences
  • Department of Informatics
  • Department of Informatics
  • Vis innførsel
JavaScript is disabled for your browser. Some features of this site may not work without it.

Improving Parallel Sparse Matrix-vector Multiplication

Tessem, Torbjørn Johnsen
Master thesis
Thumbnail
Åpne
116259352.pdf (720.0Kb)
Permanent lenke
https://hdl.handle.net/1956/8024
Utgivelsesdato
2013-12-19
Metadata
Vis full innførsel
Samlinger
  • Department of Informatics [747]
Sammendrag
Sparse Matrix-vector Multiplication (SMvM) is a mathematical technique encountered in many programs and computations and is often heavily used. Solving SMvM in parallel allows for bigger instances to be solved, and problems to be solved faster. Several strategies have been tried to improve parallel SMvM. Work has been done with regard to improved cache use, better load balance and reduced conflicts. The aim of the work conducted in this thesis is to develop new ideas and algorithms to speed-up parallel SMvM on a shared memory computer. We use a method inspired by the min-makespan problem to distribute elements more evenly. We introduce a hybrid algorithm that gives better cache efficiency, and we work with colouring algorithms to avoid write conflicts.
Utgiver
The University of Bergen
Opphavsrett
Copyright the author. All rights reserved

Kontakt oss | Gi tilbakemelding

Personvernerklæring
DSpace software copyright © 2002-2019  DuraSpace

Levert av  Unit
 

 

Bla i

Hele arkivetDelarkiv og samlingerUtgivelsesdatoForfattereTitlerEmneordDokumenttyperTidsskrifterDenne samlingenUtgivelsesdatoForfattereTitlerEmneordDokumenttyperTidsskrifter

Min side

Logg inn

Statistikk

Besøksstatistikk

Kontakt oss | Gi tilbakemelding

Personvernerklæring
DSpace software copyright © 2002-2019  DuraSpace

Levert av  Unit