Implementing linear algebra routines on multi-core processors with pipelining and a look ahead

Jakub Kurzak, Jack Dongarra

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

16 Scopus citations

Abstract

Linear algebra algorithms commonly encapsulate parallelism in Basic Linear Algebra Subroutines (BLAS). This solution relies on the fork-join model of parallel execution, which may result in suboptimal performance on current and future generations of multi-core processors. To overcome the shortcomings of this approach a pipelined model of parallel execution is presented, and the idea of look ahead is utilized in order to suppress the negative effects of sequential formulation of the algorithms. Application to one-sided matrix factorizations, LU, Cholesky and QR, is described. Shared memory implementation using POSIX threads is presented.

Original languageEnglish
Title of host publicationApplied Parallel Computing
Subtitle of host publicationState of the Art in Scientific Computing - 8th International Workshop, PARA 2006, Revised Selected Papers
PublisherSpringer Verlag
Pages147-156
Number of pages10
ISBN (Print)9783540757542
DOIs
StatePublished - 2007
Event8th International Workshop on Applied Parallel Computing, PARA 2006 - Umea, Sweden
Duration: Jun 18 2007Jun 21 2007

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4699 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference8th International Workshop on Applied Parallel Computing, PARA 2006
Country/TerritorySweden
CityUmea
Period06/18/0706/21/07

Fingerprint

Dive into the research topics of 'Implementing linear algebra routines on multi-core processors with pipelining and a look ahead'. Together they form a unique fingerprint.

Cite this