Parallel multi-projection type methods on hybrid CPU/MIC cluster

Research output: Chapter in Book/Report/Conference proceedingsConference proceedingpeer-review

Abstract

Many problems occurring in various scientields require the ecient solution of large sparse linear systems derived from the discretization of partial di.erential equations. Preconditioned Krylov subspace iterative methods based on domain decomposition techniques are suitable for solving large sparse linear systems on parallel systems. A parallel preconditioned iterative method in conjunction with semi-Aggregation based algebraic domain decomposition method for symmetric sparse linear systems is presented. .e proposed method is designed for distributed memory systems with multicore nodes, equipped with many integrated core architecture co-processors (Intel Xeon Phi). Utilizing the MIC architecture co-processors, concurrently with existing CPUs, for solving the local linear systems results in accelerating the solution process. Moreover, for large number for subdomains the proposed parallel scheme has improved convergence behavior. .e convergence behavior and the scalability of the proposed scheme are examined and numerical results are given.

Original languageEnglish
Title of host publicationProceedings - 21st Pan-Hellenic Conference on Informatics, PCI 2017
PublisherAssociation for Computing Machinery
ISBN (Electronic)9781450353557
DOIs
Publication statusPublished - 28 Sep 2017
Externally publishedYes
Event21st Pan-Hellenic Conference on Informatics, PCI 2017 - Larissa, Greece
Duration: 28 Sep 201730 Sep 2017

Publication series

NameACM International Conference Proceeding Series
VolumePart F132523

Conference

Conference21st Pan-Hellenic Conference on Informatics, PCI 2017
Country/TerritoryGreece
CityLarissa
Period28/09/1730/09/17

Keywords

  • Aggregation
  • Algebraic domain decomposition
  • High performance computing
  • Many integrated core architecture
  • Parallel solver
  • semi-coarsening

Fingerprint

Dive into the research topics of 'Parallel multi-projection type methods on hybrid CPU/MIC cluster'. Together they form a unique fingerprint.

Cite this