GPU-accelerated asynchronous error correction for mixed precision iterative refinement

Hartwig Anzt, Piotr Luszczek, Jack Dongarra, Vincent Heuveline

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

5 Scopus citations

Abstract

In hardware-aware high performance computing, block- asynchronous iteration and mixed precision iterative refinement are two techniques that may be used to leverage the computing power of SIMD accelerators like GPUs in the iterative solution of linear equation systems. Although they use a very different approach for this purpose, they share the basic idea of compensating the convergence properties of an inferior numerical algorithm by a more efficient usage of the provided computing power. In this paper, we analyze the potential of combining both techniques. Therefore, we derive a mixed precision iterative refinement algorithm using a block-asynchronous iteration as an error correction solver, and compare its performance with a pure implementation of a block-asynchronous iteration and an iterative refinement method using double precision for the error correction solver. For matrices from the University of Florida Matrix collection, we report the convergence behaviour and provide the total solver runtime using different GPU architectures.

Original languageEnglish
Title of host publicationParallel Processing - 18th International Conference, Euro-Par 2012, Proceedings
Pages908-919
Number of pages12
DOIs
StatePublished - 2012
Event18th International Conference on Parallel Processing, Euro-Par 2012 - Rhodes Island, Greece
Duration: Aug 27 2012Aug 31 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7484 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference18th International Conference on Parallel Processing, Euro-Par 2012
Country/TerritoryGreece
CityRhodes Island
Period08/27/1208/31/12

Keywords

  • GPU
  • block-asynchronous iteration
  • linear system
  • mixed precision iterative refinement
  • relaxation

Fingerprint

Dive into the research topics of 'GPU-accelerated asynchronous error correction for mixed precision iterative refinement'. Together they form a unique fingerprint.

Cite this