A Fast Mixed-Precision Strategy for Iterative GPU-Based Solution of the Laplace Equation

Stefan Lemvig Glimberg

A Fast Mixed-Precision Strategy for Iterative GPU-Based Solution of the Laplace Equation

Stefan Lemvig Glimberg (Author)

Research output: Non-textual form › Sound/Visual production (digital) › Research

66 Downloads (Pure)

Abstract

Our work is concerned with the development of a generic high-performance library for scientific computing. The library is targeted for assembling flexible-order finite-difference solvers for PDEs. Our goal is to enable fast solution of large PDE systems, fully exploiting the massively parallel architecture of Graphics Processing Units. We will detail a strategy for an iterative mixed-precision defect correction method, with p-multigrid preconditioning. We present a case study of the fully nonlinear potential flow equations, in which the bottleneck problem is finding the solution of a Laplace equation. Such large scale simulations can be of great value in coastal and offshore engineering application.

Original language	English
Publication date	2011
Publication status	Published - 2011
Event	LANS Informal Seminar - Argonne, Illinois, USA Duration: 1 Jan 2011 → …

Conference

Conference	LANS Informal Seminar
City	Argonne, Illinois, USA
Period	01/01/2011 → …

Access to Document

Slides glimbergFinal published version, 1.63 MB

http://www.mcs.anl.gov/research/LANS/events/listn/index.php

Fingerprint Dive into the research topics of 'A Fast Mixed-Precision Strategy for Iterative GPU-Based Solution of the Laplace Equation'. Together they form a unique fingerprint.

Cite this

@misc{fafaa2ddef6345d8a96a24a0f5ac23da,

title = "A Fast Mixed-Precision Strategy for Iterative GPU-Based Solution of the Laplace Equation",

abstract = "Our work is concerned with the development of a generic high-performance library for scientific computing. The library is targeted for assembling flexible-order finite-difference solvers for PDEs. Our goal is to enable fast solution of large PDE systems, fully exploiting the massively parallel architecture of Graphics Processing Units. We will detail a strategy for an iterative mixed-precision defect correction method, with p-multigrid preconditioning. We present a case study of the fully nonlinear potential flow equations, in which the bottleneck problem is finding the solution of a Laplace equation. Such large scale simulations can be of great value in coastal and offshore engineering application.",

author = "Glimberg, {Stefan Lemvig}",

year = "2011",

language = "English",

note = "LANS Informal Seminar ; Conference date: 01-01-2011",

}

TY - ADVS

T1 - A Fast Mixed-Precision Strategy for Iterative GPU-Based Solution of the Laplace Equation

A2 - Glimberg, Stefan Lemvig

PY - 2011

Y1 - 2011

N2 - Our work is concerned with the development of a generic high-performance library for scientific computing. The library is targeted for assembling flexible-order finite-difference solvers for PDEs. Our goal is to enable fast solution of large PDE systems, fully exploiting the massively parallel architecture of Graphics Processing Units. We will detail a strategy for an iterative mixed-precision defect correction method, with p-multigrid preconditioning. We present a case study of the fully nonlinear potential flow equations, in which the bottleneck problem is finding the solution of a Laplace equation. Such large scale simulations can be of great value in coastal and offshore engineering application.

AB - Our work is concerned with the development of a generic high-performance library for scientific computing. The library is targeted for assembling flexible-order finite-difference solvers for PDEs. Our goal is to enable fast solution of large PDE systems, fully exploiting the massively parallel architecture of Graphics Processing Units. We will detail a strategy for an iterative mixed-precision defect correction method, with p-multigrid preconditioning. We present a case study of the fully nonlinear potential flow equations, in which the bottleneck problem is finding the solution of a Laplace equation. Such large scale simulations can be of great value in coastal and offshore engineering application.

M3 - Sound/Visual production (digital)

T2 - LANS Informal Seminar

Y2 - 1 January 2011

ER -