An error correction solver for linear systems: evaluation of mixed precision implementations

Authors:
Hartwig Anzt;Vincent Heuveline;Björn Rocker
Affiliations:
Karlsruhe Institute of Technology, Institute for Applied and Numerical Mathematics, Karlsruhe, Germany;Karlsruhe Institute of Technology, Institute for Applied and Numerical Mathematics, Karlsruhe, Germany;Karlsruhe Institute of Technology, Institute for Applied and Numerical Mathematics, Karlsruhe, Germany
Venue:
VECPAR'10 Proceedings of the 9th international conference on High performance computing for computational science
Year:
2010

Citing 6
Cited 2

Applied numerical linear algebra

Applied numerical linear algebra
Templates for the solution of algebraic eigenvalue problems: a practical guide

Templates for the solution of algebraic eigenvalue problems: a practical guide
Accuracy and Stability of Numerical Algorithms

Accuracy and Stability of Numerical Algorithms
Numerical Linear Algebra for High Performance Computers

Numerical Linear Algebra for High Performance Computers
Iterative Methods for Sparse Linear Systems

Iterative Methods for Sparse Linear Systems
Pipelined Mixed Precision Algorithms on FPGAs for Fast and Accurate PDE Solvers from Low Precision Components

FCCM '06 Proceedings of the 14th Annual IEEE Symposium on Field-Programmable Custom Computing Machines

Optimization of power consumption in the iterative solution of sparse linear systems on graphics processors

Computer Science - Research and Development
GPU-accelerated asynchronous error correction for mixed precision iterative refinement

Euro-Par'12 Proceedings of the 18th international conference on Parallel Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes an error correction method for solving linear systems of equations and the evaluation of an implementation using mixed precision techniques. While different technologies are available, graphic processing units (GPUs) have been established as particularly powerful coprocessors in recent years. For this reason, our error correction approach is focused on a CUDA implementation executing the error correction solver on the GPU. Benchmarks are performed both for artificially created matrices with preset characteristics as well as matrices obtained from finite element discretizations of fluid flow problems.