Work

GPU-Accelerated Iterative Refinement Based on Induced Dimension Reduction

Year: 2024

Type: conference-paper

Abstract: The resolution of large scale linear systems is still a significant challenge in scientific computing.In this paper we consider the induced dimension reduction method and iterative refinement scheme for solving nonsymmetric linear systems.We formulate an IR outer scheme based on the IDR inner solver and perform a numerical study of its finite precision behavior.The algorithm is implemented on a graphics processing unit using the CUDA programming tool.

Source: Civil-comp conferences

DOI: 10.4203/ccc.8.5.4

Authors: Y. Jiang, F. Magoulès, X. Wang, Q. Zou

Institutions: Beijing University of Posts and Telecommunications, Université Paris-Saclay, CentraleSupélec

Language: English

Topic: Matrix Theory and Algorithms

Subfield: Computational Theory and Mathematics

Field: Computer Science

Domain: Physical Sciences

Open Access status: bronze

Funder: National Natural Science Foundation of China

Awards: G4806190252, G4872182317, G8939148000