GPU/CPU Parallel Computation of Material Damage
University of Michigan – Dearborn, Michigan, U.S.A.
University of Michigan – Dearborn, 2014
@article{shen2014gpucpu,
title={GPU/CPU Parallel Computation of Material Damage},
author={Shen, Jie and Vela, Diego and Singh, Ankita and Song, Kexing and Zhang, Guoshang and LaFreniere, Bradon and Chen, Hao},
year={2014}
}
In this paper CUDA (Compute Unified Device Architecture) programming and OpenMP (Open Multi-Processing) are used for the GPU (Graphics Processing Unit) and CPU (Central Processing Unit) parallel computation of material damage. The material damage is evaluated by a multilevel finite element analysis within material domains reconstructed from a high-resolution micro-focus X-ray computed tomography system. An effective computational method is investigated for solving the linear equations of finite element analysis. Numerical results show an encouraging trend in reducing the computation cost for the digital diagnosis of material damage.
June 19, 2014 by hgpu
