Employing graphics processing unit technology, alternating direction implicit method and domain decomposition to speed up the numerical diffusion solver for the biomedical engineering research

• Published in: International journal for numerical methods in biomedical engineering Vol. 27; no. 11; pp. 1829 - 1849
• Main Authors: Jiang, Beini, Struthers, Allan, Sun, Zhe, Feng, Zhuo, Zhao, Xuqian, Zhao, Kaiyong, Dai, Weizhong, Zhou, Xiaobo, Berens, Michael E., Zhang, Le
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.11.2011; Wiley
• Subjects: Algorithms; alternating direction implicit method (ADI); Biological and medical sciences; Biomedical engineering; Cell motion; Cell physiology; Computation; Computational techniques; Computer simulation; Diffusion; domain decomposition; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; graphics processing unit (GPU); Mathematical methods in physics; Mathematical models; Mathematics; Methods of scientific computing (including symbolic computation, algebraic computation); Molecular and cellular biology; Motility and taxis; Nodular iron; Numerical analysis. Scientific computation; parallel computing; Physics; Plant physiology and development; Sciences and techniques of general use; Solvers; Bioengineering; Alternating direction method; Bottleneck; Scattering; Experimental study; Parallel algorithms; Distributed computing; Chemotaxis; alternating direction implicit method (ADI); Graphic processing unit; Biotransport; Parallel processing; graphics processing unit (GPU); Parallelism; Parallel computation; Modelling; Fast algorithm; Domain decomposition; parallel computing; Biomedical engineering
• ISSN: 2040-7939; 2040-7947; 2040-7947
• DOI: 10.1002/cnm.1444

Diffusion of biological compounds, including nutrients, oxygen, and chemoattractants, is a common component of biomedical engineering models. Conventional numerical schemes, such as alternating direction implicit (ADI) for diffusion, are frequently the computational bottleneck. Our study employs graphics processing unit (GPU) technology to accelerate the diffusion model simulation. We tailor, implement, analyze, and test several parallel ADI algorithms on the highly parallel computational and data architecture of the GPU. Our study confirms that the proposed algorithms provide fast, high‐quality simulation results suitable for inclusion in numerous bioengineering simulations. Copyright © 2011 John Wiley & Sons, Ltd.