Parallel Solvers on the GPU for Large-Scale Finite Element Equations

Abstract:

We present a parallel solution based on CUDA for accelerating the computation for solving large-scale Finite Element equations in electrical and magnetic field. JCG is used for solving equations and corresponding kernel function is designed for spMV. A computation speed test for solving FE equations is taken on NVIDIA Tesla K20c GPU hardware platform, the result proves that the method of kernel can reach 17.1 times faster than the solution using CPU, however it cannot ensure the advantage with CPU if we only use the lib functions on GPU to solve equations.

• Periodical: Applied Mechanics and Materials (Volumes 701-702)
• Pages: 207-213
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.701-702.207
• Online since: December 2014
• Authors: Qing Hu Zhang*, Dong Wang, Ya Peng Jiang, Jun Quan Chen
• Keywords: CUDA, Finite Element Equations, GPU, Jacobi Conjugate Gradient Method, Parallel

References

[1] Xie Dexin. The finite element analysis of three-dimensional eddy current field [M]. Mechanical industry press, 2008. (In Chinese).

[2] Maryam Mehri Dehnavi, David M. Fernandez, Dennis Giannacopoulos. Finite-Element Sparse Matrix Vector Multiplication on Graphic Process Units[J]. IEEE TRANSACTIONS ON MAGNETICS, 2010, 46(8): 2982-2985.

DOI: 10.1109/tmag.2010.2043511

[3] David M. Fernandez, Maryam Mehri Dehnavi, Warren J. Gross, Dennis Giannacopoulos. Alternate Parallel Processing Approach for FEM[J]. IEEE TRANSACTIONS ON MAGNETICS, 2012, 48(2): 399-402.

DOI: 10.1109/tmag.2011.2173304

[4] Nvidia. NVIDIA CUDA C programming guide.

[5] Rob Farber. CUDA Application Design and Development[M]. Mechanical industry press, (2013).

[6] Zhang Lan. Research on Preconditioning Iteration Technique for Sparse Matrix Equations [D]. Guangzhou：South China university of technology, 2010. (In Chinese).

[7] Nathan Bell, Michael Garland. Efficient Sparse Matrix-Vector Multiplication on CUDA[R]. Santa Clara, California: NVIDIA, (2008).

[8] Wu Changjiang. Design of the Solver for Large Scale Linear Sparse Equations Based on CUDA[D]. Chengdu：University of electronic science and technology, 2013. (In Chinese).

[9] Zhang Jianfei, Shen Defei. GPU-based preconditioned conjugate gradient method for solving sparse linear systems[J]. Journal of Computer Application, 2013, 33(3): 825-829. (In Chinese).

DOI: 10.3724/sp.j.1087.2013.00825

[10] David B. Kirk, Wenmei Hwu. Programming Massively Parallel Processors: A Hands-on Approach, second edition[M]. Tsinghua university press, (2013).

[11] Shane Cook. CUDA Programming: A Developer's Guide to Parallel Computing with GPUs[M]. Tsinghua university press, (2014).

[12] University of Florida. The University of Florida Sparse Matrix Collection[DB/OL]. [2014-04-03]. http: /www. cise. ufl. edu/research/sparse/matrices.