Large scale finite element analysis using GPU parallel computing.

摘要

In the past years, graphic processing units have become a new abundant parallel computing resource on personal computers. In this work parallel computation of a typical case in finite element analysis for solids has been practiced. The solution of 3-D linear elastic static problems with 3 degree of freedom is fully implemented utilizing the current GPU technology. Discretization of the problem has been done for two cases of: Tetrahedral and Hexahedral elements. Acceleration of the solution has been realized using SIMT parallel algorithms. Sparse matrix storage formats as well as matrix-vector operations are investigated for optimum hardware utilization. Preconditioned conjugate gradient method has been fully implemented on the GPU device as the iterative solver. FEA GPU implementation is compared with the corresponding optimized serial version run on a conventional processor with the same technology for various mesh sizes, sparse matrix storage schemes, and choice of basis function.