Parallel Implementation Of Finite Element Codes Using CUDA
碩士 === 國立中正大學 === 應用數學研究所 === 100 === The purpose of this work is to study the performance of parallel computation of Finite Element Method using the NVIDIA's CUDA. The numerical experiments are performed only on the stiffness matrix using the conjugate gradient method. In addition, the general...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2012
|
Online Access: | http://ndltd.ncl.edu.tw/handle/12869409771252157321 |
Summary: | 碩士 === 國立中正大學 === 應用數學研究所 === 100 === The purpose of this work is to study the performance of parallel computation of Finite Element Method using the NVIDIA's CUDA. The numerical experiments are performed only on the stiffness matrix using the conjugate gradient method. In addition, the generalized minimal residual method is considered to solve the Stokes problem using both PETSc and CUDA. The speed up related to the data size will be demonstrated. Future improvements of the parallel computations will also be addressed.
|
---|