CUDA Based Parallel Computation for Gauss Elimination Method
Xiao Liu, Lei Xu · 2018
The Central Processing Unit (CPU) parallel algorithm based on Computing Unified Device Architecture (CUDA) has shown great power of computing speedup ability. What performance will the new technique show in the field of structural computation? We choose the Gauss elimination method as the research object. In this study, the parallel Gauss elimination is realized in CUDA on GPU. Furthermore, we carry out two groups of numerical experiments. The first group investigates the effect of Matrix Bandwidths (MBs) and Node Numbers (NNs) on speedup ratio. The second one compares our method with the commercial software by analyzing two actual structural problems in ocean engineering.