In order to improve the parallel efficiency of large-scale structural dynamic analysis, a hierarchical approach adapted to the hardware topology of multi-core clusters is proposed. The hierarchical approach is constructed based on the strategies of two-level partitioning and two-level condensation. The data for parallel computing is first prepared through twolevel partitioning to guarantee the load balancing within and across nodes. Then during the analysis of each time step, the convergence rate of interface problem is significantly improved by further reducing its size with two-level condensation. Furthermore, the communication overheads are considerably reduced by separating the intra-node and inter-node communications and minimizing the inter-node communication. Numerical experiments conducted on Dawning-5000A supercomputer indicate that the hierarchical approach was superior in performance compared with the conventional Newmark algorithm based on the domain decomposition method.
Summary
Clusters with the CPU‐MIC heterogeneous architecture are becoming more popular in recent years. However, it is not easy to achieve good performance on such machines. The key challenge has been the asymmetry within clusters, arising from different kinds of execution units as well as different communication latencies. To improve the performance of large‐scale structural analysis on CPU‐MIC heterogeneous clusters, a multi‐layer and multi‐grain collaborative parallel computing approach is proposed in the paper. The proposed method combines the parallel algorithm and the hardware architecture of CPU‐MIC heterogeneous clusters together. Through mapping computing tasks to various hardware layers, it both resolves the load balance problem between CPU and MIC devices and significantly reduces the communication overheads of the system. Numerical experiments conducted on Tianhe‐2 supercomputer show that the proposed method obtained better performance compared with the traditional approach. Scalability investigation showed that the proposed method had good scalability with respect to problem sizes. The findings of this paper are of help to the parallel porting and performance optimization of other applications on CPU‐MIC heterogeneous clusters.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.