云环境下的大规模线性有限元并行实现

针对Hadoop MapReduce框架实现迭代算法效率不高的问题,提出了基于Spark RDDs(Resilient Distributed Datasets)的大规模线性有限元并行算法,探索在云平台上有效地实现迭代算法。在Hadoop+Spark实验室集群上,通过空间桁架进行算例验证,并与基于Hadoop MapReduce的线性有限元并行算法进行性能比较。结果表明,在本文搭建的集群上,基于RDDs的并行算法能求解15000000个自由度的空间桁架问题,远大于Hadoop平台上的3000000个自由度;对于小模型,Spark可获得200倍以上的加速比,对于大模型,获得7~8倍加速比。

Parallel implement of large-scale linear elastic FEM in cloud computing environment

Considering the fact that iterative algorithm cannot be realized efficiently on Hadoop MapReduce platform, this paper proposed a large-scale finite-element parallel algorithm based on the Resilient Distributed Datasets on the Spark platform in order to explore how to implement iterative algorithm efficiently. The proposed algorithm was then verified using the space truss model on a 6-node Hadoop+Spark platform. Comparisons were made between a performance of Spark-based algorithms and Hadoop-based algorithms of linear elastic FEM. The results indicate that the number of the DOFs of the space truss problem that can be solved by the Spark-based parallel algorithm may reach 15000000, which is much more than that solved by the Hadoop-based parallel algorithm. Obviously, the Spark-based parallel algorithm is preferable. Moreover, the proposed algorithm exhibits an enhanced computing efficiency compared with the Hadoop-based parallel algorithm. Specifically, for a small-scale space truss model, the speed-up ratio reaches 200 while for a large-scale space truss model, it is approximately 7 or 8.