基于 VisIt与地球系统格网的并行可视化实验

利用集群架构和分布式并行可视化工具 VisIt,编写了自定义插件,实现了基于大规模地球系统格网组织下的全球科学数据并行可视化,并设计实验对其并行可视化性能进行了对比分析.实验发现: VisIt完成一次渲染的加速比及并行效率随着核数的增加逐渐降低;采用 GPU 渲染,可以很好地提高并行渲染的效率.但在核数和 GPU 个数同步增加的情况下,由于核间通信、 GPU 间通信以及核- GPU 间通信等,VisIt一次渲染的并行运行时间并无明显降低.随着数据量增加,VisIt对单位数据量的运行时间却逐渐减低.实验表明,VisIt可较高效地完成大数据量的并行渲染.该方法和结论可供地学领域大规模海量数据可视化研究参考.

Parallel Visualization of Earth System Grid Data Based on VisIt and Its Performance Analysis

Employing the cluster and distribute parallel visualization tool-VisIt,user defined plug-in was utilized.Based on earth system spatial grid,its organized global scientific data was efficiently and effectively visualized.Experiment on visual ability of VisIt was designed and carried out on the laboratory cluster.After comparative performance analysis,the following conclusions were given: the speedup and parallel efficiency of VisIt once rendering were decreased with increasing cores;the parallel efficiency was dramatically improved by using GPU rendering,however,due to communication of cores,GPUs,and cores-GPUs,running time of VisIt once rendering was not obvious reduction with synchronously increase of GPUs and cores.With enlarging data size,running time of unit data size was gradually decreased.It demonstrated that VisIt was capable of large scale data parallel rendering.Consequently,the methodology and conclusion provided an instructive exploration and reference for large scale data size.