科学计算可视化并行处理建模及其数据并行技术研究

Modeling of Parallel Visualization in Scientific Computing and Its Realization Technologies in Data Parallel Mode

科学计算规模的不断增长使得对科学数据的可视化也必须采用并行处理技术。针对可视化应用的特点,一方面,分类建立了可视化并行处理模型,给出了形式化描述,并对其并行性进行了定量的分析;另一方面,提出了一种可扩展的可视化数据并行体系结构,并深入探讨了实现该结构的数据划分技术和并行绘制方法;最后基于上述理论实现了一个数据并行可视化处理软件原型系统YH-PView,并通过实验验证了理论分析的正确性。

Parallel processing is an inevitable choice of visualization in scientific computing for it will be oriented with much larger scale data than before. According to features of visualization applications, the parallel visualizing models that are classified by parallel mode were built and formalized. And for every parallel mode of visualization, its performance was analyzed quantitatively. Then, a scalable data parallel architecture for visualization in scientific computing was proposed, and some crucial techniques of data partition and strategies of parallel rendering were discussed. At last, a data parallel-based visualization system YH-PView was implemented, and some experiment results were given.