Efficient data organization for large scale parallel numerical simulation
面向大规模并行数值模拟的一种高效数据组织方式

A new efficient data organization method was introduced in this paper which was based on a dataset I/O module developed for resolving the storing overhead problem of scientific simulation.The module exported Visualization Toolkit(VTK) XML data files for the simulation that supports random access,and portable data compression.The storing overhead was reduced significantly as well as the visualization speed was improved remarkably.The experimental results show that the data storage cost is reduced 70 percent and time consumption of loading data into the visualization system is reduced by nearly 90 percent.Another superiority of the data organization method is that out-of-core technology can be introduced into the visualization naturally to improve the data processing efficiency.