Abstract: Eliminating the executing bottleneck of intensive computational modules and increasing the processing size of large datasets is the main problems for the modular visualization environment. In this paper, inter module data paralleling and pipeline processing are presented based on data decomposition and composition, and the result of performance analysis shows these techniques can help to solve the above problems.