In this paper, from application viewpoint, a model for enhancing parallel computing performance is established. The performance of parallel computing is analyzed when different domain divisions affect convergence speed in solving the whole problem. In addition, the major factors affecting the performance of massive data parallel computing are analyzed, such as overlapping operation,the problem data size and selective algorithm. Finally some computational examples provide evidence for the usefulness of the model.