针对传统的并行核外编程复杂、虚存系统性能不理想的问题,提出一种基于PVFS的并行核外计算方法,描述了其软件层次结构、数据访问策略及实现,为工作站机群用户提供了方便的编程接口以及核外数据的全局视图.实验结果表明,该方法扩展了可有效解决的核外问题规模,大数据量时的计算性能优于虚存系统. Aiming at the problem that the traditional parallel extranuclear programming is complex and the performance of the virtual storage system is not satisfactory, a PVFS-based parallel out-of-core computing method is proposed. Its software hierarchy, data access strategy and implementation are described. And the global view of the data outside the core.The experimental results show that this method expands the scale of the out-of-core problem that can be effectively solved, and the computational performance of the large data volume is better than that of the virtual memory system.