一种动态分布数组的数据划分模式

数据划分是分布主存系统中并行编译的关键技术，它以教组和包含这些教组的嵌套循环为研究对象，以提高教据局部性和挖掘计算并行性为根本目的。对满足给定模式的动态分布的教组向量，通过选取代表元，给出数据划分模式。将单个嵌套循环内的数据划分技术和过程间投影技术很好地结合，解决了动态分布教组的数据划分问题。这种模式弥补了现有数据划分研究的不足。

Data partition mode for dynamic array distributions

Data partition is the key technology of parallel compiler on distributed memory parallel computers (DMPCs). The objects of data partition arearrays and nested loop including these arrays . The goalof data partition istoimprovelocality ofdata and mine computing parallel. Aimingat arbitrary numberarrays which aredynamic distributions , a general data partition model is presentedthrough choosing a representation, in which the data partition technologies of a single nested loop and the technology of procedure analysis are integrated in order to solve the data partition problem of dynamic array distributions. This model is a complementarity to previous research about data partition.