中文摘要：

针对处理机具有任意释放时间的异构并行与分布式系统，通过寻找最优的处理机调度顺序和任务分配方案，使得任务的完成时间达到最短．首先，在给定处理机调度顺序的情况下，分析了处理机释放时间对任务完成时间的影响，得到了任务分配方案关于处理机调度顺序和时序约束条件的解析解；然后，以调度顺序和时序约束条件为变量，以任务的最短完成时间为目标，建立了一种新的可分任务调度模型；最后，设计了高效的全局优化遗传算法求解该模型．仿真结果表明所提算法比已有算法完成任务的时间更短．

英文摘要：

For heterogeneous parallel and distributed systems with arbitrary processor release times, the make-span was minimized by finding the optimal distribution sequence of processors and the opti- mal load partition. First, the influence of processor release times on the make-span was analyzed with a given distribution sequence of processors, and the function of load partition with respect to the dis- tribution sequence and time constraints was derived. Second, a new divisible-load scheduling model was proposed with the distribution sequence and time constraints as variables and the minimum make- span as the objective. Third, an effective global optimization genetic algorithm was designed to solve this model. Finally, experimental results show that the proposed algorithm outperforms the existing algorithms in finding the minimum make-span.