根据我们课题情况,有关数据调度问题,考虑从计算方法和计算技巧两方法着手解决。由于巨型机有大容量存貯器,运算速度快,多砖能部件并行工作和流水线向量化技术的特色。本来适合于串行计算的算法,原原本本地移植到巨型机去不是很妥当的,巨型机提供的方便,尽可能发挥巨型机的特长,还得选择适当的有效方法,才可合理解决数据调度问题。 According to our subject situation, the data scheduling problem, consider from the calculation methods and computational techniques to solve the two methods. As the giant machine has large capacity memory, fast computing speed, multi-brick components can work in parallel and pipeline vectorization technology features. Originally suitable for serial computing algorithms, originally originally transplanted to the supercomputer to go is not appropriate, the supercomputer provides convenience, as far as possible to play the supercomputer specialty, have to choose the appropriate effective method, can be a reasonable solution to the data scheduling problem .