动态可重构系统任务时域划分算法

Temporal task partitioning algorithm for dynamically reconfigurable systems

为提高动态可重构系统的执行效率,提出了一种多目标优化的动态可重构系统任务时域划分方法.在可重构硬件资源约束条件下,将任务的数据流图划分成数个子模块.使用加权方法将子模块中的操作执行时间和模块间数据通信引起的延时统一为划分代价,根据每种延时在总代价中所占比例来选取优化的方向.通过建立就绪列表的方法,避免了划分过程中出现死锁问题,保证了各个操作正常的执行顺序.实验结果表明,该方法能够有效地提高应用算法在动态可重构结构上的执行速度.算法的时间复杂度为O(|V|2+|V||E|)(其中|V|为数据流图节点数,|E|为边数).

A temporal task partitioning algorithm for multi-objective optimization was proposed to improve the execution efficiency of dynamically reconfigurable systems.The task data flow graph(DFG) was partitioned into several sub-modules under hardware resource constraint.The delay cost of sub-modules and the communication cost between sub-modules were abstracted into one general cost by different weights to trade-off between delay optimization and communication optimization and guarantee the normal operation seque...