Dynamic processor assignment in a task system with time-varying load

In many applications a task is repeatedly executed on several sets of data in a pipeline fashion. For example, image processing software is frequently executed on a sequence of images. Due to varying semantic content in the data, each subtask of the overall task may experience variations in execution time for different instances of the data. The authors consider the problem of efficiently executing a task on a large parallel machine. In particular, they focus on dynamically assigning processors to subtasks in response to changing workloads seen by the subtasks. They present several processor assignment algorithms and study their performance through simulation. The simulation study is based on an application in computer vision. The results suggest dynamic re-assignment can perform very close to the theoretical optimum and distinctly better than static assignments.