الفهرس Only 14 pages are availabe for public view
 |  | from | 184 |  |  |
| | | from | 184 | | |
Abstract
Task scheduling is considered as a major issue for the parallel-pipelined vector processing systems to achieve best performance. The work of this paper provides a task-scheduling algorithm for the pipelined vector processors.This algorithm can schedule parallel applications represented in the form of Directed Acyclic Graphs (DAGs). Assuming initially bounded number of pipelines processing elements. The performance of the proposed algorithm has been reported and compared with the Bin-packing algorithm in terms of speedup, efficiency, and quality. According to the obtained results the proposed algorithm proved better performance than that of the Bin-packing algorithm .The complexity of the algorithm is as follows, the computation of the required DAG parameters demands that all (E) edges and (T) tasks on the DAG to be examined, which results in, at the worst case, a complexity of O(E+T).The time elapsed in generated the sorted list of level L and tasks T is O(LlogL)+O(TlogT), the while loops of an algorithm take time complexity of O(mT)+O(L) where m is the number of available pipelines at the worst case and in each iteration of these loops the search of task and suitable pipeline will be done, this is the overall time complexity is: O(E+TlogT+PT).