Iterative Zone Refinement 알고리즘을 이용한 파이프라인 멀티코어 시스템에서의 효과적인 태스크 스케쥴링 기법

Abstract

A new task mapping and scheduling method is developed, which resembles the zone refining process to purify crystal ingots. In our approach, the tasks are grouped and each group is iteratively optimized one by one. One of good features of our method is that the tradeoff between performance and runtime is possible by adjusting the group size. The proposed task mapping and scheduling method is applied for pipelined processing, as well as batch processing.