CAJ | 학술논문

分析了高速crossbar调度算法iSLIP在处理突发业务时性能严重恶化的原因.结合LQF/iLQF算法的思想,提出了又一种输入排队crossbar调度算法iPGQM.仿真结果表明:该调度算法在均匀业务流量下和iSLIP算法的性能基本相同;在突发业务的条件下,iPGQM算法具有更好的抗突发特性;特别在重负载的条件下,与iSLIP算法相比,不仅具有更高的吞吐量,而且平均延迟降低了10%左右.
분석료고속crossbar조도산법iSLIP재처리돌발업무시성능엄중악화적원인.결합LQF/iLQF산법적사상,제출료우일충수입배대crossbar조도산법iPGQM.방진결과표명:해조도산법재균균업무류량하화iSLIP산법적성능기본상동;재돌발업무적조건하,iPGQM산법구유경호적항돌발특성;특별재중부재적조건하,여iSLIP산법상비,불부구유경고적탄토량,이차평균연지강저료10%좌우.
This paper analyzed the reasons for which the high-speed crossbar scheduling algorithm iSLIP has a serious deterioration of performance under burst traffics. With reference to the ideas of LQF/iLQF,this paper proposed a novel input-queued crossbar scheduling algorithm called iPGQM (iterative Parallel Graded-Length Queue Matching). Simulation results show that iPGQM has the same performance as iSLIP under uniform traffics. Furthermore it has better performance than iSLIP under burst traffics. Especially in heavy load conditions,iPGQM not only achieves higher throughput,but also reduces the average delay about 10% compared with iSLIP algorithm.