基于多目标优化算法的NoC映射研究
Research On NoC Mapping Based on Multi-Objective Optimization Algorithm
功耗问题是NoC领域研究的热点,传统NoC功耗优化算法能够很好的实现映射通讯功耗单目标优化。热量平衡是NoC映射中与功耗优化相互冲突的另一重要优化目标。文章比较分析了基于多目标优化的权重算法、蚁群算法,并提出了一种基于重启多样性保存策略的多目标优化算法RMOGA,实现通讯功耗最小及热量平衡的多目标优化NoC映射。分析得出RMOGA相对于前两个算法,在NoC映射多目标优化问题上有更高的性能和效率。
Energy consumption is still a hot topic in Network on Chip (NoC). Traditional energy consumption optimization algorithm can handle mapping communication energy effectively by mapping IP cores together to minimize the energy consumption. Thermal balance is another most important optimization object, which is in conflict whit energy consumption optimization. As so, this paper proposed an improved multi-objective optimization algorithm with reboot diversity keeping strategy to map the IP cores onto the NoC using two conflicting objective optimization communication power consumption minimization and thermal balance. Analysis results show the approach is highly efficient and accurate on multi-objective NoC mapping.
袁景凌、郭宝兵
微电子学、集成电路计算技术、计算机技术
NoC映射多目标优化热量平衡通讯功耗重启策略
NoC MappingMulti-objective Optimizationhermal Balanceommunication Energy ConsumptionReboot Strategy
袁景凌,郭宝兵.基于多目标优化算法的NoC映射研究[EB/OL].(2010-03-10)[2025-08-16].http://www.paper.edu.cn/releasepaper/content/201003-272.点此复制
评论