摘要
为了将应用任务快速有效地映射到片上网络,并实现低能耗、低时延等目标,优化多目标模型并提出一种改进的量子遗传算法解决片上网络映射问题。采用加权和法考虑网络拥堵时延,并引入通信带宽量化适应值,便于类比分析映射效果;结合应用任务和片上网络结构特点,利用任务节点相关链路数和通信权重双重优先标准构建较优初始解集,使得量子遗传算法改进后映射寻优收敛更加快速高效。实验结果表明,在同样的多目标映射模型下,改进的量子遗传算法映射寻优更快更精准。
In order to map application tasks to the network on chip quickly and efficiently,and achieve the goals of low energy consumption and low delay,this paper optimizes the multi-objective model and proposes a modified quantum genetic algorithm to solve the problem of network on chip mapping.The weighted sum method was used to consider the network congestion delay,and the communication bandwidth quantitative adaptation value was introduced to facilitate the analogy analysis of the mapping effect.It combined the application tasks and the characteristics of the network on chip structure,and constructed a better initial solution set by using the task node related link number and the communication weight double priority criterion,so as to make the modified quantum genetic algorithm faster and more efficient in mapping optimization convergence.The experimental results show that under the same multi-objective mapping model,the modified quantum genetic algorithm is faster and more accurate in mapping optimization.
作者
张保岗
韩国栋
汤先拓
Zhang Baogang;Han Guodong;Tang Xiantuo(Information Engineering University,Zhengzhou 450002,Henan,China)
出处
《计算机应用与软件》
北大核心
2020年第8期115-121,共7页
Computer Applications and Software
基金
国家科技重大专项核高基项目(2016ZX01012101)。
关键词
片上网络
量子遗传算法
映射技术
Network on chip
Quantum genetic algorithm
Mapping technique
