欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
芯片上局部的高温对集成电路有很多负面影响．目前已有的行为级综合算法只是在资源绑定阶段降低峰值温度，忽视了资源数量分配阶段对峰值温度的影响．为此，提出一种资源数量分配算法，根据前一个综合结果的功耗密度以及经热分析得到的温度反馈调整各种资源的使用量，使不同种类资源间的功耗密度接近平均分布，从而达到降低峰值温度的效果．实验结果表明，该算法在增加2．0％面积的情况下能平均降低峰值温度9．4℃．
英文摘要：
Local on-chip hotspots have several negative impacts on integrated circuits. Previous thermal-aware behavioral syntheses reduce peak temperature only in binding phase, ignoring allocation phase. This paper proposes a thermal-aware resource usage allocation algorithm. According to the feedbacks which are power density and temperature of the previous solution, this algorithm improves the resources usage to create almost even power density distribution of resources of different types, so that peak temperature can be decreased. Our experimental results show that this algorithm reduces peak temperature by 9.4 ℃ on average with only 2.0 % area overhead.
同期刊论文项目
同项目期刊论文
期刊信息
位置：