高层次综合中基于整数线性规划模型的多目标功耗优化算法  

An ILP-Based Algorithm for Peak Power and Max Module Power Minimization in Behavioral Synthesis

作　　者：刘志鹏[1] 边计年[1] 赵震[2] 周强[1] 

机构地区：[1]清华大学计算机科学与技术系,北京100084 [2]北京大学数学科学学院信息科学系,北京100871

出　　处：《计算机辅助设计与图形学学报》2007年第8期966-972,共7页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(90407005);国家自然科学基金重点项目(90607001)

摘　　要：考虑峰值周期功耗和峰值模块功耗的同时优化,并尽可能地降低电路的功耗时延乘积指标.利用参数化功能单元库中各个功能模块的具体物理信息,在多供电电压、多调度周期模式下,建立了整数线性规划模型及其相应的6组约束条件,并将高层次综合中的调度过程和功能单元的绑定过程统一起来进行了模型化.文中算法已经成功地应用到自行开发的高层次综合工具之中,算法中得到的数据结果可以直接用于下一步的布图规划.对测试用例的实验进一步说明:同时优化峰值周期功耗和峰值模块功耗可以取得更好的综合结果,并且改善功耗时延乘积项的值(平均降低了30%),提高电路的可靠性和稳定性.A novel approach based on the integer linear programming （ILP）is proposed for the minimization of peak power and max module power consumption. With the support of parameterized functional units library, ILP formulations are modeled and constructed using the multiple supply voltages and multi-cycling scheme. The proposed approach has been successfully embedded in our behavioral synthesis tool and the generated corresponding results can directly serve as the input of the subsequent floorplan process. Experimental results on benchmarks indicate that simultaneously minimizing the peak power and the max module power can yield better synthesis results and better power delay product of the circuit designs.

关 键 词：整数线性规划 峰值周期功耗 峰值模块功耗 功耗时延乘积项 

分 类 号：TP368.5[自动化与计算机技术—计算机系统结构]