论文部分内容阅读
随着器件尺寸的不断缩小,电路复杂度日益增加,电源已成为微处理器等需要精确供能的系统能否较快发展的主要制约因素之一。由于功耗是诸多不确定变量的函数,因此,芯片功率及宏功率分析最为准确的方法是基于制造过程和工作负荷变化的统计分析方法。与静态时序模型可以表示为一个线性规范化的高斯分布不同,统计分析是一个依赖于过程变量、工作量波动的复杂指数函数,这对统计能量分析是一个独特的挑战。提出了一种研究方法,通过研究开关电源的静态漏电功耗和动态功率统计分布,揭示宏功率分析与芯片功率分析的特征和相关性。实验结果表明,该方法对开关电源的研究是可行的。
As device size shrinks and circuit complexity increases day by day, power supply has become one of the main constraints for the rapid development of systems that require accurate power supply, such as microprocessors. Because power dissipation is a function of many uncertain variables, the most accurate method of chip power and macro power analysis is based on statistical analysis of manufacturing processes and workload changes. Unlike a static time series model, which can be expressed as a linearly normalized Gaussian distribution, statistical analysis is a complex exponential function that depends on process variables and workload fluctuations. This is a unique challenge for statistical energy analysis. A research method is proposed to reveal the characteristics and correlations of macro power analysis and chip power analysis by studying the statistical distribution of static leakage power and dynamic power of the switching power supply. Experimental results show that this method is feasible for the research of switching power supply.