论文部分内容阅读
提出了一种基于工艺参数扰动的随机点匹配时延评估算法.该算法通过Cholesky分解将具有强相关性的工艺随机扰动转化为独立随机变量,并结合随机点匹配方法和多项式混沌理论对耦合随机互连线模型进行时延分析.最后,利用数值计算方法给出互连时延的有限维表达式.仿真实验结果表明,该算法与HSPICE仿真时延的相对误差不超过2%,且相比于HSPICE显著降低了电路模拟时间.
A method of random point matching delay estimation based on process parameters perturbation is proposed.The algorithm converts the stochastic perturbations of process with strong correlation into independent random variables by Cholesky decomposition and combines the stochastic point matching method with the polynomial chaos theory The interconnection model is used to analyze the delay.Finally, finite dimensional expressions of interconnect delay are given by numerical calculation.The simulation results show that the relative error between the proposed algorithm and HSPICE simulation delay is less than 2% HSPICE significantly reduces the circuit simulation time.