Optimization of Importance Sampling Monte Carlo using Consecutive Mean-shift Method and its Application to SRAM Dynamic Stability Analysis

Takeshi Kida,  Yasumasa Tsukamoto,  Yuji Kihara
Renesas Electronics Corporation


Abstract

With the scaling of MOSFET dimensions and the lowering of supply voltage, more precise estimation of minimum operating voltage (Vmin) of SRAM at 6-sigma level is needed. In this paper, we propose a method based on the importance sampling (IS) Monte Carlo simulation to elaborately predict Vmin for future technology node below 22 nm generation. By executing Monte Carlo (MC) simulation with consecutive mean-shift method we propose, it is shown that the most probable failure point (MPFP) is effectively and steadily derived within 0.01σ accuracy without depending on the way of random number generation. The proposed method is applied to static and dynamic behaviors of SRAM. Through the comparison with the results obtained by conventional IS method, we clarify that the method proposed is more suitable for future SRAM characterization.