Monte-Carlo-based statistical soft error rate (SSER) analysis for the deep sub-micron era

Variation in the deep sub-micron eras has made soft error rates (SERs) more statistical and difficult to capture using static analysis. Therefore, this paper presents a Monte-Carlo based SER analysis considering the statistical impact due to variation. Quasirandom sequences are also incorporated for fast convergence of SER accuracy and time efficiency. Experiments show that the proposed framework yields more accurate SERs compared to static analysis. On top of 10⁶X speedup compared to Monte Carlo SPICE simulation, an additional 2.4X speedup can also be observed in the proposed framework after applying quasirandom sequences.