Worst-case timing analysis in UDSM era

Moving to deeper in the ultra-deep sub-micron (UDSM) era continuously increases process variation. Although reliable timing analysis is necessary to ensure quality design, the increase of process variation tends to degrade the validity of the worst-case corner (WC) timing analysis. In this paper, we investigate the validity of WC timing analysis, as compared to statistical static timing analysis, under the UDSM design environment. Experimental results indicated that even the 3σ values of 10% in die-to-die (D2D) and within-die (WID) variations induced the WC timing analysis to make acceptably large overestimation, degrading its validity seriously. In addition, we found that the WID variation is more important than the D2D variation in maintaining the validity of WC timing analysis.