Implementation study of path-based AOCV model on pessimism reduction for 20nm technology

The Advanced On-Chip Variation (AOCV) model is used to systematically correct Liberty timing file for on-chip variation (OCV) based on the logic depth and distance of a path. The AOCV analysis solutions can be categorized into graph-based (GBA) and path-based analysis (PBA), which are refinement strategies over traditional OCV in static timing analysis. PBA AOCV performs recalculation on critical paths with advanced algorithm to reduce pessimism. This paper studies the implementation and parameters of AOCV in 20nm process node, specifically with the path-based approach, and compared against the conventional global On-Chip Variation for pessimism reduction. The result indicates that PBA AOCV has advantage over PBA OCV with 9% and 18% reduction in total variation paths and total negative slacks, respectively. By using recalculation algorithm, the path delay is annotated with adequate value of de-rate. Thus, pessimism can be further reduced at a cost of run time.