simulations. Since there are a limited number of variation measurements,
researchers have to use поп-precise parametric models of variations in their
simulations. Thus, simulations results might not be accurate enough. Our
method provides a fast technique to estimate variations and researchers can
use the real variation measurements in their simulations and improve their
evaluations. These models can also be integrated within power simulator
tools for accurate and realistic simulation models.
2. Post-Silicon optimization: Traditional VLSI design is based on the pre-
silicon optimizations. Various parameters of the design are considered by
the designer and they are tuned to meet different constraints of the design.
The variations are not considered at all; or only the statistical characteris-
tics of variations are considered.
The static timing analysis (STA) is an example of pre-silicon optimization.
The goal of STA is finding the longest delay in a specific circuit. The
variations in delay are not considered in the STA. Delays of the interconnect
wires and gates are deterministically modeled; then, using the graph model
of the circuit, the longest path in the circuit is found. However, in the
statistical static timing analysis (SSTA), the statistical characterizations of
variations are utilized to improve the longest delay estimation in presence
of the variations.
Today’s modern fabrication processes with high variability make the post-
60