| OCR Text |
Show 8 - Three traditional approaches to the problem of parasitic extraction are in contrast to ACRE. The first approach is to analyze a small interaction with a finite element method, where two metal layers cross or when one transistor is analyzed for gate resistance or capacitance. ACRE differs from this because it has no native features for performing finite element analysis, but this could be installed into the language through the extension capability. ACRE supports a lookup of the interactions and can capitalize on such work when it is provided as tabular data. The second traditional approach occurs when a preconfigured sequencing engine applies a set of rules to layout, and only the rules deck is under user control. ACRE differs from the second by making it trivial to program a custom sequencing engine in addition to providing a custom set of rules for the engine. The third traditional approach characterizes individual cells and sums the results to determine cumulative parasitics. ACRE differs from the third approach by flattening the layout and providing a node-to-node evaluation. ACRE may be applied on a cell-by-cell basis to achieve a hierarchical analysis, but this would not take into account capacitances between cells. Where ACRE provides a difference from other approaches is in the merging of the language and the analysis techniques. With object-oriented databases and objectoriented languages, it is observed that there is a considerable synergy when the two merge into one system of syntax and semantics, and ACRE applies this synergy to the problem of circuit analysis. Program loops, for example, which are easily envisioned in a language, may have conditionals drawn from data measurements, data base traversals, or intermediate extraction results from the |
