A wirability expert system (WES) is a knowledge based program implemented in the knowledge engineering environment (KEE) ~ on a Symbolics 3670 LISP machine. The design of ICES is based on interviews with VLSI wirability experts. The knowledge domain and the decision synthesis process of a human expert are analysed and extracted to form the reasoning basis of WES. An optimizing correlation between the chip geometry and the cell geometry is presented. Finally, possible extensions to WES are discussed briefly. computer-aided design, expert system, knowledge engineering environment, VLSI wiring Chip design with VLSI technology indicates that 50-80% of available chip space is allocated to routing. Hence, the prediction of wiring space requirements is a crucial step in the VLSI chip design process. During the early stages of chip design, a wirability expert is often consulted on the appropriate number of functions that can be placed on a chip, so that too many circuits are not committed; chip spaces are not under-used; the capability of available cell placement and routing programs are not taxed beyond their limitations; the restrictions of the fabrication facilities are well observed; and the proper cell libraries are developed.
To estimate wiring space requirements, Hailer et al 1 proposed a 1 D stochastic model, and this was extended to 2D by Patel et al 2 . The models were tested and calibrated using chip design data. This serves as a basis for wirability experts to estimate wiring space requirements. A wirability expert system is treated as a black box. The basic inputs to the box are the chip's dimensions and the device population. Inputs to the box consist of information relating to CAD technology, whether it is gate array, semicustom or fullcustom, and relating to semiconductor technology such as ECL, TTL, bipolar, etc.
WES, a decision inferencing system, draws from two domains of resources: the knowledge domain of mostly factual information and the inferencing domain of rules that depend upon the contents of the knowledge domain. An inferencing engine, coupled with a deduction mechanism, generates consequences from premises. WES is fully introduced in the next section, which is followed by two sections on the knowledge domain and the inferencing process of WES.
WlRABILITY EXPERT SYSTEM (WES)
WES is implemented in the knowledge engineering environment (KEE), which supports a frame-based knowledge structure, a semantic network, and the object orientated programming methodology 3-s. The inbuilt inference engine supports forward and backward reasoning processes.