Algorithms in computer-aided design of VLSI circuits.

With the increased complexity of Very Large Scale Integrated (VLSI) circuits,Computer Aided Design (CAD) plays an even more important role. Top-downdesign methodology and layout of VLSI are reviewed. Moreover, previouslypublished algorithms in CAD of VLSI design are outlined.In certain applications, Reed-Muller (RM) forms when implemented withAND/XOR or OR/XNOR logic have shown some attractive advantages overthe standard Boolean logic based on AND/OR logic. The RM forms implementedwith OR/XNOR logic, known as Dual Forms of Reed-Muller (DFRM),is the Dual form of traditional RM implemented with AND /XOR.Map folding and transformation techniques are presented for the conversionbetween standard Boolean and DFRM expansions of any polarity. Bidirectionalmulti-segment computer based conversion algorithms are also proposedfor large functions based on the concept of Boolean polarity for canonicalproduct-of-sums Boolean functions. Furthermore, another two tabular basedconversion algorithms, serial and parallel tabular techniques, are presented forthe conversion of large functions between standard Boolean and DFRM expansionsof any polarity. The algorithms were tested for examples of up to 25variables using the MCNC and IWLS'93 benchmarks.Any n-variable Boolean function can be expressed by a Fixed PolarityReed-Muller (FPRM) form. In order to have a compact Multi-level MPRM(MMPRM) expansion, a method called on-set table method is developed.The method derives MMPRM expansions directly from FPRM expansions.If searching all polarities of FPRM expansions, the MMPRM expansions withthe least number of literals can be obtained. As a result, it is possible to findthe best polarity expansion among 2n FPRM expansions instead of searching2n2n-1 MPRM expansions within reasonable time for large functions. Furthermore,it uses on-set coefficients only and hence reduces the usage of memorydramatically.Currently, XOR and XNOR gates can be implemented into Look-Up Tables(LUT) of Field Programmable Gate Arrays (FPGAs). However, FPGAplacement is categorised to be NP-complete. Efficient placement algorithmsare very important to CAD design tools. Two algorithms based on GeneticAlgorithm (GA) and GA with Simulated Annealing (SA) are presented for theplacement of symmetrical FPGA. Both of algorithms could achieve comparableresults to those obtained by Versatile Placement and Routing (VPR) toolsin terms of the number of routing channel tracks