A genetic approach to standard cell placement using meta-genetic parameter optimization

摘要: The genetic algorithm applies transformations on the chromosonal representation of physical layout. works a set configurations constituting constant-size population. are performed through crossover operators that generate new configuration assimilating characteristics pair existing in current Mutation and inversion also used to increase diversity population, avoid premature convergence at local optima. Due simultaneous optimization large population configurations, there is logical concurrency search solution space which makes an extremely efficient optimizer. Three techniques compared, parameters optimized for cell-placement problem by using meta-genetic process. resulting was tested against TimberWolf 3.3 five industrial circuits consisting 100-800 cells. results indicate placement comparable quality can be obtained about same execution time as TimberWolf, but needs explore 20-50 times fewer than does TimberWolf. >