We have developed a new GA hardware called GAA-I (Genetic Algorithm Accelerator-I), in which the crossover operator to be applied to each individual was dynamically selected during the algorithm execution. GAA-I has some restrictions due to the limited chip size. In this paper, we extend the GAA-I and propose a new GA hardware, GAA-II, so that large, complex optimization problems can be solved. Furthermore, GAA-II has capability of parallel processing with other GAA-II chips. The GAA-II chip has been fabricated as a CMOS standard cell chip with 0.6 m technology.