Multi-subpopulation parallel computing genetic algorithm for the semiconductor packaging scheduling problem with auxiliary resource constraints

When the scheduling is established in a semiconductor packaging factory, frequent machine-related changes can pose a serious problem because of the large number of products processed using different machines with auxiliary resources. Thus, the efficiency of the scheduling algorithm is crucial for addressing the semiconductor packaging scheduling problem (SPSP). This study proposed a novel multi-subpopulation parallel computing genetic algorithm (MSPCGA) to solve the SPSP under practical production constraints. The MSPCGA uses a multithreaded central processing unit to perform parallel computing. The graphics processing unit (GPU) grid computing method was applied to modify the genetic algorithm computing architecture to increase the efficiency of the algorithm. Finally, the proposed MSPCGA outperformed two other metaheuristic algorithms in 12 evaluation scenarios. Additionally, the existing factory method was compared with the proposed MSPCGA to verify the effectiveness of the algorithm in practical applications.