CPU vs GPU performance when solving the flexible job-shop problem using genetic algorithms

Sammanfattning: In this thesis, a comparison is made between the performance of genetic algorithms executed on the CPU and the GPU for solving the Flexible Job-shop Scheduling Problem (FJSP). The purpose is to determine whether one processing unit offers better performance than the other. The evaluation is based on execution time and makespan achieved on both processing units, as well as profiler data. The experimental setup includes an AMD Ryzen 5 3600 CPU, Nvidia GeForce GTX 1650 Max-Q and Nvidia GeForce GTX 1070 GPUs, as well as a benchmark dataset for FJSP with various problem sizes. The results obtained from the experiments shows that the GPU outperforms the CPU by a significant margin in terms of execution time, while the makespan remain very similar on all processing units.