Minimizing makespan under data prefetching constraints for embedded vision systems: a study of optimization methods and their performance

In confronting the “Memory Wall”, design of embedded vision systems exhibits many challenges regarding cost, energy consumption, and performance. This paper considers a variant Job Shop Scheduling Problem with tooling constraints, arising in this context, which completion time (makespan) is to be minimized. objective corresponds performance produced circuit. We discuss different formulations using integer linear programming point out their characteristics, namely size quality relaxation bound. To solve scheduling problem large size, we compare various approaches, including Constraint Programming model, two constructive greedy heuristics, models LocalSolver, Simulated Annealing algorithm, Beam Search algorithm. Numerical experiments are conducted on 16 benchmark instances from literature 12 real-life non-linear image processing kernels for validating efficiency.