A novel scheduling with multi-criteria for high-performance computing systems: an improved genetic algorithm-based approach

Scheduling in high-performance computing systems is experiencing potential challenges in modern computing applications due to different application sizes, computational requirements, resource utilization, rational completion time, etc. The scheduling problem is known to be an NP-complete problem. These challenges are moderated by the logical assignment of tasks to processors in a way to produce minimum schedule length and lesser load balance by utilizing system resources. In this paper, we proposed a novel genetic algorithm (GA)-based scheduling technique by considering four conflicting objectives, minimization of makespan, load balancing, and maximization of resource utilization, and speed up ratio. A novel mutation technique is proposed which helps to improve the considered multiple objectives. The performance of the proposed work is analyzed and validated through extensive simulation results using synthetic as well as benchmark data sets. It has been observed that the proposed work performs better than the existing algorithms, GA-based scheduling, priority-based performance-improved algorithm, and particle swarm optimization. A statistical hypothesis test ANOVA followed by post hoc analysis is conducted to demonstrate the significance of the work.