Load-balancing scheduling of simulation tasks based on a static-dynamic hybrid algorithm

Xiashuang Wang,Ni Li,Guanghong Gong,Xiao Song,Yanqi Guo
DOI: https://doi.org/10.1080/17477778.2020.1772023
IF: 2.543
2022-01-01
Journal of Simulation
Abstract:A scheduling algorithm is crucial for running a simulation model so that tasks can be performed efficiently.The traditionally used blade-based parallel engine system cannot be adapted to a new simulation model. This study proposed a combined dynamic priority and static method, that is, a hybrid load-balancing scheduling (HLB) algorithm. The algorithm is given priority according to the operating cycle of the model and system steps. The experimental results demonstrated that the algorithm outperformed the earliest deadline first and the time-stepped load-balancing scheduling algorithms. The results also demonstrated that the HLBhad a higher real-time operating efficiency than the other algorithms under a lower overhead guarantee. The HLB algorithm causedbetter performance while maintaining computation and communication efficiency. Simultaneously, the utilisation rate of the central processing unit was around 35%.The further study should be enhanced to generalise it so that it could be applied to incorporate load balancing.
What problem does this paper attempt to address?