An optimal semi-partitioned algorithm for scheduling real-time applications on uniform multicore processors

Uniform multicore processor is an acknowledged architecture for building complex real-time embedded systems. Having multiple cores with various execution capabilities makes it a favorable architecture to structure highperforming energy efficient systems. This paper addresses the real-time scheduling problem of uniform multicore processors and proposed a novel fixed-priority semi-partitioned scheduling algorithm: RM-SPwTS. We have shown that RM-SPwTS dominates the current semi-partitioned counterparts. It is additionally indicated that RMSPwTS is an optimal algorithm as it achieves 69 % utilization bound. To the best of our knowledge, RM-SPwTS is the first algorithm in fixed-priority multicore scheduling category that achieves this bound. The superior performance of RM-SPwTS over its existing counterparts is also established through extensive simulations. The simulation results reveal that RM-SPwTS schedules up-to 95 % more task-sets with heavy system utilization (72 %-73 %) as compared to its counterparts. Further, it improves the processor utilization (up-to 14 %) and reduces the number of cores (up-to 24 %) required to feasibly schedule the given workload.