WCRT Optimization Scheme Based on Task Mapping and Cache Partitioning

To reduce the worst-case response time (WCRT) of multi-core systems, an efficient approach based on the combination of task mapping and cache partitioning techniques was proposed in this paper. The allocating process was divided into two phases in the approach. In the first phase, the task with the worst-case execution time (WCET) based on the optimal cache size allocated was assigned to different cores. In the second phase, in order to meet the constraints of shared cache size, some mapped tasks were selected for callback of the cache size and readjustment of task mapping. At the same time, load balancing was considered in the two phases. Experiment results show that the proposed approach can reduce the WCRT of the systems and improve execution efficiency compared with previous techniques. The proposed approach can reduce WCRT of systems by up to 6.7% on average compared with GCP approach. The approach outperforms the LIP-based approach on execution efficiency.