A Shared Cache-Aware Hybrid Real-Time Scheduling on Multicore Platform with Hierarchical Cache

Abstract

Hybrid approach for scheduling on multicore processors that can effectively integrate the advantage of global scheduling and partitioning scheduling, becoming one of the hottest issues in real-time scheduling. Meanwhile, different from single-core, multicore architecture in the form of a variety of hierarchical shared caches, efficient cache usage on multicore platforms can greatly improve the scheduling efficiency. This paper proposes an improved hybrid approach for scheduling real-time tasks on multicore platforms, a multicore platform is partitioned into clusters that share a L2 cache, tasks which access a common region of memory are statically assigned to a same cluster, and those task are encouraged to co-scheduling in those cluster using improved global EDF algorithm. Simulation shows that this algorithm can improve the cache performance.