Scheduling and Response-time Analysis of Multicore and Multi-GPU Heterogeneous Systems

In this paper, we introduce a novel scheduling approach tailored for real-time tasks executing on heterogeneous computing platforms, which encompass both multi-core processors and multiple identical GPUs. The proposed approach addresses the limitations of traditional scheduling approaches for multi-core and multi-GPU systems, which typically rely on locks or semaphores to manage access to shared GPUs. The proposed approach employs a server task to manage the request to access the GPUs from other tasks. The requesting tasks are suspended on their corresponding cores during the GPU computation, thereby avoiding the limitations of busy waiting and long priority inversions often encountered in traditional lock-based scheduling approaches. The paper also develops the worst-case response-time analysis of tasks that are scheduled using the proposed scheduling approach on these systems. The evaluation results demonstrate the advantages of the proposed scheduling approach over the traditional lock-based approaches, especially in improving the schedulability rate of task sets in multi-core and multi-GPU systems.