Utilization-Controlled Task Consolidation for Power Optimization in Multi-core Real-Time Systems

Since multi-core processors have become a primary trend in processor development, new scheduling algorithms are needed to minimize power consumption while achieving the desired timeliness guarantees for multi-core (and many-core) real-time embedded systems. Although various power/energy-efficient scheduling algorithms have recently been proposed, existing studies may have degraded run-time performance in terms of power/energy efficiency and real-time guarantees when applied to real-time embedded systems with uncertain execution times. In this paper, we propose a novel online solution that integrates core-level feedback control with processor-level optimization to minimize both the dynamic and leakage power consumption of a multi-core real-time embedded system. Our solution monitors the utilization of each CPU core in the system and dynamically responds to unpredictable execution time variations by conducting per-core voltage and frequency scaling. We then perform task consolidation on a longer timescale and shut down unused cores for maximized power savings. Both empirical results on a hardware multi-core test bed with well-known benchmarks and simulation results in many-core systems show that our solution provides the desired real-time guarantees while achieving more power savings than three state-of-the-art algorithms.