Towards an Operating System Based Framework for Energy-Efficient Scheduling of Parallel Workloads

Power is a dominant obstacle for significant cost performance improvements of VLSI technology. Excessive and unbalanced power consumption affects device reliability, requires expensive packaging, and causes irreversible damage to semiconductor devices. Hence, power monitoring and thermal hot-spot elimination is a major concern of the semiconductor industry. This research addresses multicore power monitoring, management, and control via poweraware task scheduling and load balancing; supporting hotspot elimination as well as overall power balancing and reduction. The paper concentrates on the scheduling aspects of efficient power management schemes in the context of multicore systems. We utilize an in-house scheduling simulator and incorporate power and execution metrics in several classic scheduling algorithms. Results show a potential for scheduling with high level of power efficiency and tolerable degradation in execution time.