RESET: A real-time scheduler for energy and temperature aware heterogeneous multi-core systems

摘要

Nowadays, multi-core processing systems have to perform complex functionalities on densely packed multi-million gate platforms, making design issues like resource usage efficiency, energy consumption, temperature management of cores, etc. more challenging to handle. Since many of these processing platforms use batteries as their primary source of energy and are prone to an uncontrolled surge in temperatures, researchers have started focusing on these criteria in great detail. In this work, we propose a three-phase hierarchical resource allocation strategy called RESET for scheduling periodic tasks with a bounded number of migrations and context-switches. The first phase divides time into distinct intervals/windows based on deadlines of tasks. The exact proportional fairness needed for the progress of task executions is maintained at all window boundaries. The second phase performs energy-aware task-to-core assignments based on task execution requirements and operating frequency levels of cores. In the last phase, it splits tasks based on an input parameter and prepares a temperature-aware schedule for individual cores. Our experimental analysis shows that the presented strategy improves upon the state-of-the-art in terms of resource utilisation and reduces dynamic energy consumption and the average temperature of cores in the system.