GTFTTS: A Generalized Tit-for-Tat Based Corporative Game for Temperature-Aware Task Scheduling in Multi-core Systems

摘要

Temperature-constraint computing environments are emerging those years, especially in embedding computing. A game theoretic temperature-aware scheduling algorithm for multi-core systems is proposed in this paper, namely, GTFTTS (Generalized Tit-For-Tat Temperature-aware Scheduling). GTFTTS is designed to work in a resource-rich environment where resources always compete for tasks. A generalized Tit-for-Tat based method, where whether a core will corporate or not is decided by a hardness factor, is considered in this paper. The algorithm is implemented in our TASS simulator. Simulations results show that the proposed game can reduce the temperature difference between different group of cores which effectively avoids the local hotspot of a processor.