University of Veszprem
Exploiting thread-level parallelism is believed to be a reliable way to achieve higher performance improvements in the future. In this paper, the authors introduce a History-Aware, Resource-based Dynamic (HARD) scheduler for heterogeneous CMPs. HARD relies on recording application resource utilization and throughput to adaptively change cores for applications during runtime. They show that HARD can be configured to achieve both performance and power improvements. They compare HARD to a complexity-based static scheduler and show that HARD outperforms this alternative.