TECS: Temperature- and Energy-Constrained Scheduling for Multicore Systems

The widespread use of multicore architectures with decreasing feature size is causing severe increase of on-chip power dissipation in modern embedded systems. This introduces both thermal and energy management problems that need to be addressed during the system level design. In this paper, we explore the DVS scheduling problem on multicore systems under both temperature and energy constraints. We present an exact algorithm as well as a polynomial time approximation scheme, since this problem is NP-hard. When the original problem is schedulable, our approximation algorithm is guaranteed to generate a solution, which will not violate the temperature constraint, and consume no more time or energy than a specified approximation bound, e.g., within 1%, of the optimal time consumption and energy constraints. We evaluate our approach using both real and synthetic benchmarks mapped on DVS capable multicore processors. The experimental results demonstrate that our technique is able to produce schedules close to the optimal solution with reasonable execution time.