Energy-efficient elastic scheduling in heterogeneous computing systems

Reducing energy consumption has become a major goal in designing modern heterogeneous computing systems. Conventional energy-efficient scheduling strategies developed on this kind of systems mainly focus on energy saving, lacking of the consideration of users´ expectations such as expected finish time. Again, the system elasticity is not sufficiently taken into account while making scheduling decisions. In this paper, we develop a novel scheduling strategy named energy-efficient elastic (3E) scheduling for aperiodic, and independent tasks on DVS-enabled heterogeneous computing systems. The 3E strategy concentrates on adjusting tasks´ supply voltages according to the system workload, thereby making the best trade-offs between energy conservation and users´ expected finish times. Experimental results demonstrate that our 3E significantly improves the scheduling quality of the others, and is able to effectively enhance the system elasticity.