Consideration of Control System and Memory Contributions in Practical Resource-Constrained Scheduling for Low Power

摘要

This paper presents a fast and efficient heuristic resource-constrained scheduling algorithm that works to reduce power consumption by maximally utilizing the resource operating at multiple voltages while at the same time maintaining a low latency. Unlike other resource-constrained scheduling algorithms, our algorithm also takes into account the contribution from the control system or circuitry as well as the memory required. By doing so, we are able to gain more insight into the items affecting power, area, and timing as well as the tradeoffs. Scheduling with the control system and memory in mind, we see that there is a direct relationship between the number of resources, the number of control cycles, and the memory needed, which affect the overall power, area, and timing. Experimental results show that taking these extra factors into account we are able to achieve an average power savings that is better than the power savings from other resource-constrained scheduling algorithms.