ABSTRACT
The goal of dynamic voltage scaling (DVS) is to reduce CPU energy consumption without degrading applications' quality-of-service (QoS). In general-purpose systems, in which a variety of tasks are scheduled by a traditional dynamic priority-based task scheduler, existing DVS schemes tend to degrade the QoS by causing urgent tasks having low priorities to starve because these schemes ignore the behavior of dynamic priority-based task schedulers. In this paper, we propose the 'double-interval-based DVS scheme for priority-based task scheduling systems (DIP)', which better manages time-sensitive tasks' QoS by bridging the gap between CPU speed scheduling and dynamic priority-based task scheduling. We describe how DIP determines an appropriate CPU speed for a group of coexisting tasks considering tasks' priorities, using an improved interval-based algorithm so that the CPU-time requirements of the low priority time-sensitive tasks in the group can be satisfied. Another novel feature of DIP is that it separates QoS-control from throughput-control via two different interval-based algorithms, which allows more energy savings when throughput is not a primary concern. Trace-based simulations show that, depending upon target systems' primary concern, DIP generally provides better energy savings at comparable QoS or better QoS at comparable energy savings, respectively, compared to existing DVS schemes.
- L. Benini, A. Bogliolo, and G. D. Micheli. A Survey of Design Techniques for System-Level Dynamic Power Management. IEEE Transactions on VLSI Systems, Vol. 8, No. 8 (June 2000), 299--316. Google ScholarDigital Library
- I. Hong, M. Potkonjak, and M. B. Srivastava. On-line scheduling of hard real-time tasks on variable voltage processor. In Proceedings of the International Conference on Computer Aided Design (November 1998), 653--656. Google ScholarDigital Library
- P. Pillai and K. G. Shin. Real-time dynamic voltage scheduling for low-power embedded operating systems. In Proceedings of the 18th ACM Symposium on Operating Systems Principles (SOSP) (October 2001), 89--102. Google ScholarDigital Library
- K. Govil, E. Chan, and H. Wasserman. Comparing Algorithms for Dynamic Speed-Setting of a Low-Power CPU. In Proc. of the First International Conference on Mobile Computing and Networking (November 1995), 13--25. Google ScholarDigital Library
- K. Flautner and T. Mudge. Vertigo: Automatic performance-setting for Linux. In Proc. of the 5th Operating Systems Design and Implementation (December 2002), 105--116. Google ScholarDigital Library
- J. R. Lorch and A. J. Smith. Operating system modifications for task-based speed and voltage scheduling. In Proceedings of the First International Conference on Mobile Systems, Applications, and Services (MobiSys) (May 2003), 215--230. Google ScholarDigital Library
- D. Grunwald, P. Levis, K. I. Farkas, C. B. Morrey III, and M. Neufeld. Policies for dynamic clock scheduling. In Proceedings of the 4th Usenix Symposium on Operating Systems Design and Implementation (October 2000), 73--86. Google ScholarDigital Library
- Transmeta cooperation. TM5800 Data Book Version 2.1. http://www.transmeta.com/developers/techdocs.html.Google Scholar
- T. Pering, T. Burd, and R. Brodersen. The Simulation and Evaluation of Dynamic voltage scheduling Algorithms. In Proceedings of International Symposium on Low Power Electronics and Design 1998 (June 1998), 76--81. Google ScholarDigital Library
Index Terms
- DIP: a double-interval-based dynamic voltage scaling scheme for dynamic priority-based task scheduling systems
Recommendations
Dynamic voltage scaling for multitasking real-time systems with uncertain execution time
GLSVLSI '06: Proceedings of the 16th ACM Great Lakes symposium on VLSIDynamic voltage scaling (DVS) for real-time systems has been extensively studied to save energy. Previous studies consider the probabilistic distributions of tasks' execution time to assist DVS in task scheduling. These studies use probability ...
Energy-efficient scheduling of real-time tasks with shared resources
This paper explores the energy-efficient scheduling of real-time tasks on a non-ideal DVS processor in the presence of resource sharing. We assume that tasks are periodic, preemptive and may access to shared resources. When dynamic-priority and fixed-...
Dynamic Voltage Scaling for Multitasking Real-Time Systems With Uncertain Execution Time
Dynamic voltage and frequency scaling can save energy for real-time systems. Frequencies are generally assumed proportional to voltages. Previous studies consider the probabilistic distributions of tasks' execution time to assist dynamic voltage scaling ...
Comments