嵌入式多核心即時系統之排程與動態電壓調整機制

隨著科技的進步，嵌入式系統已被廣泛的使用在可攜式裝置中，可攜式嵌入式系統必須有相當好的運算能力來達成應用的即時需求。由於運算量越來越大，相對的能量消耗也隨之增加。由於系統上電池容量有限，可攜式嵌入式系統能源消耗變得非常重要，如果能有效降低系統的能源消耗，可攜式裝置將可以延長工作時間，因此如何在符合應用的即時需求的條件下消耗最小的能源，對於可攜式嵌入式系統設計上非常重要。動態電壓調整技術能有效降低能源消耗，主要是利用鬆弛時間來調整處理器的執行電壓、頻率，以減少能源消耗。多核心系統已漸漸成為主流，在相同的工作時脈頻率下，多核心比單核心有更好的產能，在多核心環境下，即時性工作排程是一個NP-hard的問題。本論文中，我們提出了一個適用於多核心的即時排程機制，並結合動態電壓調整來降低能源消耗，所提出的ME-DVS（Modified EDF with Dynamic Voltage Scaling）排程演算法能有效降低多核心環境中能源消耗，並且保證所有工作都能在執行期限內完成。實驗結果顯示在4核心環境中，當工作數增加為30個工作，處理器總利用度到達87%時，能減少大約3％的能源消耗，工作數相對減少至5個工作，處理器總利用度為18%時，能減少約52％的能源消耗，在8核心環境與16核心環境也有類似的結果。

關鍵字

嵌入式系統；動態電壓調整；多核心系統；多核心即時排程

並列摘要

With the advancement of technology, embedded systems have been widely used in portable devices. Portable embedded systems must have rather superior computing capability in order to meet real-time application demands. As the computation increases, so does the corresponding energy consumption. The energy consumption of portable embedded system is a very important issue due to limited battery capacity of the system. The working time of portable device can be prolonged if energy consumption of the system can be effectively reduced. Therefore it is very important for the design of portable embedded system to enable minimum energy consumption while meeting real-time application demands. Dynamic voltage scaling technology enables effective reduction of energy consumption by utilizing slack time to modify operation voltage and frequency of processor in order to reduce energy consumption. Multi-core systems have gradually become mainstream products providing better throughput capacity than single-core systems under the same working clock frequency. However, in multi-core environment the real-time scheduling is an NP-hard issue. In this thesis, we propose a real-time scheduling mechanism for multi-core systems which in conjunction with dynamic voltage scaling can reduce energy consumption. The proposed ME-DVS (Modified EDF with Dynamic Voltage Scaling) scheduling algorithm can effectively reduce energy consumption in multi-core environment and ensure all tasks to meet their deadlines. The experimental results have shown that in 4-core environment, the energy consumption can be reduced by 3% when increasing the number of tasks to 30, 87% of processor total utilization, and by 52% when increasing the number of tasks to 5, 18% of processor total utilization. Similar results can be found in 8-core and 16-core environments.

並列關鍵字

Embedded System ； DVS ； Multi-core System ； Multi-core Real-time Scheduling

參考文獻

[1] B. Andersson, S. Baruah and J. Jonsson, ” Static-priority scheduling on multiprocessors,” IEEE Real-Time System Symposium, pp.193-202, Dec. 2001.

[2] D. Bautista, J. Sahuquillo, H. Hassan, S. Petit and J. Duato, “A simple power-aware scheduling for multicore systems when running real-time applications,” IEEE International Symposium on Parallel and Distributed Processing, PP.1-7, Apr. 2008.

[3] T. D. Burd, T. A. Pering, A. J. Stratakos, and R. W. Broderson, “A Dynamic Voltage Scaled Microprocessor System,” IEEE Journal of Solid-State Circuits, Vol. 35, pp.1571-1580, Nov. 2000.

[4] J. Donald and M. Martonosi, “Techniques for Multicore Thermal Management: Classification and New Exploration,” IEEE International Symposium on Computer Architecture, pp.78-88, Dec. 2006.

[5] Dertouzos, M.L., Mok, A.K., ”Multiprocessor On-Line Scheduling of Hard-Real-Time Tasks”, IEEE Transactions on Software Engineering, v.15, n.12, pp. 1497-1506, 1989

國際替代計量

嵌入式多核心即時系統之排程與動態電壓調整機制

未授權

主題瀏覽