本論文探討虛擬技術執行環境中，多核心系統之電源消耗與節能議題，並提出一高效之虛擬機器環境動態電源管理機制，以此機制調整核心處理器之電壓與頻率，進而降低系統執行能耗。在過去的研究中，以靜態系統利用率或是行為分群來當成頻率調變條件，其因靜態門檻值範圍限制或群組之間頻率差異太大，導致僅能夠減少有限的能源消耗。本論文針對虛擬技術執行環境異質多核心系統，實際監控每一執行緒狀態，以系統真實需求之利用率為參數進行頻率調節，有效減少系統耗能。

本研究提出之虛擬技術環境動態DVFS電源管理機制主要包含三大部份，首先，利用核心處理器內嵌之硬體行為及效能監控模組對每一執行緒做分析與記錄，並預測下一執行緒行為，判斷適合系統執行之電壓與頻率; 其次，以各核心之間工作量平衡為考量基準，判斷各核心之間高計算量執行緒數目，為系統做執行緒遷移動作，平衡各核心之執行負載量; 第三，分析系統實際需求，以執行緒為單位，動態調整核心處理器至最合適之執行電壓與執行頻率。

在本論文研究中，我們藉由分析KVM虛擬技術執行之系統環境下，執行緒實際執行狀況，並提出一利用率計算模型獲得系統真實需求，以此定義出各核心處理器最適用之執行電壓與頻率。最後，我們將此一虛擬技術環境電源管理機制實現於Linux 2.6 作業系統中，並以具公信標準之測試程式量測比較。根據實驗結果數據顯示，本文所提出機制在僅延長平均3%的系統執行時間之狀況下，有效的降低平均近65%的系統能耗，此則證明了本機制不只顧慮到系統執行時間，更達成節省能耗的主要目標。

This thesis proposes a highly effective dynamic power management mechanism for a virtual machine environment, which regulates the voltage and frequency of core processors to reduce energy consumption during system operations.

The dynamic DVFS power management mechanism for a virtual technology environment proposed in this thesis primarily consists of three parts. First, each thread was analyzed and recorded using the hardware behaviors and performance monitoring module embedded in the core processor. The behaviors of the subsequent threads were predicted to determine the voltage and frequency that suit the system implementation. Second, based on the workload balance among various cores, the number of high computation threads was determined to maintain an execution load balance among the cores by performing thread migration for the system. Third, the actual demands of the system were analyzed; and, using the unit of threads, the core processors were dynamically regulated to the most appropriate execution voltage and frequency.

In this study, we analyzed the actual execution state of the threads under the system environment executed by KVM virtualization technology. Finally, we implemented this power management mechanism in the Linux 2.6 operating system. The experimental results indicated that the proposed mechanism is capable of reducing system energy consumption by an average of nearly 65%, while prolonging the execution time by an average of only 3%. The results suggest that, aside from considering the execution time, this mechanism can achieve the main objective of energy saving.

[1] R. Ayoub, U. Ogras, E. Gorbatov, Y. Jin, T. Kam, P. Diefenbaugh and T. Rosing, “OS-level power minimization under tight performance constraints in general purpose system”, IEEE International Symposium on Low Power Electronics and Design, 1-3 (2011), pp. 321-326.

[2] “Cloud Computing”, available at http ://en.wikipedia.org/wiki/Cloud_computing.

[3] H. Cho, B. Ravindran and C. Na, “Garbage collector scheduling in dynamic, multiprocessor real-time systems”, IEEE Trans. on Parallel and Distributed System, 20, 6 (2009), pp. 101-106.

[4] W.M. Chen, H.Y. Hung, J.M. Liang, “Energy-efficient scheduling of periodic real-time tasks for reliable multi-core system”, IEEE Conference on Electrical and Control Engineering, 16-18 (2011), pp. 5887-5890.

[5] K. Choi, R. Soma and M. Pedram, “Fine-grained dynamic voltage and frequency scaling for precise energy and performance tradeoff base on the ratio of off-chip access to on-chip computation times”, Transactions on IEEE Computer-Aided Design of Integrated Circuits and Systems, 24-1 ( 2004), pp 18-28.

[6] P. Domingues, F. Araujo, L. Silva, “Evaluating the performance and intrusiveness of virtual machines for desktop grid computing”, IEEE International Symposium on Parallel and Distributed Processing, 23-29 (2009), pp. 1-8.

[7] E. Duesterwald, C. Cascaval, and S. Dwarkadas, “Characterizing and predicting program behavior and its variability”, Proceedings 12th International Conference on Parallel Architectures and Compilation Techniques, (2003), pp 220-231.

[8] S. Ghiasi, T. Keller and F. Rawson, Scheduling for heterogeneous processors in server systems, Proceedings of the 2nd conference on Computing frontiers, (2005), pp 199-210.

[9] R. Ge, X. Feng, S. Song, H-C. Chang, D. Li, and K W.Cameron, “PowerPack : energy profiling and analysis of high-performance system and applications”, IEEE Trans. on Parallel and Distributed System, 21, 5 (2010), pp. 658-671.

[10] V.L. G, L. Wang, X. He and Y. A.J, “Power-aware scheduling of virtual machines in DVFS-enabled clusters”, IEEE Conf. on Cluster Computing and Workshops, 4 (2009), pp. 1-10.

[11] J.L. Henning, “SPEC CPU2006 benchmark descriptions”, ACM SIGARCH Computer Architecture, 34-4 (2006), pp. 1-17.

[12] M.K. Huang, J.M. Chang and W.M. Chen, “Grouping-based dynamic power management for multi-threaded programs in chip-multiprocessors”, The proceeding 7th IEEE/IFIP International Conference on Embedded and Ubiquitous Computing, (2008), pp. 56-63.

[13] “Intel VTune Amplifier XE”, available at http ://software.intel.com/en-us/articles/intel-vtune-amplifier-xe.

[14] “Intel corporation, Intel 64 and IA-32 architectures software developers manual volume 3A: system programming guide, part 1”, available at http://www.intel.com/products/processor/manuals/, (2008).

[15] “Intel corporation, enhanced Intel speedstep technology for the Intel pentium M processor”, available at http://www.intel.com/design/intarch/papers/301174.htm, (2004).

[16] C. Isci, G. Contreras and M. Martonosi, “Live, runtime phase monitoring and prediction on real systems with application to dynamic power management”, The proceeding 39th Annual IEEE/ACM International Symposium on Microarchitecture, (2006), pp 359-370.

[17] C. Isci, M. Martonosi, and A. Buyuktosunoglu, “Long-term workload phases: duration predictions and applications to DVFS”, IEEE Micro: Special Issue on Energy Efficient Design 25(5), (2005), pp 39-51.

[18] C. Isci, A. Buyuktosunogly, C. Y. Cher, P. Bose and M. Martonosi, “An analysis of efficient multi-core global power management policies : maximizing performance for a given power budget”, The proceeding 39th Annual IEEE/ACM International Symposium on Microarchitecture, (2006), pp 347-358.

[19] “KVM kernel-based virtual machine”, available at http ://www.linux-kvm.org/page/Main_Page.

[20] T. Kuo, J. Wu and H. Hsih, “Real-time concurrency control in a multiprocessor environment”, IEEE Trans. on Parallel and Distributed System, 13, 6 (2002), pp. 689-671.

[21] R. Kotla, S. Ghiasi, T. Keller and F. Rawson, “Scheduling processor voltage and frequency in server and cluster systems”, Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium, (2005), pp 234-241.

[22] R. Kotla, A. Devgan, S. Ghiasi, “Characterizing the impact of different memory-intensity levels”, International Workshop on Workload Characterization, (2004), pp 3-10.

[23] W. Kim, M. Gupta, Gu-Y. Wei and D. Brooks, “System level analysis of fast, per-core DVFS using on-chip switching regulators”, IEEE International Symposium on High Performance Computer Architecture, 16-20 (2008), pp. 123-134.

[24] H. Lin, X. Ma, W.Feng and N. Samatova, “Coordinating computation and I/O in massively parallel sequence search”, IEEE Trans. on Parallel and Distributed System, 22, 4 (2011), pp. 529-543.

[25] R. Love, “ Linux kernel development 2nd ed”, Indianapolis, Ind.:Novell Press, (2005).

[26] A. Ros, M. Acacio and J. Garcia, “A direct coherencr protocol for many-core chip multiprocessors”, IEEE Trans. on Parallel and Distributed System, 21, 12 (2010), pp. 1779-1792.

[27] E. Seo, J. Jeong, S. Park, and J. Lee, “Energy efficient scheduling of real-time tasks on multicore processors”, IEEE Trans. on Parallel and Distributed System, 19, 11 (2008), pp. 1540-1552.

[28] P.M. Shelkovnykov, S.M. Botulinskiy, “Improving Microsoft Hyper-V live migration efficiency over distance”, IEEE International Crimean Conference on Microwave and Telecommunication Technology, 12-16 (2011), pp. 473-474.

[29] T. Sherwood, S. Sair, and B. Calder, “Phase tracking and prediction”, In Proceedings of the 28th International Symposium on Computer Architecture, (2003), pp 336-349.

[30] X. Shen, Y. Zhong, and C. Ding, “Locality phase prediction”, The Eleventh International Conference on Architectural Support for Programming Languages and Operating Systems, (2004), pp 165-176.

[31] “The ondemand govemor”, available at http ://publib.boulder.ibm.com.

[32] “The performance govemor”, available at http ://publib.boulder.ibm.com.

[33] “Unixbench”, available at http ://blog.jpps.tc.edu.tw/lifetype/post/2/1308.

[34] D. Warneke and O. Kao, “Exploiting dynamic resource allocation for efficient parallel data processing in the cloud”, IEEE Trans. on Parallel and Distributed System, 22, 6 (2011), pp. 985-997.

[35] R. Weber, A. Gothandaraman, R. Hinde and G. Peterson, “Comparing hardware accelerators in scientific applications : a case study”, IEEE Trans. on Parallel and Distributed System, 22, 1 (2011), pp. 58-68.

[36] X. Wu, Y.Lin, J.Han, and J. Gaudiot, “Energy-efficient scheduling of real-time periodic tasks in multicore system”, IFIP International Federation for Information Processing, 6289/2010, pp. 344-357.

[37] D. Zhu, R. Melhem and B. Childers, “Scheduling with dynamic voltage/speed adjustment using slack reclamation in multi-processor real-time systems”, Proceedings of the 22nd IEEE Real-Time Systems Symposium, (2001), pp 686-700.