臺灣博碩士論文加值系統

雲端服務於最近在網路服務中成為一大趨勢，一些全球性的企業，如Google、Amazon、Microsoft等，釋出了多個不同種類的雲端服務，而這些服務大多分為SaaS、PaaS、IaaS三種服務架構。為了服務簡易性以及維護方便性，大多數的雲端服務平台使用虛擬化技術作為服務架構基礎。另外近年來由於通用圖形處理器的發展，高速運算也成為熱門項目，不少研究單位或是企業都選用通用圖形處理器做為高速運算平台的基礎。
因此，以圖形處理器發展高速雲端運算，一直是不少企業的目標，礙於虛擬機器難以使用圖形處理器，造成高速雲端運算平台虛擬化的困難，尤其開源的雲端平台(如OpenStack)還無法支援圖形處理器。
本研究將探討建構基於虛擬化技術的圖形處理器雲端運算架構之可能性。我們使用了新的方法，建立這個架構，並進行測試及問題探討。透過這樣的方式，我們便能夠在虛擬機器中使用圖形處理器及通用圖形處理器，並使用這個架構，進行更多的應用。

Recently, the cloud computing service has become a trend in the Internet industries. Many global firms, such as google, amazon, Microsoft, and so forth, have released a variety of cloud services. Most of these services are classified into IaaS, PaaS, and SaaS service architectures. For the simple service and effective maintenance, most of the cloud service platforms usually adopt virtualization technology to construct the infrastructure of the platforms. In addition, the growth of GPGPU is very rapid. Therefore many researchers and firms are using GPGPU to construct their high-performance computing platforms.
Therefore, for many industries to build a high-performance cloud computing platform based on GPGPU is their goal. However, it is the problem to make virtual machine to drive GPGPU; especially most of open-source cloud platforms (i.e OpenStack) have not supported GPGPU.
In this study, the possibility of building the infrastructure of GPU cloud computing platform based on virtualization technology is investigated. Meanwhile, a novel method has been proposed to build this kind of infrastructure, and the infrastructure has been test by several benchmarks, and the results show that it is useful. Through this method, the virtual machine is able to execute any application laying on GPGPU.

摘要 I
Abstract II
謝誌 III
一、前言 1
二、文獻探討 3
2.1雲端運算 3
2.2虛擬機器 4
2.3GPGPU 7
2.4GPU應用於雲端運算之挑戰 10
三、開源的解決方案 15
3.1其他的解決方案 15
3.2GPGPU程式 17
3.3以VFIO建構以虛擬機器為基礎之GPU雲端運算架構 22
四、實驗 26
4.1測試方法 26
4.2測試結果 27
五、討論 33
5.1KVM+VFIO運行限制與虛擬機器運行問題 33
5.2KVM+VFIO的可能應用 38
六、結論 40
參考文獻 41

M. Armbrust, A. Fox, R. Griffith, A. D. Joseph, R. Katz, A. Konwinski, et al., "A view of cloud computing," Commun. ACM, vol. 53, pp. 50-58, 2010.
G. J. Popek and R. P. Goldberg, "Formal requirements for virtualizable third generation architectures," Commun. ACM, vol. 17, pp. 412-421, 1974.
J. Nickolls and W. J. Dally, "The GPU Computing Era," Micro, IEEE, vol. 30, pp. 56-69, 2010.
J. D. Owens, D. Luebke, N. Govindaraju, M. Harris, J. Krüger, A. E. Lefohn, et al., "A Survey of General-Purpose Computation on Graphics Hardware," Computer Graphics Forum, vol. 26, pp. 80-113, 2007.
NVIDIA CUDA. Available: https://developer.nvidia.com/cuda-zone
Khronos Group OpenCL. Available: https://www.khronos.org/opencl/
A. S. Tanenbaum and R. V. Renesse, "Distributed operating systems," ACM Comput. Surv., vol. 17, pp. 419-470, 1985.
M. Dowty and J. Sugerman, "GPU virtualization on VMware's hosted I/O architecture," SIGOPS Oper. Syst. Rev., vol. 43, pp. 73-82, 2009.
VMware. White Paper: Graphics Acceleration in View Virtual Desktops. Available: http://www.vmware.com/files/pdf/techpaper/vmware-horizon-view-graphics-acceleration-deployment.pdf
A. Herrera. (2014). White Paper: NVIDIA GRID: Graphics Acclerated VDI with the Visual Performance of a Workstation. Available: http://www.nvidia.com/content/grid/resources/White_paper_graphics_accelerated_VDI_v1.pdf
NVIDIA. NVIDIA GRID Infomation. Available: http://www.nvidia.com.tw/object/grid-boards-tw.html
NVIDIA. (2013). NVIDIA GRID K1 Graphic Board Specification. Available: http://www.nvidia.com/content/grid/pdf/GRID_K1_BD-06633-001_v02.pdf
NVIDIA. (2013). NVIDIA GRID K2 Graphic Board Specification. Available: http://www.nvidia.com/content/grid/pdf/GRID_K2_BD-06580-001_v02.pdf
Linux Kernel Virtual Machine. Available: http://www.linux-kvm.org/page/Main_Page
A. Bakhoda, G. L. Yuan, W. W. L. Fung, H. Wong, and T. M. Aamodt, "Analyzing CUDA workloads using a detailed GPU simulator," in Performance Analysis of Systems and Software, 2009. ISPASS 2009. IEEE International Symposium on, 2009, pp. 163-174.
Folding@Home. Available: https://folding.stanford.edu/
GPU-Blast. Available: http://archimedes.cheme.cmu.edu/?q=gpublast
K. Komatsu, K. Sato, Y. Arai, K. Koyama, H. Takizawa, and H. Kobayashi, "Evaluating performance and portability of OpenCL programs," in Proceedings of the Fifth international Workshop on Automatic Performance Tuning(iWAPT2010), Berkeley, USA, June 2010.
3DMark. Available: http://www.3dmark.com/zh-hant/
Assassin's Creed: Unity. Available: http://www.ubisoft.com.tw/acu/
Crysis 3. Available: http://www.crysis.com/crysis-3
Unigine Valley Benchmark. Available: https://unigine.com/products/benchmarks/valley/
RandomControl Arion Benchmark. Available: http://www.randomcontrol.com/arionbench
Luxmark. Available: http://www.luxrender.net/wiki/LuxMark
CompuBenchCL. Available: https://compubench.com/result.jsp?benchmark=compu15d
P. Willmann, S. Rixner, and A. L. Cox, "Protection strategies for direct access to virtualized I/O devices," presented at the USENIX 2008 Annual Technical Conference, Boston, Massachusetts, 2008.
Open Virtual Machine Firmware. Available: http://www.tianocore.org/ovmf/