隨著資訊科技的進步，計算密集、與資料密集型的應用如雨後春筍般相繼問世。為了處理大量運算需求，人們提出多核心處理器架構、叢集運算、採用圖形處理器(GPU)平行加速等等的技術以提供龐大的運算能力。此外，雲端技術蓬勃發展，雲端計算服務及雲端程式整合開發環境亦成為一種趨勢。有鑑於上述發展，計算服務不再需要超級電腦或額外硬體元件的設備花費，只需使用智慧型裝置使用雲端系統所提供的服務即可，為這些高計算密集、資料密集型的應用提供了ㄧ個經濟實惠的解。
本篇論文以資料流程式設計作為叢集運算程式開發模型，整合圖形處理器運算、雲端整合開發環境，並著眼於異質機器的使用與網頁服務的建置，打造一個具高可擴充性、圖形化操作介面的雲端GPU叢集運算整合開發環境。在本論文中將描述本技術在各種不同使用狀況下的使用狀況，及本技術的未來發展可能及應用彈性。

With the progress of information technology, computing-intensive and data-intensive applications sprang up like mushrooms. Because of the demand, multi-core processor architecture, cluster computing, and parallel computing on graphical processor unit were developed to provide enormous computing power. Moreover, because of facility charge of super computer or extra hardware devices, cloud computing service and cloud integrated environment became an emerging trend. What you have to do is to connect to internet and to utilize the service provided by cloud system. It provides a cheaper and affordable solution for applications.
While various parallel programming models are available, this thesis introduces dataflow programming model due to the growing popularity of component based software design for cluster computing. Here, we also integrate GPU technology and cloud IDE, and focus on utilization of heterogeneous machines and construction of convenient web services to build a framework for cloud GPU cluster computing integrated development environment (IDE) with high scalability and user friendly graphical user interface of the programming environment. In this thesis, various use-scenarios are introduced to reveal the possible potential and broad usage of the proposed system.

[1] StreamIt, http://groups.csail.mit.edu/cag/streamit/
[2] Yu-Hung Guo, An Embedded Middleware for Component-Based Software Development and the Development Environment, master thesis, NCKU, 2011
[3] SystemC library, http://www.accellera.org/downloads/standards/systemc
[4] MPICH2, http://phase.hpcc.jp/mirrors/mpi/mpich2/index.htm
[5] Tje-Rong Liu, A Many-Processor Prototyping SW/HW Framework and Component Based Dataflow Programming, master thesis, NCKU, 2011
[6] Shih-Tun Yen, Exploiting High Speed FPGA Interconnect to Improve Performance of Message Passing, master thesis, NCKU, 2011
[7] OpenMP, http://openmp.org/wp/
[8] Microsoft Task Parallel Library, http://msdn.microsoft.com/en-us/library/dd460717.aspx
[9] NVidia CUDA, http://docs.nvidia.com/cuda/index.html
[10] Hadoop, http://hadoop.apache.org
[11] Hadoop distributed filesystem, http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html
[12] Jeffrey Shafer, Scott Rixner, and Alan L. Cox, The Hadoop Distributed Filesystem: Balancing Portability and Performance, Rice University, Houston, TX, IEEE 2011
[13] Cloud 9, https://c9.io/
[14] Node.Js, http://nodejs.org/
[15] Fancybox, http://fancyapps.com/fancybox/
[16] JNI, http://docs.oracle.com/javase/6/docs/technotes/guides/jni/
[17] SMIMS Macube manual