NVIDIA於2006年開始發展以圖像處理器進行一般程式運算的技術，稱之為統一運算架構（CUDA，Compute Unified Device Architecture）。在該架構中，將單一指令 交由多個執行緒同時處理，具有平行化程式的優點，可大幅減少運算時間。CUDA提供一組與C語言相似的應用程式介面（API，Application Programming Interface），讓程式設計者方便上手，但仍需要一段開發時間。本論文提出將原本C程式自動翻譯為CUDA程式的方法，有效減少程式開發時間。

NVIDIA developed a technique of executing general program on GPU, named CUDA (Compute Unified Device Architecture), in 2006. The CUDA programming model allows a group of same instructions to execute on multi-thread simultaneously, which has advantage of parallel programs in reducing the execution time significantly. Although CUDA provides a series of C-like APIs (Application Programming Interface) so that programmers can easy use CUDA language, it still costs certain efforts to be familiar with the development. In this thesis, we propose a tool to automatically translate C programs into corresponding CUDA programs which reduce program development time effectively.

List of Tables iii
List of Figures iv
Acknowledgments v
Chapter 1 導論1
1.1 研究動機......1
1.2 研究方法......2
1.3 論文架構......3
Chapter 2 相關研究4
2.1 CUDA（Compute Unified Device Architecture）.. . 4
2.1.1 執行緒...... 4
2.1.2 記憶體架構.....5
2.2 OpenMP...... . 7
2.2.1 範例探討...... . 8
Chapter 3 研究方法10
3.1 問題描述......10
3.2 相關定義......10
3.2.1 函式轉換...... . 11
3.2.1.1 #pragma ctocuda kernel... . 11
3.2.1.2 #pragma ctocuda loop... . . 11
3.2.2 變數指標宣告..... . . . . 11
3.2.2.1 #pragma ctocuda var decl... 11
3.2.3 記憶體管理.....11
3.2.3.1 #pragma ctocuda mem alloc.. . . . . 11
3.2.3.2 #pragma ctocuda mem copy.. . . . . 11
3.2.3.3 #pragma ctocuda mem free..12
3.2.4 函式呼叫...... . 12
3.2.4.1 #pragma ctocuda kernel invoc.. . . . . 12
3.3 範例探討......12
3.3.1 原始C程式...... 12
3.3.2 CUDA程式...... 14
3.3.3 範例解說...... . 16
Chapter 4 實驗設定與結果17
4.1 實驗環境......17
4.2 執行流程......18
4.3 實驗結果......18
Chapter 5 結論與未來目標20
5.1 結論....... . . 20
5.2 未來目標......20
Appendix A Useful CUDA API 22
A.1 Memory Management..... 22
A.1.1 cudaMalloc()..... 22
A.1.2 cudaFree()..... . 23
A.1.3 cudaMemcpy()....23
A.2 Thread Management..... 24
A.2.1 cudaThreadSynchronize()...24
A.3 Device Management..... 24
A.3.1 cudaGetDeviceCount().... . 24
A.3.2 cudaGetDeviceProperties()...25

[1] http://www.nvidia.com.tw/page/home.html, “Nvidia.”
[2] http://openmp.org/wp/, “Openmp.”