這篇論文是一個嵌入式系統上圖形處理器的大型計劃中的一部份，為了能支援這個計劃，這篇論文設計了一個簡單的編譯器給這個計劃中的圖型處理器。這個編譯器在本質上是一個C的編譯器，但我們卻能支援到OpenGLES 2.0語言(GLSL)的的編譯。因為我們有一個OpenGL到C的轉換程式可以使用。轉換出來的組合語言並不突出，因為它在OpenGL到 C的轉換的過程會遺漏一些應該有的資訊。
而這篇論文並沒有特別強調如何編譯GLSL的程式碼如何被編譯，因為那是另一位同組計劃組員的論文。在他的論文中將配合這篇論文的後端(backend)完成前端(front end)。三個主要後端的挑戰在於，一、新GPU指令的選擇。二、暫存器的分配在只有一種暫存器類型的情況下。三、保留變數中向量和矩陣的資訊給暫存器。對於產生出來的程式碼的品質，我們只能算是初期的階段。

This thesis work is part of a larger project to develop a Graphics Processing Unit (GPU) for embedded systems. In support of this larger project, this thesis describes the basic compiler for our new GPU. The compiler presented here is essentially a C compiler for our new GPU. Our compiler does, however, have indirect support for the OpenGLES 2.0 shading language (GLSL), because an OpenGL-to-C converter can be applied to the code. The final code produced by this approach is poor, however, because of the loss of information during the conversion to C.
The reason that this thesis does not focus on the issues of compiling GLSL code directly is that that is the work of another student’s thesis. Essentially, that thesis solves the front-end while this thesis solves the problems in the backend. The three primary challenges for the backend are: 1) instruction selection to the new GPU, 2) register allocation for the GPU’s unified register set, and 3) maintaining vector and matrix information from variables to the register they are stored in. Our preliminary results are shown, for evaluation of the quality of the compiled code.

1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 GPU hardware features that impact our compiler development . . . . . . . . . . . . .2

1.2 Compiler development along two simultaneous tracks . . . . . . . . . . . . . . . . . . . 7

2 RELATED WORK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.1 Interacting software packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.2 Other GPU compilers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.3 Compiler Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3 IMPLEMENTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...18

3.1 Implementing our register allocator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3.2 Implementing our instruction selector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20

3.3 Other modifications of LLVM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25

4. RESULT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29

5. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

[1] R. J. Rost, “OpenGL Shading Language, Second Edition.” Reading, MA, USA: Addison-Wesley, 2006.
[2] The LLVM Compiler. Website: llvm.org
[3] F. W. Tsai and C. L. Lee, “An Emulator for OpenGL ES 2.0 based on C-language Compiler.” Masters Thesis: National Sun Yat-Sen University, August 2008.
[4] G. F. Tseng and M. C. Chiang, “OpenGL ES 2.0 Shading Language to LLVM Compiler” Internal document, 2008. http://cooldavid.org/cooldavid.html
[5] K. A. Huang and S. Haga. “Compiler Support for Vector Processing on OpenGL ES 2.0 Programs,” Masters Thesis: National Sun Yat-Sen University, August 2010.
[6] S. C. Tseng and S. Haga. “Compiler/Hardware Codesign and Memory Management for a Novel 3D Graphics Processor,” Masters Thesis: National Sun Yat-Sen University, August 2010.
[7] S. Ryoo, C. Rodrigues, S. Baghsorkhi, S. Stone, D. Kirk, and W. Hwu. “Optimization Principles and Application Performance Evaluation of a Multithreaded GPU Using CUDA” Proceedings of the 13th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, Salt Lake City, UT, USA, February 20-23, 2008.
[8] S. Ryoo, C. Rodrigues, S. Stone, J. Stratton, S. Ueng, S. Baghsorkhi, W. Hwu, “Program Optimization Carving for GPU Computing,” The Journal of Parallel and Distributed Computing, 68(10), 2008.
[9] C. Lattner. “LLVM for OpenGL and other stuff.” LLVM Designers Conference, May 2007
[10] W. Mark and K. Proudfoot. “Compiling to a VLIW Fragment Pipeline.” Proceedings of SIGGRAPH/Eurographics Workshop on Graphics Hardware, Los Angeles California. July 2001
[11] M. Peercy, M. Olano, J. Airey, and P. Ungar. “Interactive Multi-Pass Programmable Shading”. Proceedings of the 27th annual conference on Computer graphics and interactive techniques, New Orleans, Louisiana. July 2000.
[12] A. Aho, M. Lam, R. Sethi and J. Ullman, “Compilers: Principles, Techniques and Tools(2nd Ed)”, Pearson Addison Wesley, Hong Kong, 2006.
[13] N. Sreraman and R. Govindarajan. “A Vectorizing Compiler for Multimedia Extensions,” The International Journal of Parallel Programming. Vol. 28, No 4, 2000.
[14.] D. Nuzman, M. Namolaru and J.Derby. “Compiling for an Indirect Vector Register Architecture,” Conference on Computing Frontiers, Ischia, Italy. May 2008.