M.Sc. Tezi Görüntüleme

Student: Ömer ÇAKIR
Supervisor: Asst. Prof. Dr. Cemal KÖSE
Department: Computer Engineering
Institution: Graduate School of Natural and Applied Sciences
University: Karadeniz Technical University, Turkey
Title of the Thesis: PARALLEL RAY TRACING ON NETWORKED PERSONAL COMPUTERS
Level: M.Sc.
Acceptance Date: 2/8/2004
Number of Pages: 86
Registration Number: i1502
Summary:

      Ray tracing technique is commonly used to generate realistic images in computer graphics. Ray tracing is a realistic technique to describe the interaction between rays spreading from a light source and objects. Although ray tracing is a good technique for rendering a high quality three-dimensional image, it requires rather much processing time. Parallel ray tracing technique for general-purpose multi-computer system offers the potential for realistic rendering such complex scenes in a reasonable time. Therefore, real-time image rendering may be achieved when sufficient numbers of computers are used.

      In this study, ray tracing technique is performed for high quality three-dimensional image rendering. Phong illumination model is employed to determine the intensity and color values at each pixel. The quality of the image is improved by performing texture mapping. Applying bump mapping technique may produce even more realistic textured three-dimensional images. Performing super-sampling solves aliasing encountered in the image and improves the quality of the final image. Then, back-face culling and bounding volume methods are employed to accelerate ray tracing. In addition, interactive ray tracing is achieved to allow a user to move inside the scene.

      In ray tracing, calculating the color value of a pixel is independent of others' which makes parallelisation of ray tracing technique rather simple. Thus, personal computers on a LAN can be effectively used as a parallel ray tracing platform. The ray tracing has a computational complex problem and so involves using a dynamic load balancing mechanism to ensure the best performance on a heterogeneous parallel computing platform. Therefore, the processor farm model is used for dynamic load balancing, where a task pool is maintained on the master process. In this way each application process on the parallel system requests a task from the master process, performs it, and requests another untill the task pool is empty. Result shows that parallel ray tracing achieves almost linear speedup on the parallel system. As a result the parallel ray tracing technique uses the dynamic load balancing mechanism to produce high quality three-dimensional images in a reasonable time.

      

      Keywords: Ray Tracing, Parallel Ray Tracing, Processor Farm, Dynamic Load Balancing.