Research And Design Ray Tracing Accelerated Algorithm Based On Heterogeneous Multi-core.

With multi-core technology developing, more and more researchers pay attention to using multi-core processors and many-core processor (GPU can be considered as many-core processor). With the rapidly increasing computing power and the growing of the levels of the programmability of the GPU, it has become a cost-effective parallel computing coprocessor, particularly suitable for intensive and high parallel general purpose computing. Researchers have begun to introduce the GPU to do a variety of intensive calculations based on CPU. In the Computer Graphics, as an important method for generating photorealistic images, ray tracing has been widely used in the film entertainment industry. Because ray tracing is of highly parallel, therefore, it becomes a trend to accelerate ray tracing algorithm by GPU.This paper firstly discusses three platforms to achieve the accelerated structure of ray tracing algorithm. By comparing, proves that heterogeneous multi-core platform is more suitable for building KD-tree. Secondly, implements the algorithm of completely rebuild dynamic ray tracing on the heterogeneous multi-core platform and analyzes performance of the algorithm. Lastly, proposes two dynamic ray tracing algorithms and analyzes their performance. The main content of this paper is as follows:1) Study the basic principle of ray tracing algorithm; select the suitable accelerated structure and construct ray tracing model system on the GPU which supports the CUDA.2) Study the accelerated structure of ray tracing on the CPU,GPU and heterogeneous multi-core platform. And prove that the heterogeneous multi-core platform is more suitable for building the accelerated structure of ray tracing algorithm by testing different static three-dimension scenes.3) Propose the partly reconstructed and rays coordinate transferred dynamic ray tracing after studying and achieving the completely rebuild dynamic ray tracing. Analyze the performance of these three algorithms and get the performance when these three algorithms apply on the heterogeneous multi-core platforms.