high performance computing on graphics processing units: hgpu.org

Global Illumination aims to generate high quality images. But due to its high requirements, it is usually quite slow. Research documented in this thesis was intended to offer a hardware and software combined acceleration solution to global illumination. The GPU (using CUDA) was the hardware part of the whole method that applied parallelism to increase performance; the "Lightcuts" algorithm proposed by Walter (2005) at SIGGRAPH 2005 acted as the software method. As the results demonstrated in this thesis, this combined method offers a satisfactory performance boost effect for relatively complex scenes.