high performance computing on graphics processing units: hgpu.org

Ray tracing is a computer graphics rendering technique. Different from the traditional rasterization algorithm, the ray tracing algorithm simulate the realvision process. Being able to deliver highly realistic graphics effects, it has been considered as the fundamental graphics rendering mechanism for high-end applications and is also likely to be adopted as the work-horse of future graphics hardware. However, the high computational effort is the major stumbling block for ray tracing to be deployed in real-time applications. In this paper we present a detailed characterization of the ray tracing algorithm by focusing on ray tracing algorithm based on BVHacceleration structure. Using a state of art graphics processor unit (GPU) simulator, we are able to provide key insight on improving the performance of ray tracing onGPUs. We propose thread migration technique to reduce the cache miss rate and enhance the processing throughput of primary and secondary rays.