high performance computing on graphics processing units: hgpu.org

In this paper, we present a new technique for highly efficient terrain rendering using continuous view-dependent Level-of-Detail based on hardware tessellation unit found in modern GPUs. Our technique is based on parallel local processing, in the sense that the results at each terrain patch do not depend on results already obtained at other patches. This patch-by-patch processing uses no hierarchical structure whatsoever, what makes it specifically tailored for GPU-based LOD terrain and is highly scalable.