high performance computing on graphics processing units: hgpu.org

Dealing with high-resolution imagery with billions or trillions of samples is an enormous challenge that oftenoverwhelms the graphics subsystem of any computer. Silicon Graphics, Inc. addressed this issue by providing explicit hardwaresupport for offset registers and texture sub-loads in their InfiniteReality machine. The clipmap algorithm uses sub-textures andincremental updates based on a toroidal mapping to allow a smaller region to have high-resolution data, while surroundingregions have gradually decreasing resolution. To date, this capability is not supported on most graphics cards. This paperexamines several strategies to support clipmapping on commodity GPUs using programmable shaders. We analyze issuesassociated with clipmaps and provide a taxonomy for clipmap rendering. We have implemented several shaders in our taxonomyand provide performance analysis on current GPUs. Proposals for more efficient hardware to support clipmapping are alsopresented. Throughout the paper, we provide several extensions to the basic clipmap algorithm, including support for compressedtextures, better load balancing, and better support for missing data. Results are presented using our real-time flight simulatorapplied to large-scale databases.