high performance computing on graphics processing units: hgpu.org

The adaptive subdivision step for surface tessellation is a key component of the Reyes rendering pipeline. While this operation has been successfully parallelized for execution on the GPU using a breadth-first traversal, the resulting implementations are limited by their high worst-case memory consumption and high global memory bandwidth utilization. This report proposes an alternate strategy that allows limiting the amount of necessary memory by controlling the number of assigned worker threads. This makes it possible to severely reduce the amount of necessary memory and place all intermediate data in work-group local memory. The result is an implementation that scales to the performance of the breadth-first approach while offering three new advantages: significantly decreased memory usage, a smooth and predictable tradeoff between memory usage and performance, and increased locality for patch processing.