Driving tiled displays with an extended chromium system based on stream cached multicast communication

When driving tiled displays using the Chromium’s Tilesort-SPU the performance bottlenecks are the client’s high utilization caused by expensive sorting and bucketing of geometry and the high bandwidth consumption caused by too many redundant unicast transmissions. We addressed these problems with an implementation of a true point-to-multipoint connection using UDP multicast. A transparent stream cache avoids redundant transmissions of command sequences that are generated by applications several times. In addition, frustum and hardware accelerated occlusion culling methods may be used to eliminate unnecessary transfer of invisible geometry. Finally, a software based method for synchronization of buffer swap operations at all servers was implemented. In a nutshell, for the first time an appropriate combination of our optimizations makes it possible to render large scenes synchronously on an arbitrary number of tiles at nearly constant performance.