Another performance bottleneck is the rate at which commands are transferred from the host to the OpenGL implementation. The easiest way to improve OpenGL command bandwidth is to use direct rendering if the application is running on the local graphics workstations.
One way to minimize the OpenGL command bandwidth is to use OpenGL display lists. When rendering indirectly, display lists are downloaded into the X server so that when executed, only the command to execute the display list must be transferred to the X server. This can greatly reduce the network overhead of indirect rendering. Display lists may also be downloaded into the graphics hardware so they can be executed with no host overhead to drive the display list.
It is also possible for a fast computer and fast graphics subsystem to be starved by the interface between them. Be careful that the host's peripheral bus used to connect the graphics subsystem has sufficient bandwidth to drive the graphics at your required performance.