Display List Optimization

Accumulating commands into a group for repeated execution presents possibilities for optimization. Consider, for example, specifying a texture image. Texture images are often large, requiring a large, and therefore possibly slow, data transfer from client to server (or from the server to its graphics subsystem) whenever the image is respecified. For this reason, some graphics subsystems are equipped with sufficient storage to hold several texture images simultaneously. If the texture image definition is placed in a display list, then the server may be able to load that image just once when it is specified. When the display list is invoked (or re-invoked), the server simply indicates to the graphics subsystem that it should use the texture image already present in its memory, thus avoiding the overhead of respecifying the entire image.

Examples like this one indicate that display list optimization is required to achieve the best performance. In the case of texture image loading, the server is expected to recognize that a display list contains texture image information and to use that information appropriately. This expectation places a burden on the OpenGL implementor to make sure that special display list cases are treated as efficiently as possible. It also places a burden on the application writer to know to use display lists in cases where doing so could improve performance. Another possibility would have been to introduce special commands for functions that can be poor performers in immediate mode. But such specialization would clutter the API and blur the clear distinction between immediate mode and display lists.