SGI's OpenGL implementation seeks to achieve ``maximum performance'' OpenGL rendering. For our purposes, ``maximum performance'' means that when there is no contention for rendering resources, and once utilized resources are made available, graphics rendering performance is limited only by the system's raw graphics performance and the graphics software efficiency. The maximum performance potential of the workstation should be achievable.
In practice, this means no locks need to be acquired and released when rendering. Even so, multiple OpenGL programs should be able to run concurrently. But the overhead from concurrent use of graphics should only be introduced when multiple processes are concurrently using the graphics hardware; performance of the single renderer case should not be compromised. Graphics programs should not be burdened with overhead from window clipping; in particular, multi-pass rendering for correct clipping should not be necessary. And the possibility of asynchronous window management operations such as changing window clipping or changing the window origin should not add any overhead to the normal case when clips and origins are not changing.