The MSDN is not totally correct: A8R8G8B8 and A1R5G5B5 also allow GetDC. The
main differences that have to be filtered out in d3d9.dll are GetDC on
A8B8G8R8, X8B8G8R8 and P8.

The KIL is quite expensive because it forces drivers to disable early Z
discard. It is cheaper to generate and switch between two shaders.