coverage computing using fragment program

The OpenGL 2.0 specification calls for the ‘coverage computing’ to be NOT superseeded by ‘fragment program’, therefore making hardware accelerated smooth antialiasing available exclusively on professional video cards. I’m certainly not going to spend 900 USD on a Quadro FX 1000, but I’d spend 250-400 USD for a GeForceFX, if the coverage computing would be of the same quality (even if 2-3 times slower). Does anyone know a way to perform coverage computing by a fragment program?