subpixel precision

I just read an article on tomshardware about the QuadroFX. It claims a subpixel precision of 12 bits, while the Quadro4 has only 8 bits and FireGL only 4 bits of precision. How could I find this subpixel precision information using the OpenGL API (probably it’s not possible)? I’m trying to write a simple software renderer with 16 bits of subpixel precision, but I need to know the instruction latency for MMX and SSE instructions (couldn’t find latency information on the Intel site…). Anyone can help? Thanks.

Sure, try this:

GLint subpixel_bits ;
glGetIntegerv(GL_SUBPIXEL_BITS, &subpixel_bits) ;

There’s a lot of OpenGL state you can query with the glGet* routines. Probably worth looking at in your case.