I had always thought that the geforce 2 could support 32bit zbuffer depth… but today i found that mine only supports up to 24bit… or is this becase the other 8bits are reserved for a stencil buffer?
Only two modes supported:
1: 16 bit color and 16 bit depth
2: 32 bit color and a combined 24/8 depth/stencil
But what’s the need for a 32-bit depth buffer? A single precision float (which I believe is the format the driver/hardware uses) only has 23 bit mantissa, and you can’t get more precision than the mantissa can handle.
Thanks, now i understand 
Please excuse my lack of knowledge