> I've written a 3d software rasterizer using fixed point maths.
> In an attempt to improve performance, I've reduced the number of edges
> to calculate by preprocessing my model, generating a minimal number of
> vertex pairs(edges). Rather than create three edges for each polygon.
> I'm now looking an effective method of rasterizing these.
These _what_? Are you going to rasterize edges, or filled polygons?
If you're rasterizing filled (or even shaded) polygons, reducing the
number of edges won't really help much. The maximum gain is that
you'll have half as many edges, which will only help if your
bottleneck is the transformation of edges from 3D to screen space. In
a software-only rasterizer, I doubt this can ever be the case: polygon
scan conversion doesn't use much edge information to begin with. It's
quite certainly neglibible compared to the amount of time spent doing
the actual pixels of the polygon.
If you're going to rasterize edges, I don't see how any half-way
decent algorithm of doing that can be justifiably classified bas
"slow". Not unless your expectations about what constitutes "fast"
rendering are far from realistic.
Even if all the snow were burnt, ashes would remain.