You don't need triangle strips for that and, in fact, triangle strips are generally not preferred anymore due to the difficulty of generating optimal strips from a mesh (and how little, if any, gain there is on modern graphics cards). Triangle strips have a small advantage only because they make more optimal use of the limited vertex cache.

What you really want to do in a software 3D engine is to avoid performing transformation on a vertex more than once. An indexed vertex buffer is a better choice for this. Instead of processing each triangle as 3 vertices, you put *all* vertices of a given mesh into a buffer and transform them all in one fell swoop. Then, triangles are defined by 3 indices into the output vertex buffer (which have already been transformed) so each triange is just 3 simple look-ups.


I recall this page having a lot of good information of software rendering and the software transformation pipeline in particular... The articles you want to read are "The Poly Pipeline" and "The way your 3D engine should be".