Original Post
I'm using the texture splatting technique to texture my terrain (it's relatively small: 129x129 vertices). But it is often necessary to render the terrain multiple times (e.g. to apply more terrain textures or a lightmap). So I think I should invest some time to optimize the terrain mesh. That's how it currently works: I have a vertexbuffer which holds the vertices which are written to it row by row: 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 The triangles are defined by an indexbuffer, which refers to the vertices. (I'm using triangle lists, also arranged row by row) The problem with this technique is, that nearly every vertex has to be transformed twice, because the vertex cache is to small to hold one complete row in memory (that would be 129 vertices) I also tried out the NvStrip library and some D3DX functions to optimize the mesh for optimal vertex cache usage, but the results were even worse. Maybe you could give me some advice.