Unfortunately not. I think people at this point just learn by doing. I started by writing an SSE matrix-vector multiply routine, then ray-triangle intersection code, ray-box, triangle-box... Basic, well-known algorithms that are heavy on vector math.

Grab AMD's CodeAnalyst and learn how to use "Pipeline Simulation Mode" so you can see which instructions are stalling and for how long. If you search for "SSE" in the forums, I've posted a decent amount of code.

Well, Intel's docs are only reference. They don't contain any useful code. You can search on google for sse and maybe you will find some open source libs that use it. I found few examples that way and it was pretty helpful.
This is test written by one member of gamedev (sorry, I forgot name [smile]). It contains most commonly used vector operations. It's in delphi. Maybe it will help you.

EDIT: Ahh, his name is LoreKeeper.