Most of the programs that you are going to find are examples of how to use OpenMP and not necessarily examples showing speedup. A perfect example of this, is the Appendix for the OpenMP spec, which has a lot of partial code examples that can be compiled and made to run. However, I don't think any of them are going to show a speedup. This is true for most of the examples you will find on the web. One exception floating around is example code that computes pi. I have run this on quite a few machines (different architectures from 2 to 24 processors) and have seen a speedup on them all.
As for V3.0 examples, there are not many at this time. You can find a few code examples in some papers about Version 3.0 and there are some examples on the various vendor web sites (e.g., see the openmp.org web page and go to the Sun wiki). Most vendors are currently implementing the Version 3.0 specification, so there are not too many implementations of it available and they are all new (having bugs) and have not yet really been tuned for performance. As for any book that you can find, they most likely do not yet cover Version 3, since it is so new.