Overhead for Degenerate Triangles

Hi everyone,
I’m wondering how much overhead is created during rendering by using degenerate triangles.
I’m thinking about a case where you have e.g. 10,000 triangles, with 5,000 of them being degenerate after the vertex processing stage (after the vertex shader has been applied).

I assume it would be more efficient to render only the 5,000 non-degenerate triangles then, as we will save the vertex processing overhead for the degenerate triangles. The question is how strong this effect will usually be - are there any whitepapers or so that mention some numbers about that issue?

Thanks a lot in advance!


… no ideas anyone?

I’d also be happy to receive some reference to further help sources, any comments are appreciated!

Thanks in advance.