Thanks for the reply Randy. If that's the case, then TI should simple remove the atan2sp() asm and point to the atan2sp_i() call.
Frankly, and no offense here, I find this implausible. The assembly here is hand tuned to the C674x which I'm using. Then we have some C which outperforms it? I mean... this assembly is tailor made for the part I'm using, yet the C is better? IF that's true, whoever wrote this assembly may need some major code reviews as your compiler has made him/her obsolete. Or TI needs to remove the ASM and warn customers off of using it.
I still lean toward something fishy here.