I am complaining that you don't have a C program using the AVX instructions (in C).
All the AVX functions are directly accessible from C (see: intrinsics). For a valid comparison, I would expect a C program using AVX instructions against an assembly program using AVX.
Your comparison has C-without-AVX against ASM-with-AVX. Of course, it's faster with AVX ;)
The various AVX.c versions don't use any AVX instructions. That is deceptive.