clock cycles of fmod()?

Can someone please share how many clock cycles fmod() takes?
Also, what is the penalty of casting from integer to float?

The documentation suggests that there is a severe penalty for integer mod, but I am curious what the penalty is for something like this:

iMod = (int)modf( (float)iVal, (float)iModVal );

where I cast types to float, do the floating point modulus, then convert back to int. less severe?