CODEWITHSUNDEEP
c++boost
H'H
H'H

Reputation: 1678

performance analysis for boost::bind compare to for loop

I converted 

  for(std::set<shape::Face>::iterator face_iter=vec_face.begin(); face_iter!=vec_face.end(); face_iter++)
   {
     hiddenCorner(other, bmap, *face_iter);
   }

into 

for_each(vec_face.begin(), vec_face.end(), boost::bind(hiddenCorner, other, bmap, _1));

Obviously it is less verbose, but how about efficiency?

hiddenCorner is a void function which update bmap (bmap is a std::map).

NOTE: vec_face is not a really big size set.

Upvotes: 0

Views: 55

Answers (1)

Jonathan Wakely
Jonathan Wakely

Reputation: 171263

hiddenCorner is a void function which update bmap (bmap is a std::map).

The bind version makes copies of other and bmap which are stored in the binder, then passes them to the function, so the function updates the copy in the binder, not bmap itself.

So apparently not only have you not tried to measure performance yourself, you haven't even tested to see if it works. No cookie for you.

If you write it properly, using boost::ref, it should be roughly equivalent performance, assuming you turn on optimisation.

Upvotes: 5

Related Questions