std::map: is find(key)->second faster than the [] operator?

Question

std::map<long, double> x;
x[5] = 1.2;

double y = x[5];
double z = x.find(5)->second;

Will one of these 2 assignments execute faster than the other? (supposing that the requested key is always present in the map) Is there any overhead associated with the dereferencing of the iterator when doing x.find(5)->second ?

EDIT: Thanks for the replies. In my particular function, now that I know it is not slower, I will probably go with x.find(5)->second as I need to mark my function const (the map is a member variable) and the [] operator obviously does not allow that (as it potentially modifies the map is a key is missing).

Answer 1

This doesn't answer your question, but I would point out some problem with the way you're using find .

double y = x[5];
double z = x.find(5)->second;

I cannot comment on which is faster. But I can surely say that the first approach is safe!

What if the map doesn't contain the given key?

In the first approach, it will create a new pair with the given key , and initialize the value with default value of double (which is zero), and return it.

But the second aproach? find will return map::end if the specified key is not found in the container, and you're dereferencing it. Program crash!

The correct way to use find is:

std::map<long, double>::iterator it;    
if ( (it = x.find(key)) != x.end() )
{
    double value = it->second;
}

Answer 2

Took this straight from <map> :

mapped_type& operator[](const key_type& _Keyval)
    {   // find element matching _Keyval or insert with default mapped
    iterator _Where = this->lower_bound(_Keyval);
    if (_Where == this->end()
        || this->comp(_Keyval, this->_Key(_Where._Mynode())))
        _Where = this->insert(_Where,
            value_type(_Keyval, mapped_type()));
    return ((*_Where).second);
    }

iterator find(const key_type& _Keyval)
    {   // find an element in mutable sequence that matches _Keyval
    iterator _Where = lower_bound(_Keyval);
    return (_Where == end()
        || _DEBUG_LT_PRED(this->comp,
            _Keyval, _Key(_Where._Mynode()))
                ? end() : _Where);
    }

It looks about the same. Should there be any difference between:

iterator _Where = this->lower_bound(_Keyval);
return ((*_Where).second);

and

iterator i = x.find(5);
double d = (*i).second;

I wouldn't think so.

Answer 3

The first assignment using operator[] will have to perform the same dereference to retrieve the value that is explicit in find()->second . You can profile to be sure but the performance should be close enough that you should use the form that is clearest in your code.