boost::shared_ptr<> uses a locking mechanism when copies of a variable are made in order to increment the counter correctly in a multi-threaded application. For example: int main( int, char ** ) { boost::shared_ptr< ABC > sp1( new ABC ); for ( int i = 0; i < 1000000; ++i ) { boost::shared_ptr< ABC > sp2 = sp1; } return 0; } I would like to know if there are any platform-specific timing tests that have been performed where operational speed results have been recorded. I want to know what kind of performance I should expect to achieve on different operating systems upon duplication and destruction of shared pointers. Thanks, -Sid Sacek