From page 23 in the report: “Most of the comments on runtime costs are based on a set of simple measurements performed on three different machine architectures using six different compilers , with different optimization options. Each test was run several times to ensure that the results are repeatable. Code presented in Appendix D :. The purpose of these measurements is not to obtain an accurate description of the optimal C ++ performance on a given machine and not to provide comparisons between compilers or and machine architecture, rather, the goal is to give developers an idea of the relative costs of common language constructs using modern compilers, and to show what is possible (which, in principle, is possible for everyone).
So, I think, if you really want to know, you can ask the authors of the report.
source share