Don't make claims about performance without measurements

Reason

The field of performance is littered with myth and bogus folklore. Modern hardware and optimizers defy naive assumptions; even experts are regularly surprised.

Note

Getting good performance measurements can be hard and require specialized tools.

Note

A few simple microbenchmarks using Unix time or the standard-library <chrono> can help dispel the most obvious myths. If you can't measure your complete system accurately, at least try to measure a few of your key operations and algorithms. A profiler can help tell you which parts of your system are performance critical. Often, you will be surprised.

???