The benchmarking itself was done in a simple and
fairly un-scientific fashion.
... older Pentium IV workstation. Run-times were
measured using wall-clock time.
but still a good read about tweaking c++ (I think at the time, clang and gcc were still implementing the C++11 standard features).
Submitters: please use an article's title unless it is linkbait or misleading [2].
1. Submitted title was "Google Publishes C++, Go, Java and Scala Performance Benchmarks"
2. https://news.ycombinator.com/newsguidelines.html