Am I the only one that doesn't see a HUGE difference between distros with these tests..... ever?
If they release about the same time, they perform basically the same. What I find interesting is testing the last 5 releases to see how a single distro has changed over time.
Unless I'm mistaken the PTS always builds the test package set on the test machine - as such, performance difference can only be a result of an older GPU driver, (radically different) kernel configuration and/or, older GCC versions.
However, once you start comparing the distribution binary packages (E.g. comparing Fedora postgresql RPM vs. Debian postgresql deb packages) you might actually see far bigger changes, as a result of difference compile options, different library stacks, etc.