The results arguably aren’t poor, just “good” vs “great”. That being said there are quite a few things that can cause differences across distros, so these types of comparisons are going to fluctuate a lot over time. Most notably would be stuff like security-related compile options, Fedora shipping newer software (which may have some performance gains or losses, it all depends), etc.
I’d be far more interested in benchmarks that actually show speed difference from testing similar software versions.
I just read through the benchmarks and you’ve buried the lede. The horse to beat in this race isn’t Ubuntu, it’s Clear Linux! They’ve clearly (pun intended) done some work making significant performance optimizations while all the others do some things well, other things not so well, and end up more or less the same overall.
I used to do Linux performance engineering for a living - I’m now intensely curious what’s under the hood with Clear and how that can be applied to other distros!
Yeah - they’ve compiled everything for Intel chips. It turns out they’ve a short light rail ride from my house, by the way ;-). One could get the same effect on Fedora by building all the packages locally from source RPMs using “march=native” compilation or something of that ilk.