Over clocking doesn’t count for normal users. There are reasons the systems don’t come from the factory that way. The bottom line is when I run software “x” on one machine and the same software “x” on another machine I will call the faster machine “faster”. But as stated I have not made a scientific test of this to prove it. In general all of the Broadcom chips have felt slower (rPi2 or 3) than the other variants of similarly spec’d ARM chips (Samsung, Allwinner).
I think I also need to put this another way: a 1GHz to a 2Ghz comparison is an “apples and oranges” scenario. If the broadcom part is less efficient for some reason, all you’ve proved is that doing twice as many less efficient instructions per second is faster. I would hope so! 
Still, have fun burning down the barn! 
