I disagree with the article's claim that the geometric mean lacks physical meaning.
Say you have two benchmarks with different units, frames/second and instructions/second. You can't take the arithmetic mean of these unless you divide by some baseline first (or a conversion factor).
But the geometric mean has well-defined units of (frames * instructions)^.5/second. And the reason you can divide by another geometric mean is because these units are always the same.
Having coherent units isn't exactly the same as "physical meaning", but it's a prerequisite at the least.
Geometric average is just estimating the mean of logarithmic units. Units of measurement are arbitrary, and justification for using logarithmic units for statistical purposes can be easily made with distributional considerations. E.g. durations are bounded to be positive, so there will be at least some skew in the distribution (although this can be negligible in practice).
Say you have two benchmarks with different units, frames/second and instructions/second. You can't take the arithmetic mean of these unless you divide by some baseline first (or a conversion factor).
But the geometric mean has well-defined units of (frames * instructions)^.5/second. And the reason you can divide by another geometric mean is because these units are always the same.
Having coherent units isn't exactly the same as "physical meaning", but it's a prerequisite at the least.