Bug between results and diagram on PerfTest

[roondar]

Hello, just a little message because I think that I found a little bug between your final mark and the diagram of PerformanceTest.



I guess the result is good, i'm at 54% in this categorie, but look your diagram...
It show me at 54% of the orange part, but i'm behind the "50% mark".
Because this mark is good. But the "hand" (i'm not sure in eng) is based on a linear patern, unlike the diagram, so it point behind the good point.

I hope i help you for your very good software ^^

Et en français ça me sera plus simple La marque des 50% est bonne car c'est une échelle un peu particulière. Mais la flèche semble codé de façon linéaire sur la zone orange et ne pointe donc pas au bon endroit. Ce qui est sûr, c'est qu'il y a incohérence entre la marque "50%" et ce que pointe la flèche.

Merci de me tenir au courant si jamais j'ai vu un truc, ce serait cool
Encore meilleur pour votre travail. roondar.

[David (PassMark)]

The dial graph is linear for the CPUMark value (called "Note UCT" in French). This is the white circular scale going from 0 to 63,000 in your example.

But the percentage of people getting these scores (percentile, or "Centile" in French) is not linear. Most people get scores around the average value. So the green bar is much larger than the orange and red bars. This means the top 25% of scores covers a very big range of values. The other quartiles cover a smaller range.

Your percentile value represents the number of people who got better or worse scores than you. So 54% percent of people with this CPU got scores lower than you. 46% got higher scores. Importantly this doesn't mean the 50% percentile is the average score. It would if the distribution of results was a perfect bell curve. But this isn't the case in the real world. The average benchmark score is somewhat higher that what 50% of people got.

Imagine the scores being: 2, 2, 3, 8, 16
The 50th percentile is 3
The average is 6

In short: The average result and the 50% percentile aren't always the same.