Intel Corp (INTC)

[chipguy]

Let's give him the benefit of the doubt.

Then at the very least they should have scaled the numeric scores
of the mobile and desktop versions differently to make it obvious
they weren't comparable (like SPEC did between CPU 2000 and
CPU 2006). There is no obvious or even subtle indication which
version a result came from. That is completely unacceptable.

We aren't talking quantum chronodynamics here. The difference
between good and bad benchmark practices has been long and
widely known. Geekbench is effectively a hidden multi-tier
Dhrystone on steroids with strong influence by specific idiomatic
niche instructions thrown in for good measure. A poor outing on
version one is understandable if regrettable. Still doing this crap
in version three is something else altogether.

There is no benefit of the doubt to be given.

BTW, since you visit RWT then why don't you do a search on what
Linus Torvalds had to say about it there. He is probably even less
of a fan of it than I am.