AM (myname4rwt.delete@this.jee-male.com) on August 17, 2018 2:45 pm wrote:
> Michael S (already5chosen.delete@this.yahoo.com) on August 17, 2018 3:13 am wrote:
> > Gabriele Svelto (gabriele.svelto.delete@this.gmail.com) on August 17, 2018 1:00 am wrote:
> > > ... so it's misleading by definition. That being said it's a shame because it seems to me
> > > that A76 has merits and ARM shouldn't have to massage the numbers to make it look good.
> >
> > I beleive it when I see it.
> > This particular design team promised power efficient cores in the past. But until now never delivered.
> > It is true that 76 is not a yet another descendent of 15, but I still want measuremnts.
>
> I thought you didn't follow ARM tech,

Strange thought. In today's world one can't expect to find many people that are interested in CPUs and don't follow ARM. at least at shallow level.

> but in case you do, and there's some actual knowledge behind
> your skepticism, how about you be more specific as to what ARM's claims wrt A76 make you skeptical
> and, thus, what are your well-educated predictions (in figures please: ARM's projections you don't
> believe vs what you consider likely). Naturally for the case when A76 is unconstrained by factors
> like end product manufacturer's artificial limit imposed on a product's top clock rate or lower-than-expected
> mem lat/bw or piss-poor dvfs config (as was the notorious case of Exynos 9810).
>
> I'm also curious about the "promised but never delivered power-efficient cores"
> bit. Are you referring to your own measurements or published figures?

I don't believe that A76 will have better perf/watt at the same performance than A75 when both manufactured on the same process and A75 not pushed behind its natural frequency limits. Same for A76 vs A73.
Of course, you can say that "natural frequency limits" are vague words. Yes, they are. But not too much.