By: anon (anon.anon.delete@this.anon.com), November 25, 2020 5:41 pm
Room: Moderated Discussions
Dummond D. Slow (mental.delete@this.protozoa.us) on November 25, 2020 10:27 am wrote:
> Doug S (foo.delete@this.bar.bar) on November 25, 2020 9:51 am wrote:
> > Dummond D. Slow (mental.delete@this.protozoa.us) on November 25, 2020 9:47 am wrote:
> > > ARM Cortex cores don't lose on energy efficiency, you are mistaken.
> > >
> > > Look here: https://www.anandtech.com/show/14892/the-apple-iphone-11-pro-and-max-review/4
> > >
> > > The "joules" values clearly tell you the Cortex A76 in Snapdragon 855 (and the Kirin 990) consumed
> > > less energy than the Apple core even though it had to run longer to finish the computation. And
> > > the difference is not small (after all, it can be 2W versus 5W in power consumed).
> > >
> > > Hence, the power/watt ratio a.k.a. efficiency is better with Cortexes. ARM designs
> > > the cores for higher efficiency (besides small silicon area) and it shows.
> >
> >
> > If you don't care how fast it runs then you could either clock down the M1 core or use Apple's efficiency
> > cores (~1/3 the speed of the big core at ~1/10th of the power) to beat A76 in performance/watt.
>
> Or you could downclock the Cortex too, if we are moving goalposts.
>
> The point is, the Axx worshipping club likes to pretend the gap in absolute performance between
> Apple and ARM H. cores is all about skills and/or magic of the Apple's tean (the dumb "ARM
> is 2 years behind" narrative that certain poster here loves) and nothing else.
>
> But it is in fact ARM's targetting low silicon area and low power consumption,
> because that is what the licensors want while Apple cares not. It might even
> be overwhelming majority of the factors involved in the difference.
Why talk about these things in hypotheticals when nuvia has published actual data on how these cores compare across their voltage/frequency curves?: https://nuviainc.com/blog/historical-perspective-on-the-past-20-years-in-power-management
> Doug S (foo.delete@this.bar.bar) on November 25, 2020 9:51 am wrote:
> > Dummond D. Slow (mental.delete@this.protozoa.us) on November 25, 2020 9:47 am wrote:
> > > ARM Cortex cores don't lose on energy efficiency, you are mistaken.
> > >
> > > Look here: https://www.anandtech.com/show/14892/the-apple-iphone-11-pro-and-max-review/4
> > >
> > > The "joules" values clearly tell you the Cortex A76 in Snapdragon 855 (and the Kirin 990) consumed
> > > less energy than the Apple core even though it had to run longer to finish the computation. And
> > > the difference is not small (after all, it can be 2W versus 5W in power consumed).
> > >
> > > Hence, the power/watt ratio a.k.a. efficiency is better with Cortexes. ARM designs
> > > the cores for higher efficiency (besides small silicon area) and it shows.
> >
> >
> > If you don't care how fast it runs then you could either clock down the M1 core or use Apple's efficiency
> > cores (~1/3 the speed of the big core at ~1/10th of the power) to beat A76 in performance/watt.
>
> Or you could downclock the Cortex too, if we are moving goalposts.
>
> The point is, the Axx worshipping club likes to pretend the gap in absolute performance between
> Apple and ARM H. cores is all about skills and/or magic of the Apple's tean (the dumb "ARM
> is 2 years behind" narrative that certain poster here loves) and nothing else.
>
> But it is in fact ARM's targetting low silicon area and low power consumption,
> because that is what the licensors want while Apple cares not. It might even
> be overwhelming majority of the factors involved in the difference.
Why talk about these things in hypotheticals when nuvia has published actual data on how these cores compare across their voltage/frequency curves?: https://nuviainc.com/blog/historical-perspective-on-the-past-20-years-in-power-management
