Sorry, but Intel didn't do their job properlyI expected talk about Anandtech accepting payola in the mobile forum, not in the CPU forum. :\
A9 is still no competition for Core m (or Mobile Maxwell on the graphics side).
Will never see Core M in a mobile phone. But, what happens when A9 or A9X is put in a notebook with better cooling? At a higher voltage, can it go to 2.5GHz, TSMC in 2014 did a test that indicates that their FinFet+ can be taken that high with the basic ARM reference design much less Apple's optimized designs. And what speed can they clock the GPU with better cooling.
A9X should be very telling.
Well said. Apple is well on its way to moving their entire product stack to ARMv8-A. Apple is waiting for the right time to make the transition. Apple should get Symmetric Multi threading up and running in their custom ARMV8-A core next year. Now that would really crank the multithread performance up. SMT is hard to get done right. Only Intel and IBM have it running now on their CPU cores. AMD Zen and Apple A10/A10X should have it too.
http://www.fool.com/investing/general/2015/10/02/is-there-any-truth-to-this-apple-inc-a10-rumor.aspx
Right now its very clear that Apple is the king of the mobile world. When somebody wants to compare the performance of Core M against A9/A9X they need to look at the performance in the relevant form factors and most importantly sustained performance. Intel chips are good at bursty performance but suck at sustained performance and throttle like mad over long time periods (1+ hr) .
A9X will beat Core M for CPU performance and be competitive for GPU performance when you test with similar form factors. Anybody who is serious about Core M vs A9X needs to look at passively cooled 12-13 inch Skylake tablet vs iPad Pro. Apple SoCs are now the high standard against which any other mobile SoC is measured. Apple has arguably the best mobile SoC in the world and their custom ARMv8-A CPU core leads the competition by a huge margin in single thread performance which btw is still the most important gating factor for a superior smartphone user experience.
Well said. Apple is well on its way to moving their entire product stack to ARMv8-A. Apple is waiting for the right time to make the transition. Apple should get Symmetric Multi threading up and running in their custom ARMV8-A core next year. Now that would really crank the multithread performance up. SMT is hard to get done right. Only Intel and IBM have it running now on their CPU cores. AMD Zen and Apple A10/A10X should have it too.
http://www.fool.com/investing/general/2015/10/02/is-there-any-truth-to-this-apple-inc-a10-rumor.aspx
Right now its very clear that Apple is the king of the mobile world. When somebody wants to compare the performance of Core M against A9/A9X they need to look at the performance in the relevant form factors and most importantly sustained performance. Intel chips are good at bursty performance but suck at sustained performance and throttle like mad over long time periods (1+ hr) .
A9X will beat Core M for CPU performance and be competitive for GPU performance when you test with similar form factors. Anybody who is serious about Core M vs A9X needs to look at passively cooled 12-13 inch Skylake tablet vs iPad Pro. Apple SoCs are now the high standard against which any other mobile SoC is measured. Apple has arguably the best mobile SoC in the world and their custom ARMv8-A CPU core leads the competition by a huge margin in single thread performance which btw is still the most important gating factor for a superior smartphone user experience.
If A9 is "a monster", "enormous", "incredible" and other stuff said in the iPhone "sponsored by Apple" 6s review, then I wonder which adjectives exist to properly describe SKL ().
A9 is still no competition for Core m (or Mobile Maxwell on the graphics side).
Also, my calculation of cache http://forums.anandtech.com/showpost.php?p=37805764&postcount=5390.
They will test SPEC06 on iPad Pro.SPEC 2000 shows more gain than Geekbench. Time to move to SPEC 2006, especially given L2/L3 cache sizes...
What evidence do you have that the iPhone 6s review is "paid for"? I own both phones (which I paid for, of course) and the AT review is spot on.
The iPhone 6s and 6s Plus are the only phones I recommend to anybody in the market for a high end phone. They are that good.
I expected talk about Anandtech accepting payola in the mobile forum, not in the CPU forum. :\
I know that AT isn't sponsered by anyone. That's why it was between quotation marks. But really, the review was very, very positive. I won't blame anyone for thinking such a thing. I don't have the phone so I can't judge, but when you hear a reviewer talking about "perfect (phone)"...
If A9 is "a monster", "enormous", "incredible" and other stuff said in the iPhone "sponsored by Apple" 6s review, then I wonder which adjectives exist to properly describe SKL ().
A9 is still no competition for Core m (or Mobile Maxwell on the graphics side).
Also, my calculation of cache http://forums.anandtech.com/showpost.php?p=37805764&postcount=5390.
the only benchmark that didn't look great was again 3dmark physics...
And that's the only one being computational CPU bound.
What would be great is a benchmark that abuses every advantage that can be taken from both cpus (short of any half-precision shortcuts) including NEON, AVX, etc, etc. Basically, a true measure of raw power under ideal circumstances for both architectures.
Do you mean SPEC isn't CPU bound?And that's the only one being computational CPU bound.
Do you mean SPEC isn't CPU bound?
Actually, 3DMark is using Bullet physics, which does not use threads:3dmark physics is a joke on iOS devices. It was hard coded to use 2 threads so a8x saw no speed increases despite it having 3 cores That just shows the quality of the code from 3dmark
Actually, 3DMark is using Bullet physics, which does not use threads:
https://github.com/bulletphysics/bullet3/issues/126