Awesome, thanks so much!
It seems certain from what you’ve said then then that the MacBook (and Mac mini?) are the machines that Apple are going to target for ARM (if they are going to go ARM on the Mac in some sheep or form).
Anytime, dude. Yeah, I see this. The MacBook could easily benefit from the A-series, but it's gonna need more....damn I hate buzzwords....but ARM is gonna need more teraflops. Teraflops is such an empty word. The Galaxy S9 has more teraflops than the iPhone X, but the iPhone X outperforms it. However, ARM just does not have the floating point of an i5/i7 yet. It is going to need Floating Point (FLOP) to achieve through-put parity.
And it looks like our professional users can sleep more easily at night with the commonalities between Pro Mac and iOS more about sharing the investments in software ie Metal rather than hardware.
I absolutely could see Apple wanting to move to ARM in the MacBook so it can easily share similar components to the iPad ie to put Face ID a 4g/5g modem etc. Ditto the low end iMac.
I get what you’re saying about the iMac - I wonder though if Apple does make an ARM Mac if its professional use will be in creating AVR and VR applications? (let’s assume that apple will be creating processors & GPUs completely optimised for those applications on iOS devices).
I could also see Apple wanting to widen the gulf between its professional and consumer Macs both in performance and in price. ARM at the lower end could be a good way to do this.
It's not about separating them on price or about diverging in power. From a business perspective, I'd want to make sure all of my products are distinct in function. MacBook cannot do MacBook Pro work. MacBook Pro 13 cannot compete with MacBook Pro 15. iMac is inferior to iMac Pro, but superior to Mac mini. I do not want any of my products to bleed into another with function.
Mini is for advanced Home Entertainment or Mac OS Servers. MacBook Air is for old fashioned travelers who need a USB port. MacBook is for travelers who are completely wireless. MacBook Pro 13 is for College Students who need status. MacBook Pro 15 is for getting the job done on the go. iMac is for education or basic work. iMac 4k/5k is for desktop publishing. iMac Pro is for apologizing to customers who have been left behind. Mac Pro should be for "I need to make a AAA Big Budget Hollywood Film and spare no expense.
If I may be greedy and ask you another question(as you know so much about this):
Could it be that ARM’s roadmap in 3-5 years is going to look a lot more impressive and be able to handle more RAM and more general number crunching tasks (more cores?)? I.e. that Apple knows something that we don’t?
Or is it that the ARM architecture is simply not about raw processing power as you say, so it just won’t be able to go in this direction.
Thus will Intel x86 always have the advantage here (as far as the the desktop goes)?
Btw I’ve no skin in the game as to the outcome ie will Apple go ARM or not? I’m just curious!
I don't know a whole lot about ARM, but I understand RISC and the use case for it (worked with IBM PowerPC a lot back in the day). I also knew about PA Semi, the company Apple bought as they built great little processors that can be used in a whole host of great deployments (military especially, but also medical, transportation grid, mobile workstations, etc). PA Semi made these wonderful little chips going back into the 1990s that you could put in a military drone or a heart monitor or those inventory scanners used in BigBox Retail.
3-5 years is an ok assessment for these chips and their power. But you have to also remember their FLOP may not be at parity. Floating Point clock cycles are extremely important for the processing of complex instructions such as those in image signal processing (video, photo), complex interactive chaos (AI, physics simulations, weather, etc.), complex and high quality audio signal processing, spatial awareness processing (AR, VR, etc).
Floating point allows you to control with a more granular hand the sensitivity of your code. Code twenty years ago could make an apple bounce. Code today can make an apple bounce with a certain individual character based on how hard you swipe on screen. This random, chaotic entry by the user can be better calculated and used if it is not an Integer, which is a rigid number with no variation (1 is not 2, etc). By Floating the Point (.) from the Integer to a Rational Number, you can turn a 1/100 scale to the next level. Tiny variations in your flick will reproduce totally different outcomes under a Floating Point scale. Not so much in an Integer Scale, which is just levels of bounciness. What if you want force behind it? Gotta do another scale. Not with Floating point. You can set levels of bounciness in the code based on fractions, the weirder the better.
So, ARM will need to increase its floating point output before it can do AR/VR. But I think Apple has reached a new plateau with iPhone 8/X. Maybe they are almost at parity with the i3? Who knows? But this year we will definitely find out when the new benchmarks come out. Maybe Apple is onto something and AR is their way of showing off just how powerful their A-series now is, as a means of negotiating with Intel.