Can anyone explain from a technical standpoint of why the ARM transition would be a welcome change. The thing people seem to tout is laptop battery life should improve due to less power requirements and getting away from being dependent on Intel. But as the end user outside of possibly needing to plug in a laptop less frequently what benefits are there to us?
Intelās processors are running against walls in terms of speed, power efficiency, heat dissipation. Not to mention they are seriously lagging behind AMD and Apple now in terms of innovation.
ARM processors, especially Appleās, are shaping up to be more efficient, hence more powerful and silent. Also, since Apple wouldnāt depend on Intelās regularly missed targets, we could see more frequent upgrades to the lineups.
If its about Intel falling behind the curve of AMD wouldnāt a move to AMD be better since it already uses x86 architecture? I guess my point is what are the benefits for their computers to be on ARM other than they get control and power users can claim they run on 20 hours of battery?
Or better yet as a desktop user what benefit do I derive from this transition that staying on x86 is holding them back.
More power, a more modern architecture that will evolve better, more silence under load.
Itās far to early to state that definitively. But Apple should be good to go for at least another 6-8 years.
1 Like
For both the low and high end it lets apple set the focus on what the cpu is good at. It also lets apple leverage all of the investment they have put into TSMC. TSMC are well ahead of intel on node size etc. But the big win for apple is being able to control the schedule to often have they been stuck waiting for a new cpu from intel.
We will see a lot of focuse on the low end but there are real strong advantages of the ARM cpu on the high-end as well. Intact it is much easier to make a high core count ARM chip than a high core count x86 chip. Apple could easily produce a 64core (or higher) ARM cpu with basically a copy past operation of there existing destines. And for the very high end users that can use many many many threads at once this will be a killer cpu.
The midrange is where the ARM cpus will have less of a strong advantage, eg the iMac. Currently the iMac has the fastest single core performance and if apple want to avoid people saying it regressed in single core speed they will need to push out a ARM core that can match that single core.
That article does not suggest that we will not see an ARM based macPro within the next 2 years (it is possible that they do a anounce at WWDC 2022 and ship in dec 2022 like all macPro releasesā¦)
It would be less offort for apple but would not help with the issues they have been having with intel. AMD would also have schedule issues. Also apple get TSMCs nodes a good 6 months before AMD get them so by using ARM apple is able to make use of all the funding. I think what apple realy want to be able to do is fine tune the cpus to the exact needs that apple has, they would not be able to do this with AMD.
|
1 Like
Working for a company that owns its own instruction set (S/360 and massive evolution thereof (over 55+ years so far) into current z15 instruction set) I note that IF the switch to ARM happens Apple need never switch again. Thatās got to be worth quite a lot to them - if they are that forward looking.
(My touch point for ARM assembler was Acorn Archimedes about 30 years ago.)
2 Likes
Right. My point is that carefully evolved architectures can go for a VERY long time. Carelessly evolved ones, howeverā¦ 
Glad we agree. But you did say ānever.ā
1 Like
Just rewatched Steve Jobs announcing the Intel transition in 2005. He mentioned the OS X transition as being the second big one in Appleās history. One comment he made about this perked my ears: it was the OS to lay foundation to Appleās ecosystem for the 20 years to come.
I know this was by no means a figure set in stone, but it made me wonder what could happen in 2025ā¦ 
1 Like
Ha. I think weāre going to see a platform of new/next-gen technologies tied to all new hardware: WiFi_6, ThunderBolt_4, and on ARM Macs neural cores and machine learning a la current iPad/iPhone. I also think ARM Macs will be the only Macs to which Apple finally, natively adds cellular. Depending on pricing (thereās flexibility without an āIntel Taxā) and you can see the beginning of a compelling proposition for mass-market notebooks that have more power efficiency, significantly better battery life, more power, more functionality, more developers (assuming tools in place to migrate iOS apps).
4 Likes
Iām (supposedly) in Marketing so Iām trained to use hyperbolic such as āneverā.
1 Like
As you mentioned the āIntel Taxā a caution on pricing: Processors are getting increasingly expensive to produce, particularly with the latest manufacturing processes.
Good thing Apple has uses class-leading tech for smaller SoCs which reduce cost, combined with competitive scaling for its iOS devices. The A12X is manufactured by TSMC using a 7 nm FinFET process, the first to ship in a tablet. Upcoming chips appear to remain class-leading in power/performance, too.
And TSMC is only one of two companies working toward a 2 nm process. VentureBeat just published an interesting article on Appleās ARM roadmap:
2 Likes
To nitpick, I wonder if A10 was really 16nm. 14nm seems more likely.
Gus Mueller, decades-long Apple developer (Acorn, among other apps over the years) authored this āeducated guessā blog post about what he expects to see in the ARM transition. Itās apparently making the rounds - I just noticed it was linked to at SixColors and 512Pixels.
Given the recent refreshes of MBP (e.g. the 16), I think an ARM CPU might show up in a new Air or Macbook 12, testing the switch on lower-end Mac risking problems for power users and app developers. Given the butterfly keyboard fiasco, Apple may be more risk-averse than before in regards to the Mac product line.
2 Likes
The chip change may be great hardware-wise at both high-end and low-end, but what about performant software? Even if the dev chain makes software transition relatively easy for the common case, I think high-end chip transition might come later simply to give pro software devs the time to optimise their code.
1 Like
Perhaps. Although increasing the default/maximum core count on something like the Mac Pro would buy a lot of headroom as far as software performance.
I mean, the iPad Pro has 8 cores. Some of these rumors are talking about 12-core chips right out of the gate. And yes, I know those are split between high performance / energy efficient cores - but there are already high-end ARM workstations with 32 cores, with some on the way rumored to be up to 80 cores. So thereās a lot of room to boost the speed of massively parallel tasks.
2 Likes
It seems like it would be hard to make the change-over piecemeal. If you only have one machine migrated over, wouldnāt that likely result in a dearth of software available for the new platform? I suppose if a developer could just re-compile for an ARM target, then it would be a no-brainer. But I donāt think porting software has ever been that easy. So, wouldnāt there be a risk of the ARM platform being ignored by developers as long as there remain intel machines āin circulationā side-by-side with the ARM machine? Maybe Iām looking at this the wrong way?
1 Like