Let's hope not! The problem PPC had was it wasn't able to scale like x86 did. Hopefully Apple will be able to continually improve ARM.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MP 7,1 What does Apple's switch to ARM based processors mean for the future of the Mac Pro
- Thread starter Korican100
- Start date
- Sort by reaction score
Honestly Intel haven't really made massive improvements since the release of Sandy Bridge back in 2009.
If it wasn't for AMD I am sure Intel would still be selling Core i7 as quad core and Core i5 with disabled HT at the high-end while charging $700 for it.
Although 2 different architectures, I wonder if it would be somehow possible to have an ARM low power core, and another for high intensity tasks. Probably impossible or at least too inefficient to do that.
Perhaps Apple will stay with Intel for the Mac Pro until they develop a chip that can rival one that Intel would provide. I see them starting out with the lower end Macs, and working up.
Perhaps Apple will stay with Intel for the Mac Pro until they develop a chip that can rival one that Intel would provide. I see them starting out with the lower end Macs, and working up.
This suggests a marketing decision, not a technical one.
[automerge]1592515319[/automerge]
It's possible they already have.
It's entirely possible the 7,1 will be like the quad PowerMac G5s—cut down a bit from a longer life by circumstance.
But the whole point of the Mac Pro is that it's extendable and upgradable. It might have *less* life than it would if they stuck with Intel, but even if Apple transitions their entire line over in 2021 it doesn't obsolete the hardware overnight.
But the whole point of the Mac Pro is that it's extendable and upgradable. It might have *less* life than it would if they stuck with Intel, but even if Apple transitions their entire line over in 2021 it doesn't obsolete the hardware overnight.
Given Apple has never used an i7 or i6 in a Mac Pro ( or iMac Pro ) whether AMD got back on track or not is rather narrowly focused. In the Xeon SP space ( which basically impacts the Xeon W 3000 series ) the Graviton2 represents a class of competitors that was coming even if AMD failed to show up.
The mainstream desktop processors may have had issues competitiveness but the space the Mac Pro is in is coming from multiple directions.
AMD isn't making any massive year over year improvements either. What AMD is closer to what Intel used to do with tick-tock which is scoped and managed innovation at a regular pace. There is no "swing only for the fences" solution. That is primarily where Intel failed. They entangled the design updates with a specific process and then cranked up the complexity of a process shrink move. ( if the current 10nm density+complexity target had been the original plan instead of "a bridge too far' plan for 10nm the delay would have been closer to what happened on 14nm )
Intel fumbling the ball on cellular modems also probably is a major contributing factor too. ( a bit of the same root cause where Intel merged that product onto the 10nm rollout path also for 5G. Having to eat crow and go back to Qualcomm is probably a bigger deal than Mac processor issues.
Possible just not highly probable given Apple's investment and priority weighting for the Mac Pro over the last 10-11 years ( 2009-2020 ). Over that 10 year span they have done two Mac Pros. The 2013 Mac Pro was timed to enitrely skip Intel Xeon E5 v1 generation and didn't appear until v2 ( in 2012 Apple was shipping "not new at all" 2012 Mac Pro while every other major workstation vendor rolled out systems with the substantive changes that Xeon E5 v1 brought).
The Mac Pro then went into another even deeper Rip van Winkle sleep until 2017 where got two years of "dog ate my homework" excuses until finally got a new model very late 2019.
That track record and did a major quantum leap investment over Intel and AMD to jump the Mac Pro into new stratosphere of performance? Quite probably not. Over the same time period of doing nothing on Mac Pro Apple spent 3-4 years trying to make butterfly keyboards work. They pushed a one port wonder MacBook out . Touch bars. etc.
Most likely this move to ARM is going to be targeted at the same set of stuff they have been mainly targeting these last 6-7 years.
Every Bloomberg pund-the-drum article on this has point to laptops as being the major motivator for Apple. That probably means that the Mac Pro is once again in a "we'll get to it in our copious spare time" status. If Apple kicks the can for 1-2 years some 3rd party processor vendor could show up and they don't have to put in any special Mac Pro work. They might do something for the iMac and Mini but leave the other two floating ( after bumping the iMac Pro with a "refresh" update later in 2020) . There is extremely minimal indicators that Apple is itching to getting into a high end desktop processor battle with anyone.
Same exact feeling I have. Yet there’s still those mocking us who bought the 7,1, calling us “stupid” and saying how nobody will want this machine come next year. Sounds like envy more than anything rooted in evidence. If I were a betting man I would risk the house that the people say this do not own a 7,1, nor have they ever even used one.
Yet another article out today.
"...
Instead, Rockwell’s team planned to sell a stationary hub, which in prototype form resembled a small Mac, that would connect to the headset with a wireless signal. In Rockwell’s early version, ....
.... Rockwell pushed back, arguing that a wireless hub would enable performance so superior that it would blow anything else on the market out of the water.
...
... (The technology in the hub didn’t go entirely to waste: Some is being recycled to build the powerful processors Apple plans to announce next week for its Macs, replacing components made by Intel Corp.)
..."
Apple’s Secretive AR and VR Headset Plans Altered by Internal Differences
Mike Rockwell’s team of 1,000 engineers aims to create the company’s biggest hits since the iPhone.
www.bloomberg.com
Mac Processor a deliberate plan to build a processor for higher end desktops .... or primarily a side-effect of other projects investing tons of engineers, personnel, hardware, and money into ? [ And this power hub died for something lighter, less performant, and more mobile. ... Entirely in line with the Captain Ahab lighter, thinner laptop primary focus of Mac line . ]
Possible that Apple took a N1 ( some Neoverse baseline design) and tossed some base clock tweaks on it and use that as a baseline for the Mac Pro so don't have to put tons of time into it. That is one option. Probably could get away with that even though a bit behind leading edge Intel/AMD stuff in workstation space.
Here’s another interesting post.
Also consider, Intel and AMD CPU’s are designed to socket into a wide variety of generic systems, each of slightly different quality and purpose. Apple’s solution will be a processor that‘s tuned to run macOS. So, even if the benchmarks show parity or even a decrease in raw speed, as long as it beats comparable Intel macOS system performing macOS specific tasks, it’s still a winner.
Last edited:
If Apple updates the MPX modules and makes incremental additions to the GPU drivers ( and doesn't disrupt that flow with radical driver interface that causes and many problems as it fixes ) then could keep this "core infrastructure" going for a while. Much of the years was between 2014-2019 was folks groaning about how PCI-e slots were "everything". So they got them. Just have to pay for them.
Mac Pro is probably going to have sales growth problems though. However, there will be probably enough to keep it alive for a year or so while Apple hems and haws about what to do long term.
Apple was going to do a "big bag" transition of all the operating systems over to APFS in a year. Getting that done on a Mac failed when the additional complications of Fusion drives and the much broader scope of storage options to longer to do than they though.
Last WWDC Apple said was going to get some real time ray tracing renders out the door later "after 10.15 release"
https://forums.macrumors.com/thread...-new-mac-pro-news-updates-chat.2212822/page-8
Octane X still sitting there waiting. Perhaps some "dog ate my homework" moment next week at WWDC 2020 , but all too indicative of why don't see Apple retiring the Mac Pro 2019 any time soon. Even if there is some Apple hand waving roadmap about getting it done ... just probably won't happen at the speed Apple outlines.
sounds more like same stuff different day of similar complaints of Mac Pro since was realized as the Power Mac. Can go back to 2007-9 era threads and find a small vocal crowd talking about how Mac Pro didn't have enough slots . Wasn't as good as "insert the blank" vendor. etc etc. etc. If it is not exactly the build they would do then it is all wrong and doom to failure.
More likely it is the historic groups plus a large addition of the folks slighted when Apple cranked up the entry about 100% . There are lots of folks in the $2,500-4,500 system price range who won't do iMacs and are more than mildly annoyed. For much of the MP 2013 lifetime there was a constant drown of folks who wanted that system to fail in hopes that would force Apple to go back and build what they wanted. Apple would "wake up faster" when they discovered they were making no money at it and change course faster. That doesn't work when Apple is making money. The margins on the MP 2019 are large enough , that they probably will even at the much lower volumes than older volumes of systems in this space. ( 45-75K per year would probably work for them. )
Last edited:
There isn't much socket variety in Intel's dedicated workstation offerings. Only doing something for macOS is a dual edged sword once get into the part of the macOS ecosystem with relatively very low volumes. ( Intel services niches but the sell to all the system vendors in a "niche" which collective is a much bigger pond. )
Also, only all the "deep insider" rumors so far indicate that isn't what they are doing. The baseline cores are still being primarily design for iPhone (and possibly iPad OS ) systems. There might be some Mac like I/O subsystem(s) tacked onto a scaled up core count ( app cores and GPU cores ), but there is no indicator that they have forked the core design. At best it will be tuned to Mac workloads that laptops run ( and there is high overlap with the desktops. Folks run FCPX on set in mobile settings. Photoshop in the field, etc. ).
If talking about optimizing the OS and not the apps the kernels of the various Apple "xxxOS" flavors aren't that different. Nothing that would majorily change the core design. Applications and memory I/O scale/footprint might, but the OS iteself? No.
If Apple picked up N1 (so some Neoverse iteration) those too are tuned to something else. That would be a bitter fit to the 16-28 core buying Mac Pro users. Not a optimized fit but better than what the iOS/iPadOS OS+app baseline workload traces are going to present.
Intel macOS system comparable system is where Apple's standard practices will be sweeping it under the run. if they wait until Mac Pro 2019 is two year old and only compare a 2022 ARM Mac Pro to 2019 Mac Pro that will totally miss the contemporary workstations of 2022 and their processors.
It won't be hard for Apple to cherry pick some benchmarks that look good. Intel is spending lots of time talking about ML/AI workloads for the Cooper Lake roll out they are doing now. Stand in under that pariicular streetlight's illumination and they are doing pretty good.
I'm not sure what you mean by this statement... can you explain a little more?
I agree with a lot of what you've said here, especially to parts which refer to their starting with mobile devices first. I might even throw out the idea the Mac Pro will remain x64 for a few years. However the only problem with that is it would mean maintaining two versions of macOS...one of the ARM based systems and one for a small production x64 system.
I am looking forward to next week and hearing what, if anything, Apple will be doing with ARM.
Last edited:
Look at Apple’s marketing pages for the Mac Pro to see how concerned they are about being in alignment with what’s current for PC’s.
For EVERY comparison they provide, it’s against some other older Mac. In this case, a Mac that was released in 2017 (iMac Pro) and a Mac (”trashcan“ Mac Pro) that was most recently refreshed in 2013!If Apple sold only 75,000 Mac Pro’s a year, that would be $450,000,000, may be enough money to stay in the business
Last edited:
The $ sign in front of the 75,000 confused me. I see that he was talking about units and not dollars. Thanks.Look at Apple’s marketing pages for the Mac Pro to see how concerned they are about being in alignment with what’s current for PC’s.
If Apple sold only $75,000 Mac Pro’s a year, that would be $450,000,000, may be enough money to stay in the business
If Apple moves to ARM desktops and real software performance declines all but the most dedicated Apple users will abandon IMac, Mackbook and Mac Pro computers. You are already paying more for a Mac computer than for a Windows version with roughly the same chipset. Comparing an ARM Mac to an Intel/AMD that’s a few years older isn’t going to keep users. Having a single compatible OS across phones/Pads/desktops IS a selling point for both users and developers, but if the desk/laptop devices aren’t as fast and are still more expensive than X86 compatible devices then they can spin comparisons all they want they will still lose a significant part of the smaller market share that they have now.
It’s got to have advantages for users and developers and not just Apple. I don’t pretend to know enough to say flat out that Apple’s ARM will be the same/better/worse than whatever AMD or Intel offers when actually running software, but I think “same” is the lowest level Apple should accept and if they can’t get there, REALLY get there on release then they shouldn’t switch architecture. Or decide to just get out of the desktop/laptop market and concentrate on phones and iPads.
I THINK they know this. Monday we’ll find out.
People miss the point of buying a Mac is to get both the operating system that enables them to do their job and also the support for when things go wrong that it gets fixed. Sure you could build yourself something for less, go through figuring out what parts you want and putting it all together but that's a different experience. When it breaks, you're on the hook for getting it fixed as well. You could also pirate MacOS and put it on the device as well but that's another cost saving: you're not paying for that operating system either. Apple could probably make less of a profit on the device but that's their own balance.
If you have a purpose for buying a Mac Pro, then you've got a purpose for it. If you buy it brand new when it's released then I don't think that's even problematic. When they get a little long in the tooth is where I have a much harder time thinking it's a good idea but that's more down to Apple not shipping even routine updates to the desktop line. I can't help but think they play on the uncertainty to get people to buy when they need to instead of waiting for the next random update.
I don't see Apple coming out of the gate and shipping an ARM powered Mac Pro until late 2021 or even 2022 to give time for the higher end apps that would run on those devices to update. I don't expect all of them to be ready in a year or two but I do wonder if they haven't been doing something with Adobe already to help get them across the line at least. I'd expect the pro apps Apple wants to keep updated by the time a theoretical ARM powered Mac Pro appears.
The one wildcard is that Intel might try to screw Apple some how and Apple decides to prematurely shift their timeline. I'd hope not but Intel is increasingly being put on the back foot by AMD and with Apple shifting to ARM that could actually help normalise it in a way that Microsoft's ARM porting attempts haven't.
In an even crazier world I could see Apple continuing to support Windows on an ARM Mac via Bootcamp because it's still in Microsoft's interests to make that possible (they keep pushing and failing) whilst being a hedge that the device can still run Windows.
I feel for a large number of users performance will be maintained. The iPad Pro in it's space has shown that it can handle tasks that are towards the more demanding end with Linus Tech Tip's making a video on editing a video on the iPad. I think they can push comparable performance on day to day desktop activities and they'll likely pull a rabbit out of the hat to compare render performance in one of their highly optimised apps. I think the third party applications picking up and seeing if they optimise for what ever hardware ends up shipping will be more interesting to see.
In the portable space the advantage to users is potentially even better battery life and I'd expect at least comparable performance. The Mac Mini and iMac are essentially portables without a battery so I'd expect slightly more power draw but a device that performs at least comparably. Apple might pass on some sort of cost saving there but I doubt it.
The Mac Pro line is the more interesting one on more what sized core they ship. I could see on the Mac Pro line some architectural changes actually leading to more overall performance by leveraging multiple cores to handle processing. Instead of competing on single core speed (which Intel owns), focus on having more cost effective cores that can handle just as much work potentially quicker. Focus on shipping a bunch more dedicated graphical processing cores so that multiple video streams can be composited at the same time for example. Leveraging all of that power would likely require third party developers to take some time to come to the party, not dissimilar to the console gaming space, but could make an ARM powered Mac Pro just as competitive as the existing Intel ones. The blueprint for how this works is already in the iPad Pro.
Maybe Apple will surprise everyone and announce they’re buying AMD...
That would be quite the “one more thing...”
That would be quite the “one more thing...”
Getting a custom processor out the door is usually a 3-4 year timeline. Apple wouldn't be "prematurely" shifting much of anything. That said Intel completely blowing their roadmap timelines over the last 3 (or so ) years hasn't helped their relationship with Apple. Execution consistency Intel is starting to turn the ship. It is a giant ship so it turns slowly. There is little upside in Intel purposely "screwing" Apple as they separate. Intel doesn't gain anything there at all. In fact making the transition longer will help them. ( So Tiger Lake with integrated 4 port Thunderbolt will be hard for Apple to pass up for one last MBP 13" update. )
Intel has competition from multiple directions, not just AMD. A couple of ARM implementers ( not just Apple) are coming in from the low power consumption, "always one with cellular modem" direction. [ There should be 3-4 Qualcomm 8cx laptops out in retail by this Fall well before Apple has anything out commercially. ] There is Qualcomm and MediaTek in the Chromebook space ( which Intel is trying to hold onto. ) And there are three ARM server vendors ( Ampere , Marvell Thunder X3 , Amazon Graviton2 ) all pushing in from the higher end server space.
What Apple is doing is the least of Intel's problems. Other folks are already shipping in 2020... Apple really won't be.
This is exactly backwards. Microsoft doesn't "chase" OEM system implementers. OEM have to submit drivers and software to Microsoft to be certified and signed. Booting Windows 10 ARM on is something that Apple has to put substantive work into. ( Yes, Microsoft has to do the ARM build of Windows, but that would run in a VM and/or other validated OEM ARM systems.). As noted above there will be multiple folks shipping ARM based Windows system this year. Apple probably isn't going to be the dominate numbers of ARM laptops shipped.
Windows on ARM boots differently than regular Windows. The options to "deploy" , netboot , and UEFI configure an ARM system with Windows 10 is different. Can see the differences outlined for the Surface X specifically here:
https://docs.microsoft.com/en-us/surface/surface-pro-arm-app-management
Enhanced security, UEFI booting is the baseline norm for ARM Windows devices. Pragmatically that means there are more "hoops" for OEM vendors to go though to get a Windows certified for their systems.
[automerge]1592662296[/automerge]
That would actually be worse than just giving away that amount of money to people on the street.
Apple in no way shape or form generates enough business to consume all of AMD's products .
And Apple doesn't want to be a component parts supplier for other OEM's. ( which is AMD's primary business). Without all the contributions for R&D funding from the vast majority of AMD's customers, AMD can't do what they do.
Nevermind there is a problem with actually holding onto the AMD x86 license if they are acquired.
If Apple wants to hire AMD as a component contractor they simply just can go ahead and do that. The new Radeon Pro 5600M with custom HBMv2 memory controller is illustrative of exactly that. Apple didn't need to buy AMD to do that. Just put a large stack of money on the table and sign a contract for a large order(s).
A more possible surprise might be to bring AMD's CEO into the video stage and announce Apple is handing over desktop line to AMD processors and that dumping Intel over next 12-18 months didn't mean dumping x86-64 completely and quickly.
More likely still though is that Apple will have lots of hand waving as to what is going to happen with the desktop space ( especially the Mac Pro part of it). Much like the 2017 "dog ate my homework" meeting where Apple says something along the lines of " Yeah, were really should work on something in this space that is competitive. We'll start working on that now. It won't happen soon. We'll get back to you when we finally finish. "
Last edited:
I remember similar hand wringing when Apple transitioned to Intel. The G5 was never going to fit in a laptop, and the Power architecture wasn't going where Apple needed it to go, but the G5/970 was very competitive with contemporary Xeons. The G5 was a beast, it was just *very* power hungry. Many professionals hung on to them for years—I would have too but I switched career paths and sold off my G5 video editing rig in 2006.
I wasn't impressed with the first generation Mac Pros but the second iteration was a very solid step up. Ampere and Fujitsu have shown that HPC and professional/performance ARM based processors are very possible—and neither of them have a custom core as fast at single threaded operations as Apple's latest A-series.
I think the Mac Pro will be the last machine (again) to get updated. The hardware will not change significantly aside from and upgrade to PCIe 5.0 or 6.0.
My personal guess, is that Apple (perhaps with AMD's help) will use chiplets for their macARM systems and the Pro desktop variant will be similar to the standard desktop variant (performance vs power saving TMSC process) with several more core chiplets added to the package—for a core count far higher than Intel's offerings. For many, if not most, multi-threaded workflows, I think it will be very competitive with Intel, perhaps AMD in ~2022 when it is released.
I wasn't impressed with the first generation Mac Pros but the second iteration was a very solid step up. Ampere and Fujitsu have shown that HPC and professional/performance ARM based processors are very possible—and neither of them have a custom core as fast at single threaded operations as Apple's latest A-series.
I think the Mac Pro will be the last machine (again) to get updated. The hardware will not change significantly aside from and upgrade to PCIe 5.0 or 6.0.
My personal guess, is that Apple (perhaps with AMD's help) will use chiplets for their macARM systems and the Pro desktop variant will be similar to the standard desktop variant (performance vs power saving TMSC process) with several more core chiplets added to the package—for a core count far higher than Intel's offerings. For many, if not most, multi-threaded workflows, I think it will be very competitive with Intel, perhaps AMD in ~2022 when it is released.
Here is the real question. Will this be the last intel based Mac Pro or will there be another iteration?
If this is the last intel based Mac Pro, it's going to be quite coveted in that lots of people still have Intel needs, virtualization and otherwise.
I think they will need to present an arm with 64cores to be 'competitive' in this high end space. That does exist.
venturebeat.com
Apple might even be able to top that 80core unit, who knows. However, I suspect the first arm will be either a 12" MacBook or the MacBook Air. The reason is the current A11 is a better chip than what's in those laptops, so it will be an easy and big boost in horsepower. Then iMac, and Mac mini, MacBook Pro, and Mac Pro will be last (not sure if the iMac Pro will be continued).
So last time they did the transition it took them about 9 months to update the entire line. So, if they release new machines in January and finish converting the Mac Pro in December of 2021, there is a 50/50 chance they dont bother updating the intel Mac Pro.
However, if they need more time to develop a huge core processor, like 2 years, then we may get one more revision.
If this is the last intel based Mac Pro, it's going to be quite coveted in that lots of people still have Intel needs, virtualization and otherwise.
I think they will need to present an arm with 64cores to be 'competitive' in this high end space. That does exist.
Ampere Altra is the first 80-core ARM-based server processor
Ampere has unveiled the industry's first 80-core ARM-based 64-bit server processor today in a bid to outdo Intel and AMD in datacenter chips.
Apple might even be able to top that 80core unit, who knows. However, I suspect the first arm will be either a 12" MacBook or the MacBook Air. The reason is the current A11 is a better chip than what's in those laptops, so it will be an easy and big boost in horsepower. Then iMac, and Mac mini, MacBook Pro, and Mac Pro will be last (not sure if the iMac Pro will be continued).
So last time they did the transition it took them about 9 months to update the entire line. So, if they release new machines in January and finish converting the Mac Pro in December of 2021, there is a 50/50 chance they dont bother updating the intel Mac Pro.
However, if they need more time to develop a huge core processor, like 2 years, then we may get one more revision.
I think they’ve already got a plan in place to scale up to a ’Pro’ appropriate ARM processor. Something that, A12 for phones and A12X for iPP, could work in both an iMac Pro an Mac Pro. I think 64 cores is likely, but Apple’s will run faster (although perhaps without some of the HPC-oriented instructions). Apple’s processor is unlikely to be multithreaded (at first at least) so it will need at least a slightly higher core count. Of course, Apple’s SOCs have a lot of logic that Intel doesn’t have like an NPU or media accelerator.Here is the real question. Will this be the last intel based Mac Pro or will there be another iteration?
If this is the last intel based Mac Pro, it's going to be quite coveted in that lots of people still have Intel needs, virtualization and otherwise.
I think they will need to present an arm with 64cores to be 'competitive' in this high end space. That does exist.
Ampere Altra is the first 80-core ARM-based server processor
Ampere has unveiled the industry's first 80-core ARM-based 64-bit server processor today in a bid to outdo Intel and AMD in datacenter chips.
Apple might even be able to top that 80core unit, who knows. However, I suspect the first arm will be either a 12" MacBook or the MacBook Air. The reason is the current A11 is a better chip than what's in those laptops, so it will be an easy and big boost in horsepower. Then iMac, and Mac mini, MacBook Pro, and Mac Pro will be last (not sure if the iMac Pro will be continued).
So last time they did the transition it took them about 9 months to update the entire line. So, if they release new machines in January and finish converting the Mac Pro in December of 2021, there is a 50/50 chance they dont bother updating the intel Mac Pro.
However, if they need more time to develop a huge core processor, like 2 years, then we may get one more revision.
I agree that the last Intel Mac Pro (probably this one) will be very coveted for a number of years. Samething happened with the G5 and high-end Quadra 68K Macintoshes before that. I wouldn’t be surprised if the 2013 Mac Pros retain their value as well for the same reason.
Everyone keeps citing cloud servers/processor designs as examples of how it is "easy" to have a MacPro style replacement, but that's not the correct comparison. The big Intel processors are actually not super ideal for cloud computing environments. There's a reason that the cloud computing vendors have been looking elsewhere. I spent 15+ years in the high technology space dealing with data center "big iron" and the problems being dealt with there are not comparable to what you do on a high-powered workstation.
A large percentage of the tasks those servers are handling are not very compute-intensive at all and are able to be highly parallelized. That's why Google ran huge farms of low-end processors on bare motherboards rather than big Xeon-based servers (or even Opterons--it wasn't about saving money, it was about the right tool for the right job). Massive amounts of money are spent on power and heat dissipation--vastly more than are spent on the actual servers themselves.
That's a perfect environment for ARM-based computers, which is why so many players have built or are building servers based on ARM technology for this space. But the tasks that are done on most workstation-class computers are not quite the same. Many of them don't parallelize as well, or the software isn't anywhere close to being optimized for multiple cores--not even 8 cores, never mind the 28 cores in the top of the line MP, and 60+ cores? No way.
Is it possible Apple could pursue this and make that high-end ARM processor? Probably yes. But the economics of it make no sense at all. This isn't the days of the 68K -> PPC or PPC -> Intel; in both cases Apple was basically forced to take a new direction. In the PPC move they made a bad bet on an architecture that didn't become mainstream, hence the shift to Intel. In this case they don't have a compelling reason to make a move, so the approach they can take doesn't have to be the same.
The obvious reasons and uses for ARM in the Mac world are in the small low-power devices where they have been using them--and ultralight laptops. I think it's an obvious leap for Apple to move to ARM on their laptops, especially ones like the MBA. As to what they do beyond that, we'll find out soon enough, but I don't expect them to do some wholesale migration of all of their computers over to ARM any time soon.
A large percentage of the tasks those servers are handling are not very compute-intensive at all and are able to be highly parallelized. That's why Google ran huge farms of low-end processors on bare motherboards rather than big Xeon-based servers (or even Opterons--it wasn't about saving money, it was about the right tool for the right job). Massive amounts of money are spent on power and heat dissipation--vastly more than are spent on the actual servers themselves.
That's a perfect environment for ARM-based computers, which is why so many players have built or are building servers based on ARM technology for this space. But the tasks that are done on most workstation-class computers are not quite the same. Many of them don't parallelize as well, or the software isn't anywhere close to being optimized for multiple cores--not even 8 cores, never mind the 28 cores in the top of the line MP, and 60+ cores? No way.
Is it possible Apple could pursue this and make that high-end ARM processor? Probably yes. But the economics of it make no sense at all. This isn't the days of the 68K -> PPC or PPC -> Intel; in both cases Apple was basically forced to take a new direction. In the PPC move they made a bad bet on an architecture that didn't become mainstream, hence the shift to Intel. In this case they don't have a compelling reason to make a move, so the approach they can take doesn't have to be the same.
The obvious reasons and uses for ARM in the Mac world are in the small low-power devices where they have been using them--and ultralight laptops. I think it's an obvious leap for Apple to move to ARM on their laptops, especially ones like the MBA. As to what they do beyond that, we'll find out soon enough, but I don't expect them to do some wholesale migration of all of their computers over to ARM any time soon.