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
Ah, you see, but that's exactly the thing. You are entirely correct that there is this notion of "efficiency" being delegated to mobile parts. Which has a historic precedent. In order to consume less power, engineers designed simpler circuitry, with less components and features. Thing about earlier Intel Atom CPUs which had large blocks like out-of-order execution cut out of them in order to make a smaller chips that runs slower but uses much less power. This is also where the entire RISC story comes from, and the popular confusion that "ARM is for slow mobile chips only".
Apple however went the opposite direction. They didn't make their chips efficient by making them simple. They made their chips efficient by making them complex. Their CPUs can process more instructions per single clock than any x86 chip and they can do it by consuming very little power. The result is a mobile phone CPU core running at sub 3Ghz while consuming 5 watts of power that can trade blows (in synthetic benchmarks) with a desktop x86 core running at 4.5+Ghz consuming who knows how much.
What if you have more room to spare than just 5 watts however? What if you can give the CPU 30 watts? 50 watts? 150 watts? You see what I am getting at? Apple claims very confidently that their CPUs scale up. Personally, I doubt that we will see an Apple CPU running at 4Ghz anytime soon. But they most likely can run at 3.5Ghz — which will give it a healthy performance boost over anything in the x86 world while still consuming very little performance in the relative terms.
So let's have a look at a practical example. Intel Xeon W-3275M, which is the largest Mac Pro CPU currently packs 28 cores running nominal 2.5Ghz into a 205Watt TDP package. Well, Apple could pack 40 A13 cores running at 2.5Ghz into the same package. And an A13@2.5Gz core is considerably faster than a Xeon@2.5Ghz core.
Of course, making large chips like that is not trivial at all, and Apple probably still has a lot of work to do before they can deliver it. But this is where their efficiency can be turned into some serious power.
This is one of those things I’m content to just let time provide the answer.
Apple could be providing Just enough information for devs to work on for now, OR they could be providing a very clear direction on their future plans. Either way, once it happens it will have happened!I think Apple may reset expectations on what a “desktop” system should be. Sure you COULD consume 150 watts, but, if the performance is still there, there’s no downside to consuming 50 watts. High Performance desktops currently require big fans and expensive CPU cooling apparatus. What if you could get the same performance with much less cooling effort?
I don't think that the modular MP is going anywhere. They are probably going to sell some sort of CPU boards as MPX modules... but I have no idea what will happen to third party integration. Strictly speaking, it is probably not in Apple's interest to support third-party GPUs in order not to fragment its GPU capabilities. But that is a very dangerous path as well...
Better thing would be to allow IHVs to make Metal drivers and leave that to them completely, at least for the Pro platform.
Better thing would be to allow IHVs to make Metal drivers and leave that to them completely, at least for the Pro platform.
The Mini has been thermal constrained for a while. It is also price constrained. A new Mini with 8 P cores is still 2 cores better than it has now. A 12 core SoC will be cheaper to make than a 18 core one and Apple is probably going to want to increase or lower the Mini price all that much. Apple passing up a opportunity for "thinner and lighter" probably has decent traction in the Mini and iMac space. To go back to some of the Mac Pro 2013 principles and declare "can't innovate my ass" again with a new moniker for the system? They put up a chart saying they were primarily interested in lower power so that would "free up" they ability to "design new systems". That is probably primary where "Apple Silicon" is going. Apple will probably do an "edge case' variant for the Mac Pro (and not necessarily return to the 2013's full constraints ), but it will be anchored on where the base core design being leveraged is trying to go.
The Mac Pro is probably going to use something different. In part, that is why a solution probably isn't coming for another 2 years (and not surprising if doesn't arrive for 2.2-2.5 years). In 2022 starting at 12 cores ( with 8P ) would probably not be competitive. However, the Mac Pro isn't whole desktop line for Apple. It probably isn't even 10% of desktops.
The Mac Pro 2019 is as high a power draw as can get from a normal plug ( in the USA and many other places). So if the discrete GPUs get more greedy with power drawn CPU will need to drop down a bit. There is no more go get if want folks to plug them up most anywhere ( and not only in special provisioned "Machine Rooms" ).
when the Apple Silicon has to drive 12 (or more ) DDR5 DIMMS slots and 72+ PCI-e v4 (v5) lanes , multiple 10GbE feeds , etc. it isn't going to be a double watt chip anymore. The current ARM Neoverse N1 derivaties ( Graviton2 , Ampere , etc.) ) aren't. Apple doesn't have magic pixie dust to do order of magnitude better.
Which is relatively incomplete. Just can't crank core counts. If don't keep those additional cores feed with data then they aren't going to be effective only workloads which aren't skewed to drop as much into the CPU package cache as they can.
But is Apple really going to make something that big and skewed to P cores for such a small number of users? Apple cut the Mac Pro adrift for 6 years ( 2013 - 2016). It isn't a high volume, high priority target market for them.
Eh? The primary R&D funding for these cores is iPhones and iPad. Apple is doing a narrow tweak to cover the Mac also, but there is no evidence that this is a whole 'build from scratch" microarchitecture of processors just for the Mac. It has been laptop targeted for a while. Apple and the fanboy chorus has been arm flapping about "desktop class" A-series but it really isn't if not anchored on Intel's woes and execution problems.
What I'm saying is that not done just because take the laptop TDP limiter off. There is very substantive work to do scaling up the I/O. Dealing with a much higher amount of off-chip interactions.
Apple "artificially" limits themselves all the time because they aren't trying to sell everything to everybody. There is no technical limit as to why there is not mid-range xMac ( to fratricide with the iMac systems ). Apple wasn't broke and couldn't assign a team to get a revised Mac Pro out in 2015-17 timeframe ( No they assigned folks to pay more attention to the iMac Pro instead).
If Apple's 14 core solution appears by end of the year and can't do much with a discrete GPU then there really isn't tons of investment evidenced there with respect to the Mac Pro solution space at all.
Fujitsu is not interested in selling at the price points and volume levels that the Mac Pro / iMac Pro sell at. It is much higher priced solution for folks with generally bigger bank accounts. Ampere is a bit of the opposite. High end Mac Pro volume would be quite bad for them. It is too low. For Ampere to work long term they are going to have to sell at least 1/2 dozen system implementors who will buy in bulk.
Amazon will probably get away with relatively low volume and not a high upfront price because they aren't selling any of the Graviton2's. They are renting them. That is a long term reoccurring revenue stream.
Increase R&D cost. You seem to think that Apple is going to drain the Scrooge Mc Duck money 'billions' pit with as many Mac SoC designs as possible. That probably isn't true. More costs more likely means that Apple will do fewer , not more, designs.
Probably not. The much higher the core count the much fewer systems sold. Which means there is fewer of those SoC to amortize fixed costs on. At some point the processors get expensive enough to drive up system costs where just not going to be as competitive. There highly likely will not be a "ridiculous core count" option at all.
macOS can't handle "ridiculous number of cores. Apple probably isn't going to branch-fork just to chase a subset of a subset of the Mac Pro market. The rest of the Mac , iOS , and iPad OS ecosystem can do with the kernel scheduler they have now. So the Mac Pro will very probably be limited to that. ( Linux and Windows have orders of magnitude larger server market ecosystems which make having a separate scheduler for that make it worthwhile to do. The Mac has the Mini as the primary driver of its "server market". it is a completely different scope and unlikely to change much even with the introduction of the rack model Mac Pro. )
The Mac Pro would far more likely get incrementally more cores (and more up clocked P cores ) than just blindly chasing core counts. If Apple got into a core count pissing match, I wouldn't be surprised if they did that with E cores more so than P ones (and still would be capped under 64 count. 32 if added SMT. ).
As far as SVE2 versus AVX512 ... two issues. One that isn't gong to be almost Intel exclusive by late 2022. Second, there are more than just Apple as ARM workstation/server vendors here. This is quite likely one of those "the iPhone/iOS is so far ahead Android will never catch up contexts". Apple may start off selling more ARM workstations than others but in a couple of years they'll get passed up. Apple being only able to sell to Mac Pro ( and iMac Pro ) users is a boat anchor on what they can do in this space. The other players selling to an order of magnitude more users is battle Apple isn't likely to win. So it is bit dubious that they would do a "bet the farm" on trying to do that. They'll do enough to sell enough, but supreme "domination" in the workstation space? That is probably just fanboy hype. There isn't enough money in it for Apple to do that ( in fact good chance they could loose substantial money trying to do that. )
Apple hasn't sold anything like that in numerous decades. And didn't particularly want to sell it even when they did.
MPX is for supplemental cards not a path to a new "host system".
CPUs aren't coming to MPX modules. A whole new motherboard that still have MPX bays for MPX modules? Yes. but the CPU (and security enclave) are likely solder to the new primary board. And that board is probably not "screw compatible" to the current case. (conceptually it could made to be, but Apple isn't in the retail motherboard selling business any more than are in the retail CPU selling business )
There are more than a few Mac with eGPUs attached to them getting work done. Apple is going to fully consume that market too? Not likely at all. There is extremely little motivation for Apple to drop down to a singluar mono-system for GPU drivers. They don't have to bend over to kiss Nvidia's ring but it makes zero sense for them to kick both iNtel and AMD all the way out when Apple doesn't want to fill all the discrete GPU card needs.
They want fewer discrete GPUs attached, but not zero.
Also have showed zero signs of being interested in non unified (shared) memory GPUs. If don't want to do that work then going to hard pressed to kick out the folks that do want to do it.
the 3rd party GPU markers have to make Metal driver contributions now. Have to want to do it Apple's way and mesh well with their kernel and other drivers but "3rd parties doing work" wouldn't be a change.
Apple is squeezing Intel out of the iGPU space. But the pro platforms never had Intel as the primary GPU vendor.
I agree that all things being equal, lower power consumption is unambiguously a good thing. And the A13 P cores seem to perform similarly to those of the best desktop CPUs, albeit in synthetic benchmarks. So it's not a big stretch to imagine Apple can put 8 or 16 of them in a single chip, and it will be very impressive. I am a little wary about comparing a chip running iOS (which barely supports multitasking), to one running a full desktop OS on much more complex hardware, though. A typical x86 laptop CPU supports 32GB RAM; a workstation might support 1TB. How much does an iPad Pro support? 6GB total? That's less than most graphics cards. Adding bells and whistles to support a proper computing experience will start levelling the playing field a bit.
At the end of the day, though, I don't see much point in speculating on AS performance - we'll know soon enough, when the first laptops start coming out. I admit I am sceptical that any chip can really be that much better than it's contemporaries, but it's true that iPhones spank Android phones for performance (even on the same ISA) and just a few years ago, Intel's x86 chips were stomping AMD's. So it's certainly possible.
Having lived through the PPC era, though, I am a bit worried there will be an initial fanfare, and then things will stagnate - especially at the high end. The Mac has often felt like a lower priority to Apple in recent years, and moving to ARM is likely just as much about rationalising costs as giving us faster Macs. Given Apple's design priorities, we're also more likely to get thinner Macs than faster ones. Whilst laptops and lower end iMacs will have a lot of overlap with iPad Pro processors, so are virtually guaranteed yearly updates, this won't be true at the high end. A double whammy of relatively low sales and the need to significantly rework their mainstream iOS chip, means Apple will be tempted to drag out updates for as long as possible, to recoup costs. Even today, the iMac Pro is approaching 1000 days without an update, and that uses off-the-shelf PC parts.
Meanwhile, it's not inconceivable Intel will regain their competitiveness in manufacturing, prompting strong competition again in the PC space. Mac Pro's could easily wind up sitting around with three+ year old ARM chips (based on past history, I'd be staggered if they don't), whilst AMD and Intel bring out powerful new chips every year.
At the end of the day, though, I don't see much point in speculating on AS performance - we'll know soon enough, when the first laptops start coming out. I admit I am sceptical that any chip can really be that much better than it's contemporaries, but it's true that iPhones spank Android phones for performance (even on the same ISA) and just a few years ago, Intel's x86 chips were stomping AMD's. So it's certainly possible.
Having lived through the PPC era, though, I am a bit worried there will be an initial fanfare, and then things will stagnate - especially at the high end. The Mac has often felt like a lower priority to Apple in recent years, and moving to ARM is likely just as much about rationalising costs as giving us faster Macs. Given Apple's design priorities, we're also more likely to get thinner Macs than faster ones. Whilst laptops and lower end iMacs will have a lot of overlap with iPad Pro processors, so are virtually guaranteed yearly updates, this won't be true at the high end. A double whammy of relatively low sales and the need to significantly rework their mainstream iOS chip, means Apple will be tempted to drag out updates for as long as possible, to recoup costs. Even today, the iMac Pro is approaching 1000 days without an update, and that uses off-the-shelf PC parts.
Meanwhile, it's not inconceivable Intel will regain their competitiveness in manufacturing, prompting strong competition again in the PC space. Mac Pro's could easily wind up sitting around with three+ year old ARM chips (based on past history, I'd be staggered if they don't), whilst AMD and Intel bring out powerful new chips every year.
Last edited:
I don't know how to reply to this without just repeating myself. We seem to be talking at cross purposes. If you can't believe that Apple will invest in faster, higher wattage chips (despite what Apple themselves said at the keynote) then nothing I say will convince you.
You are right we don't know exactly where Apple will take their new designs but this isn't Apple's first major transition. I seriously doubt Apple will go back to something like the 2013 Mac Pro—at least not initially. If you look at the previous two transitions, Apple kept the basic chassis/design of their high-end professional system and just updated the internals to the new architecture (Performa 9600 to PPC 9600; G5 to Mac Pro). After all the drama leading up to the 2019 Mac Pro, and the inherent risks of losing professionals they've been actively courting back, do you really think they would risk upsetting that?
There is also the iMac Pro that will need a Xeon-caliber CPU. So, yes, altogether I think there is a lot of incentive for them to develop tech that competes with Xeon systems like SVE2. And I think they've already got it on the drawing board if not in the fab.
Everything you're saying was said when Apple replaced the G5. They gave themselves two years for that transition and finished in (about) one. I think it maybe take a little longer than a year this time, more because of the state the world, but I'm confident they've already got the technology for a full replacement of their lineup. That isn't fanboyism that's good business sense and history.
Apple's leverage point is that they can optimize more elements in the equation by using a more holistic approach to improving performance per watt. By coding for a specific processor architecture, and making more tasks multi-threaded, Apple has the opportunity to make low power/long battery life devices that have enough muscle to meet most consumer needs.
IMO, the workstation space (desktops and high end laptops), has devolved into a siloed environment where software developers have limited incentive to take on the expense of rewriting from the ground up. Hardware vendors are happy to sell you new gear able to overpower weak code. Improving resource utilization improves the bottom line of whom exactly?
If I were running Apple, I'd look at the A series chips and see the opportunity to push them well beyond their initial market segment by taking on the cost of writing new code, including financial incentives to key software developers if they fully embrace Metal. Since this kind of evolution is hardly an overnight proposition, I think having an Intel based workhorse like the 7,1 available keeps them from being pressured to get an ARM 8,1 out the door ready or not.
Until then, for all the complaints about cost, the 7,1 is a great fit for a segment of the marketplace. It's powerful enough for use cases that would melt a laptop - content creation, etc - but doesn't need to live in a machine room. For tech folk it may not matter, but there are plenty of people who want a nice looking, quiet computer in their homes - not a whining eyesore. I expect more people working from home, rather than commuting to an office, even after COVID-19. Especially since available bandwidth in many residential settings has increased enough that a workplace with a T1 line is no longer as critical.
IMO, the workstation space (desktops and high end laptops), has devolved into a siloed environment where software developers have limited incentive to take on the expense of rewriting from the ground up. Hardware vendors are happy to sell you new gear able to overpower weak code. Improving resource utilization improves the bottom line of whom exactly?
If I were running Apple, I'd look at the A series chips and see the opportunity to push them well beyond their initial market segment by taking on the cost of writing new code, including financial incentives to key software developers if they fully embrace Metal. Since this kind of evolution is hardly an overnight proposition, I think having an Intel based workhorse like the 7,1 available keeps them from being pressured to get an ARM 8,1 out the door ready or not.
Until then, for all the complaints about cost, the 7,1 is a great fit for a segment of the marketplace. It's powerful enough for use cases that would melt a laptop - content creation, etc - but doesn't need to live in a machine room. For tech folk it may not matter, but there are plenty of people who want a nice looking, quiet computer in their homes - not a whining eyesore. I expect more people working from home, rather than commuting to an office, even after COVID-19. Especially since available bandwidth in many residential settings has increased enough that a workplace with a T1 line is no longer as critical.
I think this is excellent analysis. I believe that Apple's goal is to create a new proprietary ecosystem for this kind of workloads and they hope to persuade the developers to invest with promises of new levels of performance. In particular, their ML accelerators seem to be really performant and the way how different components of the system can interact with each other is also fairly unique.
Interestingly though, I would expect them to go after TensorFlow and implement their own Metal and CoreML-driven backend for it. But I didn't see any of the popular ML frameworks on their "open source ARM transition" list...
Feels to me that the Mac Pro is more likely to be canned in a 5 year time frame than replaced by an ARM version in a 2 year time frame. I can’t see how it makes any economic sense whatsoever for Apple to make server/workstation class ARM cpu‘s for the trivial volume that would get used in the Mac Pro.
The decision to revamp the Mac Pro was surely made long before the decision to switch to ARM.
The decision to revamp the Mac Pro was surely made long before the decision to switch to ARM.
I seriously doubt this. Apple made a HUGE mistake with the 2013 Mac Pro. And they said so themselves. The 2019 Mac Pro was created due to the issues/complaints of the horrible 2013 Mac Pro. I do not see them making the same mistake again. Pros want what the Mac Pro provides - PCIe slots, upgradeability, customization, and more. I do not see an Apple Silicon competing with two dedicated Radeon Pro Vega II Duos
Hopefully, Apple will soon shed light on professional sector that they are dead serious and come up with Apple silicone contendor for xeon and nvidia/amd dgpu parts. Their current road map, or whatever they shared so far, isn’t really convincing. SoC will not cut it for professional machine. Apple may have considerable technological advances already, but, if so, that technology can be better served with CPU and dGPU with vram separated.
Had a chance to use the dev kit. I know people are going to say "oh it's not going to be real hardware", but w/e.
-It feels like a Mac, but somethings do feel slower.
-The UI feels slower, but maybe that's Big Sur need to compare with Intel Mac.
-I wasn't logged into the App Store, and it was sitting idle in the background using 40% of CPU doing nothing.
-Strange spikes in Window server usage. Maybe Big Sur?
-Things like Home-brew don't really work well, partially Big Sur issue.
-Had a weird crash / hang. Where the screen just blacked out when the CPU was full load.
-I was able to compile OpenCV pretty easily with a few changes.
-One high note at full CPU load it only pulls 22W from the wall.
-It feels like a Mac, but somethings do feel slower.
-The UI feels slower, but maybe that's Big Sur need to compare with Intel Mac.
-I wasn't logged into the App Store, and it was sitting idle in the background using 40% of CPU doing nothing.
-Strange spikes in Window server usage. Maybe Big Sur?
-Things like Home-brew don't really work well, partially Big Sur issue.
-Had a weird crash / hang. Where the screen just blacked out when the CPU was full load.
-I was able to compile OpenCV pretty easily with a few changes.
-One high note at full CPU load it only pulls 22W from the wall.
I agree with much of this. Apple is primarily the manufacturer of the iPhone, and is essentially porting its legacy desktop platform to use the same chip as its main product. As the iPhone generates huge amounts of money, and Apple ploughs much of it back into R&D, the iPhone's SoC is very fast and efficient. This bodes very well for laptops, which have similar design requirements, but there is almost no synergy with workstations. Workstations don't use integrated graphics, have no particular requirement to save power (within reason), need to be able to support huge amounts of RAM and so on. At best Apple will bring out a new workstation CPU every few years, trying to amortise the development cost over as long a lifespan as possible; at worst they'll just can the whole segment. They have a long history of ambivalence with the Mac Pro.
The only thing I would disagree with is that the switch to ARM has surely been on the cards for quite a long time - just looking at a graph of A-series performance year on year, an eventual intersection with x86 was likely. Moving to one ISA has obvious appeal to Apple, as it would let them put macOS on the same codebase as iOS, as well as generate higher margins due using their own CPUs.
One can assume Apple started work on the 2019 Mac Pro the week before the meeting with the friendly journalists at Apple HQ. They had clearly intended that an iMac variant replace the 2013 MP, but something must have changed their mind. Given that by that point an ARM transition was on the horizon, surely they wouldn't have bothered to do a Rolls Royce job on the 7,1 if it was just going to be discontinued after one generation? In addition, the whole point of the MP was to demonstrate their commitment to pro users. Making an exorbitantly priced pro machine, then announcing an architecture transition 6 months later, then let the machine whither before being discontinued after one generation, would be practically trolling said users.
Last edited:
I understand that the socket for the Intel W series has no future. Does that make an ARM based retrofit impossible, or just difficult? Since an ARM solution would presumably have lower TDP requirements, I can imagine a custom frame that uses the existing screw holes to host it.
I understand that an SoC is designed to avoid engaging sub-systems by doing more on die. What if the MacPro operated like a much smaller machine when the demands were modest - let's call it straight ARM mode - but could readily access other resources as needed?
It's hard to imagine any current or near term SoC that could accommodate many of the specialized requirements currently handled by PCIe cards. Unless or until such a "super" SoC exists, heavy duty use cases will rely on PCIe resources. Because of this, I don't think a well engineered slot box will become useless anytime soon. Moreover, Apple hedged their play by making sure that the power supply, cooling, space, etc could handle a wide swath of potential evolutionary changes.
I understand that an SoC is designed to avoid engaging sub-systems by doing more on die. What if the MacPro operated like a much smaller machine when the demands were modest - let's call it straight ARM mode - but could readily access other resources as needed?
It's hard to imagine any current or near term SoC that could accommodate many of the specialized requirements currently handled by PCIe cards. Unless or until such a "super" SoC exists, heavy duty use cases will rely on PCIe resources. Because of this, I don't think a well engineered slot box will become useless anytime soon. Moreover, Apple hedged their play by making sure that the power supply, cooling, space, etc could handle a wide swath of potential evolutionary changes.
By the end of 2022 Apple will either have a replacement for the 7,1 MP or officially discontinue the line. They’ve given themselves a hard two year deadline to complete the transition and while they have missed deadlines in the past, this transition is a huge deal and Apple will not want it to end with egg on their face. I also think they know that dropping the headless pro workstation will cost them the goodwill among pro users they’ve worked so hard to entice back. So I really, really doubt this is the end of the Pro or the cheese grater design.
[automerge]1593801582[/automerge]
On the flipside, many of Apple’s designs have been constrained by Intel’s lack of sufficient PCIe lanes for things like thunderbolt and expansion. I believe that’s one reason why the 2013 Mac Pro was never updated. Intel never released a single socketed processor that had enough PCIe lanes to support thunderbolt three.
[automerge]1593801582[/automerge]
Apple already uses PCIe tech for their SSD interface on their SOCs so no reason to think they’ll drop PCIe support. Like PCIe in the PPC days, specialized drivers/firmware will prob mean fewer supported cards.
On the flipside, many of Apple’s designs have been constrained by Intel’s lack of sufficient PCIe lanes for things like thunderbolt and expansion. I believe that’s one reason why the 2013 Mac Pro was never updated. Intel never released a single socketed processor that had enough PCIe lanes to support thunderbolt three.
Last edited:
You never owned a G5 Powermac did you? If your computer sounds like a 747 taking off, someone somewhere has failed to achieve sufficient efficiency.
Waiting for the 8,1 ARM. It's unlikely APPLE will provide a an ARM CPU chip that can replace an Intel CPU or a up-dated motherboard. Use what you have chasing the the future is expensive. We have 5 years to save up for 8,1. To date I am still waiting for my cad software to catchup to my existing cMP or iMacPro so if you happen to be running software that can benefit from the power then COOL.
You forget that Apple publicly apologized for the Mac Pro situation in 2016/2017 causing the iMac Pro to be created while the 2019 Mac Pro was being developed. Think about that, they had so many complaints that they could not wait until 2019 for the Mac Pro and they had to release a system in the mean time to satisfy the pros. That alone tells you something about the demand for a Mac Pro system. The 2013 Mac Pro gave Apple a very very bad reputation in the Pro market due to this horrible version. There is zero chance the Mac Pro will be dropped. iMac Pro definitely, but not the Mac Pro. They cannot play with the Pro market again and cause more issues. At that point the entire Mac line would be suffering with Pros getting too fed up with Apple's games. If the 2019 Mac Pro was a once and done system, they would not have spent so long engineering it. It would have just been a slightly modified version of the 2010 Mac Pro than a new cooling, MPX cards, new everything.
And for a company that publicly came out and apologized and said "we are going to make things right again", you don't end the product you created out of vast complaints of the Pro industry. It will essentially be destroying the Mac platform.
While I am looking at getting the 2019 Mac Pro in a few months, if a new 27" iMac is very impressive I might get that instead. If Apple does come out and say they are dropping the Mac Pro line, then I will be done with Apple, even if I do not use the system. Just because of the way they treat their Pro users if that does happen, I will be gone. And I know many others will be too.
2013 Mac Pro was a BIG mistake, and lots of people did leave because of it. They want these people back which is why they created the 2019 Mac Pro to begin with. If Apple does it again? I do not expect Macs to be as popular as they are now.
It's pure speculation, but I don't expect the ARM transition to percolate all the way up to the MP until late 2022. What will be interesting is to see when they can get a shredder laptop shipping. In terms of sales volume and proving out ARM for more demanding tasks, I see that as the inflection point. Until that train is rolling...
The latest Linux drivers have Apple OS and Navi 32 - so you should see at least 1 RDNA 2 GPU.
The $64 question is will an ARM Mac have PCIe slots......
Anything makes economic sense if you charge appropriately for it. It wouldn’t make economic sense for them to produce the current Mac Pro AND the custom made stuff, (including a custom afterburner card)... if they charged $2000 for it.
Any future ARM enabled Mac Pro will just have a similarly appropriate price.
But it didn't fix anything.
Every subsystem in the 7,1 was obsolete on launch day.
Yes, we want PCIe lanes - but not 3.0, we want 4.0 now, and 5.0 in 2021.
Upgradability is a mirage - MPX is the 7,1s version of thunderbolt.
You won't see Apple ARM with Vega, because AMD has already stopped making VEGA chips (You might see RDNA2 however, NAVI 32 appears to be tied to Apple OS).