Speculating here, but I would expect Mac Pro to be the last Mac to move to 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 Is it worth it to buy the Mac Pro now (considering ARM is coming soon)?
- Thread starter dapa0s
- Start date
- Sort by reaction score
What's the likelihood that Apple will be able to create server-grade workstation CPUs with their first attempts?
We all have fears.
But in the end, it's all about getting your work done.
So if you fear moving forward and it hinders you to create then maybe its time to give up this field.
Thoughts from an old man that started on an Apple LC III with 4MB of RAM
I think you can be pretty sure that GPU support - if important at all - will drop off as soon as all the bread and butter Macs are switched to the new platform and all the attention is focused there. The CPUs - you can probably beat those with a Ryzen already for most things, in a few years inevitably they will run rings around them and a fraction of the cost.
Which leaves you with a computer that can address a lot of memory as the USP.
So if the intent is to buy a computer that can last as long as these older cheesegraters then you are out of luck. I think these only stayed around due to not all that much going on with CPU performance advancements in the past. Looks like these times are over though. Also no more GPU drivers for Nvidia cards which also helped to keep the computer relevant.
If you need it now for work then it should be a no brainer though. Question is is it really the best choice if you disregard upgradeability.
Which leaves you with a computer that can address a lot of memory as the USP.
So if the intent is to buy a computer that can last as long as these older cheesegraters then you are out of luck. I think these only stayed around due to not all that much going on with CPU performance advancements in the past. Looks like these times are over though. Also no more GPU drivers for Nvidia cards which also helped to keep the computer relevant.
If you need it now for work then it should be a no brainer though. Question is is it really the best choice if you disregard upgradeability.
Unless there are major shifts in the kernel + GPU driver interface then the GPU support should not drop of at all.
if talking about "new" MPX GPU for older Mac Pros, then Apple has officially left previous Mac Pros out even when doing Intel to Intel transitions. MPX modules that arrive with a post transition , 2nd generation/iteration ARM Mac Pro working with a 2019 Mac Pro. Probably not. I suspect it more so depends upon how much Apple is kick Intel based Mac with eGPUs in the shins or not than on the Mac Pro. However, once no there are no new Intel based Macs with an embedded third party GPU the number of Apple 'backports' to Intel is probably going to be very slim (to none).
The primary motivators for GPU drivers for macOS is inclusion in some standard Mac configuration for sale. The range of GPUs Apple writes drivers for 3rd party GPUs is probably going to get narrower as Apple tries to cover more workloads with their own GPU.
Those more so stayed around longer because Apple was disinterested in the product. ( more resources on other new form factors). If Apple sits and squats on the Apple Silicon Mac Pro replacement that may have side effect of extending out these 2019 Mac Pros. if they immediately nuke the MP 2019 from the new MPX module coverage than not. But if they do one last Intel driver build for those ( to goose out selling some more MPX modules short term) then could be the windows for an extended baseline.
But all of that is looking at "new modules". More than likely the Pro Vega II and Pro Vega II Duo will come way down in the used market prices in 3-4 years. Upgrading performance if start off with a 580X isn't going to be hard in 3-8 years by just moving up in the options that current exist but are temporally priced too high.
Not sure why that is a fear. Apple clearly said there were transitioning the Mac line up over the next two years. The replacement is coming. The only unknown is whether that is 2 , 6 , 12 or 24 months from now. The Mac Pro 2019 wasn't going to see a 6 years transition window the Mac Pro 2013 saw at all.
Five years from two years from now is seven years. it might be five years from when they get dropped. It they aren't going to be dropped soon than right now that means more than five years to de-support status on macOS.
Even of the last DECADES there has been absolutely NOTHING in Apple's policies that the more you pay the longer we will support you. How much you pay for the system has to do with costs that Apple built into the components of the system. the software support costs are bascially fixed ber system. Mac Pro users don't pay much more than MBA users for macOS. Never really have.
It is far so based on replacement cycle time.
"...Owners of iPhone, iPad, iPod, or Mac products may obtain service and parts from Apple or Apple service providers for 5 years after the product is no longer sold ..."
https://support.apple.com/en-us/HT201624
systems that are replaced slower get a longer cycle because the active sales lifecycle is longer. It is how long they are sold. It just happens that Apple has hit the Rip van Winkle 'snooze button' on the Mac Pro far more than most Mac products.
There are lots of folks who rationalize that they are getting more by paying more, but Apple never said that at all. That is just stuff that other folks just made up for themselves. It isn't a contract with Apple.
Unless something is showing up in the console log that the GPU is choking somehow that is about as likely a quirky installed software problem as a "hard driving gaming" problem. Things should just thorottle down if fans can't keep up.
If a software interaction then "cool down" may not matter as much as quit the other apps and reboot.
Nvidia buying ARM is hilarious how? Because they "outsmarted Apple".? Apple would have been way past dumb to pay $40B for ARM. It would have been billions down the drain. Apple doesn't want to be a general system component design vendor. it would certainly drain huge regulatory fire to buy ARM and threaten to nuke the open licensing model the company currently has. Apple doesn't want to be in that business and they need more regulatory scuritiny like another hole in the head.
$40B is a beyond goofy price to pay for ARM based upon how much money they make directly from their business model. How Nvidia is going to make that work to pay off that amount will be interesting. Nvidia has an inflated stock price and irrational exuberance is what is "paying' for most of this deal. Whether than pans out long term isn't really clear. if the stock stays on a growth path it will get covered up without too much fuss. If Nvidia runs into a hiccup over the next 1-3 years it will be a problem. ( some aspects of their path is out of their control).
Apple has architectural licenses. The Nvidia deal probably won't close so if they don't already have an "ARM v9 " license, they have months of room to pick one up for a sliver fraction of $40B. That is highly likely all Apple needs for next 5-10 years of roadmap. There is no "Nvidia has them in a trap now" here at all. Yes, Apple will have to pay some very small per chip produced license fee that is largley a fixed cost wouldn't be much different whoever owned ARM.
(and it will in no way pay off anything even remotely close to $40B at all. )
As far as Nvida and AMD somehow getting upper hand on Apple in that link reference.
" ... Apple already intended to move to 5nm in 2H 2020, freeing up space for companies like AMD. ...
... so AMD will not only be getting the Apple leftovers, ..."
A more than hefty chunk of the "excess capacity" that they are buying is what Apple is leaving behind. Unless getting most of their performance line up on 5nm was somehow a problem for Apple., those two are in Apple's wake. Nvidia is showing up Apple by finally getting onto a process that Apple jumped onto two years ago. Not really. I guess in the "making supersize dies" business yeah it is a problem. But it won't significantly hurt Apple in the slightest to squat on the Mac Pro 2019 until 2022 when doing very large dies ( or collection of medium sized ones) work better on 5nm. The Mac Pro isn't a strategic product for Apple.
That isn't really true.
It wouldn't be very hard for Apple to do a pared down version of Poseidon outlined here.
Apple doesn't have to shot for more than 32 cores. They don't have to cover Xeon SP just the core of the Xeon W range.
The primary point is that ARM is going to compete with Xeon SP ( and maybe top end of Xeon W) in 1-2 years. Apple being able to also do it in that time frame isn't a big , mysterious leap at all.
The bigger issue for Apple is whether they can economically do something in the top end of the Xeon W space and whether they will be able to iterate at the same space as AMD/Intel. As your post srated out if the Mac Pro keeps the same $5999 starting point Apple will have money to throw at at expensive SoC. It is more a matter of whehter they are motivated to do it or not.
IMHO what is dubious is whether that will come inside of 2021 time frame or not. Late 2022 would be my expectation.
Apple would roll out on very mature (at that point) 5nm to get the yield cost overhead costs down .
This won't be lie the MP5,1 There was a long gap between MP5,1 and MP7,1 that extended out Apple putting the 5,1 on the Vintage list. Apple bent the span rules. Similar with the MP6,1. by end of 2022 the vintage/obsolete countdown clock on the current Mac Pro will have started. If Apple drags there feet until the end of 2022 then this Mac Pro will get something longer. If it is in the first half of 2022 then it will end a bit quicker.
I highly doubt though that used Mac Pro 2019 (7,1) prices are going to sink quite as low as 5,1's did over 4-6 years out and that will limit how long they 'stick around". Apple isn't going to sell anywhere near the same number "new". While many folks will think they should be cheap because old there also wan't be that many of them. and if the ARM (and refactored kernel ) driver support is far better 6 years down the road than on the abandoned Intel side besides those using it in "frozen time bubble" mode . (cue story about some old PPC Power Mac G5 in the corner of a studio that has some frozen in time snapshot of some old software that still records music. And is hugely detached from most mainstream set ups in common use. )
So, what you're saying is that getting it now is not the best option, since it will be obsolete soon enough?
If you need for making music right now, just buy it.
Also, Apple Silicon Macs are relevant for portable for now, since Apple targeting mostly laptop first for battery efficiency, then consumer desktop such as iMac/Mini.
I guess Pro desktop of Apple Silicon would came long, very very last.
Also, Apple Silicon Macs are relevant for portable for now, since Apple targeting mostly laptop first for battery efficiency, then consumer desktop such as iMac/Mini.
I guess Pro desktop of Apple Silicon would came long, very very last.
I asked accountant today and says Mac Pro 7,1 with XDR between February 2020 to now going to October 2020 already paid for with funds from projects to include Adobe CC year subscription (throwing up now at yearly subscription scam). He too lazy to look up electricity bills. If you are making money and running a business with 7,1 - then not an issue of purchasing now. If it takes 7-years to pay off 7,1 then maybe your business is not proper for the computer or other way round...We all have fears.
But in the end, it's all about getting your work done.
So if you fear moving forward and it hinders you to create then maybe its time to give up this field.
Thoughts from an old man that started on an Apple LC III with 4MB of RAM
Eh, I need something better than my macbook pro 16, and the only options are mac pro or iMac pro (which is very outdated and not modular).
So between apple silicon coming out soon and the imac pro having such an old design, it's kinda difficult to choose. I guess I'll get the mac pro.
The problem is basically that I am very stingy, be it with cars, computers, houses or whatever, but I guess there's no way around it, I'll have to get a mac pro.
All in all, it's kind of a weird time to upgrade an apple desktop computer.
Mac Pro 7,1 seems better option going forward than iMac Pro even considering 8,1 AS model. Most important thing is to wisely choose CPU and GPU combination for your workflow. If you can't wait 2 years, then be satisfied you did best job/choice possible in your situation - it's sign of mature mind.
I have a 7,1 on the way, and will be selling my beastly (but elderly) maxed-out 5,1 when it arrives. However, I'm currently debating whether I should also offload the Cubix PCIe x16 expansion box that I've been using to squeeze so much into my 5,1, or hold on to it.
The 7,1 is brilliant and offers a tonne of expansion. But it's still limited (by accessory power ports) to only two high-powered 3rd party GPUs. If I keep the Cubix, I can add up to an additional four GPUs on top of the internal two, down the line.
If Apple decide to push even their workstation class machines over to ARM, and the 7,1 does end up being the last intel Mac, then keeping the Cubix will at least allow me to add a lot of additional processing power to the machine down the line. And since Metal accelerated computing and multi-GPU support is only becoming more prominent in video post production software, there's a good chance that future software revisions will be able to take advantage of the additional GPUs.
But who knows? Maybe if we don't have to pay a Xeon tax on a future ARM-based Mac Pro, the prices will drop back down to a more broadly accessible level, and upgrading won't be such a financially-ruinous ordeal as the 7,1 is. ?♂️
The 7,1 is brilliant and offers a tonne of expansion. But it's still limited (by accessory power ports) to only two high-powered 3rd party GPUs. If I keep the Cubix, I can add up to an additional four GPUs on top of the internal two, down the line.
If Apple decide to push even their workstation class machines over to ARM, and the 7,1 does end up being the last intel Mac, then keeping the Cubix will at least allow me to add a lot of additional processing power to the machine down the line. And since Metal accelerated computing and multi-GPU support is only becoming more prominent in video post production software, there's a good chance that future software revisions will be able to take advantage of the additional GPUs.
But who knows? Maybe if we don't have to pay a Xeon tax on a future ARM-based Mac Pro, the prices will drop back down to a more broadly accessible level, and upgrading won't be such a financially-ruinous ordeal as the 7,1 is. ?♂️
Do you not do serial processing? If so the Xeon processor may disappoint you.
Realtime performance in a DAW is extremely related to CPU clock frequency, not so much to core count.
The iMac has a much higher clock frequency than any Mac Pro or iMac Pro.
Nah, I need a lot of cores. At least 12, but ideally 16 if I can afford it.
I mean, 16 would be great, but 12 is already excellent and almost certainly enough.
I would love to be able to afford/justify Mac Pro 7.1. I work mainly with Linux and produce music as a hobby on my AMD hackintosh. I came across a series of interesting videos recently https://www.youtube.com/playlist?list=PLi2i2YhL6okTb7-pQbRk9rSWrw0JknLxT
As you are working with audio I thought it maybe of interest.
As you are working with audio I thought it maybe of interest.
I'm guessing you work more with orchestral sample libraries then, without the need to process tracks in groups. For those working with grouped tracks of intensive VSTi's, more cores are almost always diminishing returns and rarely help beyond 8-10.
These videos provide an excellent explanation why, and can help save you thousands of $.
TDLR: Audio doesn't utilise cores like video rendering does, due to how an audio buffer processes in real time. Many musicians have purchased a Xeon processor to be quite disappointed by the relative performance difference.
great videos, thanks!
Yeah, I work with orchestral sample libraries and my mbp 16 struggles a lot.
that is what you are saying. Not me.
If Apple doesn't replace the Mac Pro 2019 with an ARM variant until mid-late 2022 how can the system be obsolete in the very near term future? No replacement means "obsolete" isn't even a viable adjective. Over a year from now is not "soon" either. Even when there is a replacement in the Mac line up "obsolete" is a extremely dubious adjuective also. Apple doesn't immediately classify them as obsolete so why should users?
Even when Apple does replace it that is only the start of the countdown clock that runs several year. Five years if looking at hardware support is not "soon". Even if Apple cut off Intel macOS in four years after the transition that is not "soon" either.
For anyone whose software stack is slowly moving on updates (and upgrades) then being on Intel code means that the system can get more work done than a system were the software doesn't exist. LogicX Pro will move quickly on the Apple Silicon. Hand crafted , highly optimized vector (DSP) assembly code probably won't. So whether the system goes "obsolete" or not from the user perspective depends upon the software mix also.
If have real business system workloads now/soon then a different alternative is not coming soon ( by any sensible definition of 'soon') for the Mac Pro. There is a slim chance might get either an Intel iMac Pro (to help "kick the can" into 2022) or 24" Apple Silicon iMac ( good chance will hit the 12 core count metric when toss in the "energy efficient" cores ) in next 6-8 months. But if have future needs for triple digit GB RAM and/or PCI-e cards ( more DSPs , larger faster SSD storage at reasonable, non-Apple prices, etc. ) then the current Mac Pro is the only "soon" option.
Apple building a "big die" on 5nm any time soon is pretty unlikely. Apple putting more effort into the Mac Pro than the rest of the Mac line up (especially the laptop side) is also highly unlikely any time soon. The Mac Pro is most likely going to be last to transition. Apple gave themselves two years to take that last step. Over the last 10 years every time Apple put a vague extended window of time for delivering on a new Mac Pro ..... they took all of it. Little indications that they are shifting in strategy now.
If the Mac Pro is really too high for your budget then don't buy it. Some hand waving rational that you are paying thousands more to "future proof" the system is on extremely thin ice. Especially, if you have a requirement that you have to stay highly concurrent with the latest , greatest LogicX Pro software release for 5-7 years into the distant future.
If have demands now and the revenue from work will cover the Mac Pro, then there isn't much else in the short term window likely.
As soon as they support (and big IF) ray tracing compute units on a MPX card I plan on transitioning to this current MacPro. It'll be the last MacPro to be able to use native x86 Windows under bootcamp which I occasionally need for 3D work.
If they come out with a Big Navi generation of MPX cards that'll be good enough for me and my work. Otherwise if the hardware is competitive for your work and it's in your price range I'd say go for it.
If they come out with a Big Navi generation of MPX cards that'll be good enough for me and my work. Otherwise if the hardware is competitive for your work and it's in your price range I'd say go for it.
That really isn't much of an 'if'. Apple gave a timeline for doing the Mac transition. The only way the Mac Pro gets "left off" is that the Mac Pro dies. Depending upon how Mac Pro sales go from early 2020 - late 2021 , that may happen. If Apple is way under sales projections they could kill off the Mac Pro. ( and probably salvage work sunk into the large SoC with variations on the iMac Pro).
If the Mac Pro ( "box with slots" ) is dead then the Cubix is still the only window to dramatically cranking up TFLOPS. But the future GPU options will probably get even more narrower.
But it won't be any GPU.
From 2012-2017 there really wasn't a "Xeon" Tax for the Xeon E5's that the Mac Pro used (or could have used). What occurred with the Xeon W was far more a lack of competition tax . The "$3K for > 1TB RAM capacity" tax was at first because AMD wasn't keeping up and later because Intel doing a money grab because they couldn't scale in core count to the same level.
That "non competition" tax is probably going away on the x86 side in late 2021 - end of 2022 also.
However, the Apple Silicon Mac Pro SoC package probably won't be super cheap. The Mac Pro costs are not based on the CPU. The 8 core that Apple is using lists for around $700. That CPU cost isn't driving $6K. That's an order of magnitude gap. If the CPU dropped to $500 the overall system price would be relatively still the same. Apple doesn't sell millions of Mac Pro's. They probably don't even sell 100K Mac Pro (per year). Which is only going to make the Mac Pro SoC a relatively very low volume die. And likely a large one. That combination of big die , very leading edge process , and very low volume typically leads to higher package cost for end users. There is little volume to amortize the high fixed costs over which means each die has to "pay for" substantively more stuff.
The more features the Mac Pro (and iMac Pro) have that the rest of the mac SoC doesn't have is going to drive costs higher. ~60 PCi-e v4 lanes probably isn't trickling down to the rest of the line up. 400-1,500 GB RAM capacities aren't trickling down to the rest of them either.
The systemic "low volume" tax on the Mac Pro likely isn't going away with an instruction set change. Still going to be a low volume system. And Apple is OK with that as long as it isn't "too low".
Mac Apple Silicon is primarily about bigger Apple leverage on the high margin laptop market. The Mac Pro is "hobby product" along for the ride.
Well, anyway, I bought an 18 core iMac Pro, since it's good enough for my work.
If it doesn't work out well, I'll just be investing in a Mac Pro, and in any case, I'll be using the iMac Pro for about two years, and then switch to ARM. But this way, I also won't be selling my 16-inch Macbook Pro as I've planned, so I'll still have a Macbook on the go.
Also, getting the iMac Pro means I'll be getting a beautiful 5k screen, and it was hard to me to justify buying a pro XDR display, since I absolutely don't need it in any way... but getting a Mac Pro without it just felt sad.
If it doesn't work out well, I'll just be investing in a Mac Pro, and in any case, I'll be using the iMac Pro for about two years, and then switch to ARM. But this way, I also won't be selling my 16-inch Macbook Pro as I've planned, so I'll still have a Macbook on the go.
Also, getting the iMac Pro means I'll be getting a beautiful 5k screen, and it was hard to me to justify buying a pro XDR display, since I absolutely don't need it in any way... but getting a Mac Pro without it just felt sad.
The iMac Pro doesn't drive the XDR Dispaly at its native resolutions. That's one of the specific quirks that Apple could fix with a "speed bump" , incremental Intel powered upgrade. ( nothing wrong with the GPU in the iMac Pro. That specific instance of the Thunderbolt v3 controller is older and has the limitation. The same controller as in the current Macs and that would be cleaned up. )
If there is a Pro XDR display in the set up primarily for audio and part time gaming, then 'stingy' really isn't indicative.