MP 7,1 Is it worth it to buy the Mac Pro now (considering ARM is coming soon)?

[Schismz]

Personally I was thrilled to see the Cheesegrater return because I never thought I'd see it again. 5,1 to 7,1 took literally a decade. I purchased it because I needed it right now. Or to be more accurate I wanted it and the price doesn't really matter that much to me.

If your Mac Pro hasn't more than paid for itself within 2+ years of time, then you don't need a Mac Pro.

Apple's choice to turn it into over-engineered heavy metal sculpture where the entire case is essentially a luxury item probably doesn't help sell more units. On the flipside I really don't think they care.

If this is the last cheesegrater, that's great. I'm happy with my purchase. If they release another cheesegrater- that's wonderful and I'll upgrade. Looking at what products and services make them money, and their track record for the past decade, I'm really not holding my breath; but always receptive to being pleasantly surprised.

 

[jscipione]

Seems doubtful that Apple has some contract that binded them to be required to buying only from Intel . Apple probalby had contract(s) to buy what they wanted , but that is different than "have to buy Intel stuff".

Apple, like other OEMs, gets discounts from Intel for shipping only Intel processors in their computers. Presumably these discounts are ending with the upcoming Arm switch, but up until now Apple has continued to receive them including for the Mac Pro.

CPU development typically takes 3-4 years. So yeah timeline wise some Mac Applie Silicon work would have started about 2017 ( when Apple was doing their "dog ate my homework" sessions and starting on a new Mac Pro requirements. ) The Mac Pro Applle Silicon design work may not have started to a deep stage but the stuff coming out by the end of this year would have been in flight. The Mac Pro SoC would have started before the Mac Pro shipped though. ( Even if stretching out to 2022 there would be some start time in 2019 even if reusing large chunks of the baseline targeted 2020- 1H21 core design. )

Apple’s plan was to replace the trashcan Mac Pro with iMac Pro, but in April 2017 they showed off their yet-to-be-released iMac Pro to some key pro customers. The customers were thoroughly unimpressed and pointed to their cheesegrater Mac Pros. Having already committed to iMac Pro Apple quickly threw together a round table press event and announced the upcoming modular Mac Pro in order to prevent their pro customers from abandoning them when they released the underwhelming iMac Pro. Apple then threw together a team, told Johnny Ive to build them a cheesegrater, and 2 years later we got the current Mac Pro.

Apple gave themselves a 2 year window to transition the Macs over. They didn't explicitly bind to a Mac Pro product. ( If they think it doesn't work in the line up anymore they can drop it. Again not explicitly contractually binding themselves for specific future products. ) . If starts the transitions in Dec 2020 and gets to the Mac Pro in October 2022 then that won't be anywhere close to "as soon as" Apple Silicon is released at all . macOS 11.1 or 11.2 would be the second (or third) generation on ARM.

If Apple as trying to do the whole ecosystem before June-August 2021 then yeah... that is quite likely very problematical. Stretched over two years should be substantively less of a problem quality and stability wise.

I wouldn’t expect Apple to release a Mac Pro update until at least 2022, by then we will know more about what to expect from Apple’s ARM Mac processors.

 

[OkiRun]

I suspect that Apple won’t update the Mac Pro except possibly the GPU until at least 2022

Will the GPUs need to be rebuilt/restructured from the 7,1 to the 8,1? I don't see the need to offer a different GPU for the 7,1 - what's in it for Apple moving forward to AS...? Other GPUs fit within the 7,1 not offered by Apple.

 

[jscipione]

Apple might update the GPU offered in the base model 7,1 between now and 2022, perhaps swapping out the RX 580 for the W5700 or they may offer new GPU options in the future, as they have already with the W5700. An Nvidia option may be possible if Apple and Nvidia can get the Metal API driver support worked out.

There is no reason that Apple couldn’t continue to offer MPX GPU modules for the 8,1, even if it is based on Apple silicon instead of Intel silicon AFAIK.

 

[deconstruct60]

Sure, so going by that logic what they have done now is to show their customers that the left hand in their corp really does not know or care what the right hand does (spending all that time on a new design when the general push is for a new platform and releasing it just before that announcement - wouldn't a simple update to the iMac Pro have served as a stop-gap measure? Instead they made custom GPU modules and an all new design that seems to put expandability at its core.

That is your 'logic'. There is no disconnect from the 'left' and 'right' hand going on here at all. One of ehe current Mac Pro system design objectives is to deouple the CPU package from the overall system constraints. The major problem which Apple essentially explicitly said they were trying to avoid with the new system is tight coupling of the CPU thermal with the other parts of the system (including the GPU). That was one of the primary "painted ourselves into a corner" that they did with the Mac Pro 2013. This new overall system design doesn't say much about the future CPUs other than it can be a relatively large package. If Apple wanted to go with Intel, AMD , Ampere (ARM based ) , or their creations for a workstation SoC they all would work with the current baseline parameters for the system design.

In 2018-2019 there largely was nothing to do a major update on the iMac Pro with. The iMac Pro was more closely a replacement for the Mac Pro 2013. Very similar self-imposed thermal constraints provisioned as a literal desktop system solution ( sits on a desktop taking minimal space ). It was not intended to be a replacement for a subset of the hard core Mac Pro 2008-2012 users who made no moves to the Mac Pro 2013 or to iMacs and MacBook Pros in the interim.

Also completely moot point after Apple sat down in the April 2017 "dog ate my homework" pow-wow and said they were going to do something more high end modular. A "more Pro iMac " was discussed there as part of what they talked about doing in addition to that effort. There was zero logic in Apple ignoring the context that they were in being more than grossly late in getting something else to fill this product space. So they went somewhat conservtative and picked what Intel and AMD would have "done" by the end of 2018 ( or so).

What they have shown me as a Mac Pro customer is that there's no way to rely on any of what they have been doing since 2012. No apparent roadmap, no overall strategy, just flapping around from one concept to another like a teenager with mood swings.

Apple has a strategy. May not want to 'hear' or 'listen' to the strategy , but they are not wandering aimlessly. Macs are focused on laptops. That is mainly it. Frankly, that is highly aligned with where vast majority of Mac users are going. Apple generally following where most folks are going.

The trend of more embedded GPUs and processors is fully enveloped by the Mac Pro 2013. Followed up on by the iMac Pro 2017. There is no flapping around there. It is just not primarily oriented to creating classic boxes with slots.

With them you don't know if there's even still a workstation product line going to stick around a few years from purchase date and if that would be the kind of box you'd expect or some new crazy concept they'll just pull out of thin air.

Apple doesn't promise anybody future products in anything they sell. That predates 2012 by more than a decade.
The more commoditized a product gets, the more likely Apple will just leave that market. Apple isn't trying to be a sell everything to everybody company. Again that isn't "new". If anyone was looking for a system vendor that sold a wide variety of boxes with slots in 2005 , that wasn't Apple either. From 1997 to 2010 Apple sold progressively fewer box with slots models. They aren't flopping around on that trendline much. Over the very long term, it will probably hit zero.

When processors get down past 3nm it is pretty likely the trend is going to shift far more toward being more specialized than general. There isn't a huge amount of room to go down on the general path that hasn't been optimizing by more the several generations of improvements. That won't completely collapse the general workstation market, but it probably will shrink it were folks with more specifically targeted workloads "move down" to smaller systems. Same general trend that "took out" big iron Mainframes from being the 'center of the universe' of computing is going to come for the 80's vintage "box with slots" form factor.

Very different from the laptop and iMac situation.

in part that is because where people are going. If peephole only look around in a niche then not going to get a picture of where most folks are going. Apple is making no promises there but as long as most people are going in that direction then their is robust growth there. ... and Apple will likely follow.

And definitely the opposite of what I can expect from Boxx, HP, Lenovo and
[so on.

If expecting Apple to be HP , Lenovo , or Dell and sell everything to everybody then those expectations are exceedingly dubious foundation. Apple has explicitly said that's not their strategic objective. So expecting them to do what they said they don't want to do is just poor expectation setting. Again this isn't "new", that has been Apple's position for decades.

Apple isn't Boxx either. While not trying to sell "everything" , they also aren't also trying to do relatively low scale either. The Mac Pro is a "hobby product" where they are partially extending the envelope but it is not a strategic objective. Apple will probably 'kick the can' on the Mac Pro 2019 until they have a "copious spare time" window to do another iteration. After which they'll go into another more dormant mode while measuring is there is "still enough" folks buying the system to do another iteration. If there is then they'll do another long cycle iteration. If not then they'll stop. It is just as much about what customers do (as in buy) as it is what Apple plans to do.

 

[deconstruct60]

Apple, like other OEMs, gets discounts from Intel for shipping only Intel processors in their computers. Presumably these discounts are ending with the upcoming Arm switch, but up until now Apple has continued to receive them including for the Mac Pro.

Getting discounts for bulk buys is significantly different than a lock out contract that bars Apple from buying from another vendor. Apple makes lots of major investments not to end up in single vendor lock-in situations long term. As for as "needing" the Intel discounts, Apple sells Macs CPUs for more than normal ark.intel.com 1,000 tray prices. Mac doesn't have "Intel inside" stickers on them for Intel kick-back discounts either. Apple isn't financially "pressed" to need those kick backs to have a profitable system business.

If was/is financially more convenient for Apple to max out their bulk discounts with Intel while AMD didn't have products that were an equally high performant and Apple objective aligned products. AMD has been cheaper for a long while. If Apple was primarily concerned about cheaper they could have moved to AMD a long time ago. AMD being somewhat hostile to Thunderbolt was probably a bigger issue than cost. Intel willing to co-work on projects that Apple cared a lot about mattered. AMD has come around on competitive CPU product delivery and less hostile against Thunderbolt ( now that basically part of USB 4), but they haven't dug themselves out of the hole faster than Intel dug themselves into another one that Apple doesn't like. AMD is also primiarily focused on desktop where Apple sell 75+% laptops.... so unless myopically looking into the iMac 27" BTO , iMac Pro, and/or Mac Pro space.

Apple also generally avoids selling "loss leader" products where many system vendors need Intel's discounts and kick backs to make them closer to break even ( less underwater ).

Apple’s plan was to replace the trashcan Mac Pro with iMac Pro, but in April 2017 they showed off their yet-to-be-released iMac Pro to some key pro customers. The customers were thoroughly unimpressed and pointed to their cheesegrater Mac Pros. Having already committed to iMac Pro Apple quickly threw together a round table press event and announced the upcoming modular Mac Pro in order to prevent their pro customers from abandoning them when they released the underwhelming iMac Pro. Apple then threw together a team, told Johnny Ive to build them a cheesegrater, and 2 years later we got the current Mac Pro.

Highly likely that isn't completely true. The vast majority of the 'die hard' customers that were pointing at their Mac Pro 2009-2012 models had been pointing at it since 2014. It isn't like most of them they had some sudden preference change in 2017 that Apple had never heard of before. Apple new the Mac Pro 2013 had issues in 2014-2015. They largely went about 'solving' them with the iMac Pro as opposed to completely back tracking. But it is highly doubtful that Apple was not getting feedback in that time period from a substantive subset of users that another "box with slots" was also a preferred option. [ For example, there were multiple thread that each spanned , muliplte years years (and thousands of posts) complaining about the Mac Pro 2013. Mulitple times in each thread where folks suggest sending grumblings off to apple feedback. Mulitple sites where the same thing. Additionally, there were multiple threads on multiple sites complaining about failed GPUs . Similarly to the butterfly keyboard thing... it wasn't that Apple didn't 'hear' earlier that they had a problem. There is difference between when they 'hear" and when willing to walk back some design decision that has a large chunk of ego attached it. (e..g, "Can't innovate my a** " , "Courage to kill the audio jack". Apple is in now way oblivious to the range of feedback about those moves. )

One primary reason for the April 2017 "dog ate my homework" meeting was that the 2013 Mac Pro was about to hit the 1100 days mark in age. It was going to be a wide tech press "moan and ground" festival of complaining about how comatose the Mac Pro was a product. It was far more a bad press management event than a "sudden discovery" issue. In April 2017 Apple tweaked the Mac Pro default configurations and sites like Macrumors mercifully reset the "age" date on the system (temporialy). Apple had set a signification 'record' for inaction. Every addition 100 days was a new "even number" record to grumble about. They needed to make some overt excuses. Apple knew they hadn't put in work here. That they suddenly discovered no work is just probalby just excuse making for them on the outside.
( the Mac Mini was also well on track to be yet another desktop product setting records for delay at that point also. It wasn't that the Mac Pro was the only comatose product in the line up. )

Similarly the Mac Pro 2013 didn't try to cover both halves of the Mac Pro 2009-2012 design that. There was a "go large" element to the dual processor system. In terms of RAM capacity and an even higher core count. ( MP 2013 covered the MP 2012 max 12 core count. It didn't really move it up. ).

In one of the interviews with Apple folks after the Mac Pro was released it got revealed that the iMac Pro and Mac Pro have the same product manager. That's is pretty indicative that these products are pragmatically being single tracked. It probably wasn't that Apple had to "scramble" to get the Mac Pro started in 2017. It is at least just as likely that it was the Mac Pro's 'turn' to get started in 2017-2018 after iMac Pro got primary focus in 2015-2016.

Previewing the iMac Pro to those customers was probably more so about identifying "what's left uncovered" than some kind of scrambling . Apple probably had not decided exactly how far more upscale bandwidth wise they were going, but the notion that they were completely blind that there wasn't more is exceedingly dubious. Xeon W in the iMac Pro was leaving loads of bandwidth on the floor. One x16 PCI-e bundle to the GPU , two x4 bundles to Thunderbolt controllers , and on serious overkill x4 link to 1-10GbE controller left 20 PCI-e v3 lanes completely unloaded. The iMac Pro was leaving performance on the "floor" just to fit inside the general 27" iMac enclosure constraints. The fact the iMac Pro was constrained to use only a lightly modified iMac enclosure is also indicative of how little independent , concurrent product development is done in the Mac division.

Apple had time to kick the can on a Mac Pro 2012 replacement because it had a 7 year Vintage/Obsolete count down that started in 2013. The folks who squatted on the model the base 2009 infrastructure for 4 years from 2013 to 2017 could probably squat longer. So the Mac Pro was ordered after the iMac Pro in priority. Especially if Apple enabled some new GPU cards to be added to the system with supported drivers for 3rd party cards. ( which the eGPU effort needed anyway).

I wouldn’t expect Apple to release a Mac Pro update until at least 2022, by then we will know more about what to expect from Apple’s ARM Mac processors.

I don't expect much new clarity before WWDC gives some public insight into macOS 11.1 . If the macOS 11.1 on Apple Silicon version still has "zeroed out" 3rd party GPU support then a new Mac Pro probably is deeper into 2022 than sooner. If the first Apple Silicon Mac lacks a Thunderbolt port ... similarly the Mac Pro is probably further out. ( as there is no easier access "developer and/or tester" test harness to work on enabling that 3rd party GPU support. )

Not only new SoC has to arrive to enable the Mac Pro , Apple is also doing a major disruption of the software drivers for macOS at this point also. The driver disruption impacts the x86-64 Macs also, but there is going to be lots more of those that need fixing for more than a year or so of rolled out Apple Silicon system sales. There are multiple holes to plug in multiple dikes.

 

[deconstruct60]

Apple might update the GPU offered in the base model 7,1 between now and 2022, perhaps swapping out the RX 580 for the W5700

Apple has already dropped the W5500 into the line up as a "low end". And somewhat boldly charging $200 more for the W5500 over the 580X ( for a card with less rock solid drivers and only corner case performance increases. )

Far more likley that Apple will replace the W5700 with another "$800-1000" card move it all the way down to "$400" card range as the default card. Maybe Apple does the $200 drop in price of the W5500 into the base default placement. But discounting $600 in one year. Apple generally doesn't do that. They relatively rarely drop prices at all.

Apple needs a $1000 GPU upgrade replacement more so than the 580X. Furthmore the 580X and W5500 modules are "half width" MPX modules. The W5700X is a "full size" MPX module. It isn't a direct substitute even with a price cut. Even Apple downsided the VRAM and used a mainstream 5700 (instead of 5700XT) as the baseline, they would probably still would have issues putting it into a "half width" MPX module and still have a creditable solution.

By the end of this year Apple is going to be in more "hot water" with the competition at the mid and upper end of the MPX line up; not the bottom.

or they may offer new GPU options in the future, as they have already with the W5700. An Nvidia option may be possible if Apple and Nvidia can get the Metal API driver support worked out.

Nvidia ship has extremely likely already sailed (Nvidia has sunk gobs amounts of money into other avenues of increasing revenue. ) . A "maybe other option" is more likely for an Intel dGPU than Nvidia at this point. Intel at least has a working driver development relationship with Apple at this point (and probably will for the next several years as the bulk of Macs will remain on Intel iGPUs. )

There is apha or beta support for "Navi 2" driver support in macOS 11.0 so there is a pretty decent change will get access to some dervatives of AMD's updates that will start rolling out later this year. There is probably a delay to showing up on macOS in a completed driver state so 2021 timeline is more likely.

There is no reason that Apple couldn’t continue to offer MPX GPU modules for the 8,1, even if it is based on Apple silicon instead of Intel silicon AFAIK.

But that doesn't guarantee that future modules will get support in the older system. It is more likely could move older MPX modules forward into a newer system than the other way. Depends upon what else Apple adds to the future modules and also driver support ( e.g. Thunderbolt controllers also possibly present on the card).

The basic foundation framework that MPX design lays out is solid though. Should be hard to adapt it to PCI-e v4 and/or provisioning Thunderbolt v4 discrete controllers. Apple has to put in the work though on not sure the Apple Silicon support front but on the software/firmware side also.
[automerge]1603311074[/automerge]

 

[ssgbryan]

I wouldn't bet on getting very much (if anything) from RDNA2. Apple has burned their boats, not to mention the fact that the 7,1 userbase is microscopic in size. And then they will have to design the card themselves.

I'll have an RDNA2 card before the Mac Pro community.