Yes and No, every 'pro' is different... and part of being a 'pro' 'modular' computer is the ability to configure it to the job that you want to use it... and to an extent this includes the ability to repurpose the use of the computer to a different task (some computers are bleeding edge and should be updated/upgraded as soon something changes for the better) -- including the ability to repurpose this to another professional task within the company. Take Apples' most common 'pro' usage... the creative.. in this case the video professional. If you are GPU bound and not processor bound as soon as a new generation of GPUs come out - it would be reasonable to be able to take out the GPUs and replace them (time is money). A pro could be a creative (audio/visual), a pro could be scientific (ability to extend tensor cores etc). So it is not necessarily DIY, but it is reconfiguration... no, I don't need to select the enclosure, tinker with specific power supplies, change the processor, but pretty much everything else falls in the space of configurability to the task. The Mac Pro should really be the catch all for all those that don't fall into a given niche... and thus its ability to be shaped to the task should not be inhibited.
I’m a little “worried” Apple won’t replace the cMP with something decent. It seems most are keeping healthfully low expectations and are starting to make adjustments as well.
I’ll continue to use their iPads, iPhones maybe MacBooks (they’re good, just iPads are a better fit for me in mobile), but for work I may have to make the transition to a hackintosh or god forbid windows.
The reason? Well my current 2012 cMP (granted with some heavy Mods) is running 12- cores at 3.46, 128gb ram, a 1080ti, a highpoint 7101 4x- m2 ssd blade NVMe raid card on 16 lanes, usb C, 16 TB internal storage on Ssd, 4K 10 bit broadcast I/O. It’s gone through several iterations before that.
Anything that can do less than that or have a shorter lifespan potential is a serious downgrade and an absolute no go.
If the new MP ends up just a stack of pricey Apple parts intended to replace a series of PCI slots it really seems like a solution looking for a problem.
I’m thinking of the Black Magic EGPU. It’s cool, but it’s custom design compromises what the original concept was designed to accomplish (modularity) at a extreme price premium, just to optimize the platform. I guess it depends on what you want. Flashy and bespoke or something that works “off the shelf” and comes in about 20-30% less expensive that may or may not be “fit and form” perfect.
A tower about half the size of the cMP with PCI 3.0 seems like a much simpler solution. Apple can design and fabricate it without much hassle (we know they don’t like spending time on these “pro” systems anyway.
If they do stick with a similar look in a smaller package, it would be a clear signal to pro’s Apple is listening and knows what they want. Make it available in space grey and we’ll all have orgasms. “Another classic in the making”. Great tag.
They already have enough to scrap from the cMP (ditch the giant optical drives, the giant heat sink, the ridiculously huge power supply, the 3.5” Hd’s and replace them with SSD and M2. The new iMac pro is a great template in terms the internals. There’s not much to work out on that end. (IMO they are wasting Xeons on that machine).
A nice “core” package with a few upgrade options like processor (single or multiple cpu configs) would make for really nice expandable package/ range or options.
For those that need even more expansion, make a really nice expansion chassis available where the user can add a second GPU, more drives, more expansion cards etc. Maybe they can work out a unique interconnect to take advantage of native pci performance. If you don’t need all of that there are plenty of thunderbolt 3 third party options for lighter weight work, albeit you may have to make several purchases and may have to share TB3 bandwidth.
I feel like the expansion chassis and how it connects to the “core” are what would define a real “pro” system more so than most of the parts inside. The user wants to add those and they don’t want to not can they rely on Apple for all of those parts. The more they lock it down the fewer third party options would be available.
I couldn’t imagine buying a whole box when all I want to do in the future is a “usb4” or new 9K broadcast monitor card. I would have thrown out my tower at least 3 times in that scenario.
Also whatever chassis they conceive should work with the IMac Pros an the MBP’s for those that need it, thus expanding the range of “pro-ness” to three and not just one of their most respected lines.
TBH, I think Apple care about the Pro-sumer market way more than the Pro market.
A MacBook PRO that care about thinnest more than preformance? A Mac Pro is not really upgradable?
Yes, they are powerful when they are newly released but after a few years. People will want to upgrade. Maybe the REAL pro with deep pockets can buy a new one every year. But not all pro will do that. Just look at how often Apple updates the Mac Pro make me sad.
Yes and No, every 'pro' is different... and part of being a 'pro' 'modular' computer is the ability to configure it to the job that you want to use it... and to an extent this includes the ability to repurpose the use of the computer to a different task (some computers are bleeding edge and should be updated/upgraded as soon something changes for the better) -- including the ability to repurpose this to another professional task within the company. Take Apples' most common 'pro' usage... the creative.. in this case the video professional. If you are GPU bound and not processor bound as soon as a new generation of GPUs come out - it would be reasonable to be able to take out the GPUs and replace them (time is money). A pro could be a creative (audio/visual), a pro could be scientific (ability to extend tensor cores etc). So it is not necessarily DIY, but it is reconfiguration... no, I don't need to select the enclosure, tinker with specific power supplies, change the processor, but pretty much everything else falls in the space of configurability to the task. The Mac Pro should really be the catch all for all those that don't fall into a given niche... and thus its ability to be shaped to the task should not be inhibited.
Of course it would be reasonable to take out the GPU and replace them, but it would not be reasonable to build the whole architecture on that capability. If, for the sake of immediate performance, the GPU needs to be soldered then let be it, don't reject it. I want the performance right now, don't sacrifice today's performance for the sake of some better performance in some undetermined future. That is the point.
The reason? Well my current 2012 cMP (granted with some heavy Mods) is running 12- cores at 3.46, 128gb ram, a 1080ti, a highpoint 7101 4x- m2 ssd blade NVMe raid card on 16 lanes, usb C, 16 TB internal storage on Ssd, 4K 10 bit broadcast I/O. It’s gone through several iterations before that.
If you don't mind me asking, what kind of performance do you get out of your upgraded Mac Pro?
Different Pro's need different things. While some pro's won't bother doing anything to alter their system, others expect it to be changeable.
The word Modular can be seen as both internal as well as external. The 2013 Mac Pro is mostly a modular system externally (excluding RAM & SSD). Whereas the 2012 and older crates are internally modular. I can also see the hybrid idea of blocks interconnecting as a modular system as well.
The need of a given subgroup of Pro's also needs to be looked at from what has been stated Apple is going to focus the new Mac Pro on image and video editing as well as AR/VR production. Given their tight relationships with Pixar and Disney that makes sense!
Sadly, that leaves a large group of Pro's hanging in the wind like engineering, code & web development and lastly music editing.
So looking through my crystal ball I see a semi-closed architecture similar the the 2013 Mac Pro, where the panels within the system are interchangeable at time of build and not likely afterwards. So you can get different CPU & GPU modular boards, RAM will be still serviceable and boot and application storage (secure enclave) all other storage will be external. The interface between these modules with be PCIe 5.0 as well as the option for InfiniBand interlink between Mac Pro's so they can be ganged into a compute-cluster for still higher computing (small scale super computing) for CGI and animation.
The sign posts are clear! Apple is sticking within the visual fields which is still the most dominate pro space.
Truly, it shouldn't have taken them this long to design and build the next Mac Pro unless they needed to wait for standards as well as the technology to mature.
Will Pro's buy it? Large companies will for sure, small shops might, and the home user likely not. The new Mini is for them.
Now, if Apple would only create a working Pro's MacBook Pro instead of slapping the Pro moniker on what was clearly the next generation MacBook. The Real MacBook Pro would have what the engineering, code & web development and lastly music editing pro's want. Ports (no dongles), MagSafe and performance the i9 could offer if it had the needed cooling the current MacBook Pro's don't have do to the size. As well as a bigger battery option.
Connecting peripherals (a display is not a peripheral. It’s part of the basic definition of what makes a computer). Is not modular.
Back when computers were really modular -- the first computers I used did not have a display - so no, a display is not part of the basic definition of what makes a computer... maybe a workstation, maybe a personal computer -- but not 'a computer'.
Yeah... no.
The display is part of the input/output.
Today (and it’s been this way for a long long time), personal computing is done via GUI EVEN KEYBOARD SHORTCUTS display the result of their irrational on a display.
And if you want to drill drill down to the topic at hand, the Mac Pro is a GUI driven computer.
A display is not an accessory. It’s a required component of the computer, regardless of the fact that it’s not built in.
In fact, responses like yours, which neither refute nor add anything to the discussion is why I qualified my comment with “personal” computing.
Connecting peripherals (a display is not a peripheral. It’s part of the basic definition of what makes a computer). Is not modular.
Back when computers were really modular -- the first computers I used did not have a display - so no, a display is not part of the basic definition of what makes a computer... maybe a workstation, maybe a personal computer -- but not 'a computer'.
Yeah... no.
Today (and it’s been this way for a long long time), personal computing is done via GUI EVEN KEYBOARD SHORTCUTS display the result of their irrational on a display.
And if you want to drill drill down to the topic at hand, the Mac Pro is a GUI driven computer.
A display is not an accessory. It’s a required component of the computer, regardless of the fact that it’s not built in.
Again, You did not say 'personal computer' you said 'computer'.
And guess what one of the use cases that Apple showed that Mac Minis were useful or used in? Yep, you guessed it - headless configuration (sans monitor)... rows and rows of Macs housed to act as a computer/rendor/colo backend. Having a monitor in that configuration would be ... unmanageable. It is still an accessory.
I just bought a new monitor on Sunday (3 hour delivery at 10:20 at night - good service) -- and it was... on the accessories - monitors tab ;o
One thing is clear: the GPU must be put as closer to the CPU as possible, Xeons don't have integrated GPUs. Once you get the GPU soldered on board (or socketed, preferably), it does not matter much whether the second, third, nth... GPUs are attached via TB3, or PCIe direct slot or M.2, M.n... or whatever... The Thermal Core architecture provides a unified cooling solution for both the CPU and the GPUs (even double!) and I don't expect that will change much in the new Mac Pro. If you separate the GPU from the main board just because your expensive titanic GPU card comes with its own cooling motor, then this is a big compromise.
One of the reasons computer makers prefer soldered CPUs/GPUs is that socketed processors have a tendency to develop loose or intermittent connections over time. Big negative when it comes to reliability and data integrity! For that reason sockets are a definite no-no for laptops. Desktops might be considered a little more reasonable. But you risk damaging the sockets when upgrading a processor.
I agree completely with this article except for one point. When the article indicates that eGPUs don't lose a lot and then indicate that eGPU performance is about 80%-85% of the card's natural profile.
I can't imagine anyone that wouldn't balk at having 20% less of something they care about.
20% less salary?
20% value on your home?
20% less value on your life savings?
20% less food given at a restaurant?
20% less product for your money?
20% diminishment is *significant*.
Basically, that means that to get Vega 64 performance externally is:
A. Impossible.
B. You're going to get Vega 56 performance using a Vega 64 plus a 500-600 dollar carriage.
That's outrageous. And that 20% loss of performance *for twice the price* doesn't even factor the crazy markup Apple has been pushing with their newest systems (Read: Mac Mini -- which is marked up 80% already!)
So yeah, I agree that the Mac Pro won't be what we want. I also predict it will be a flop like the *last* Mac Pro. And yes, Apple themselves admitted it was a complete flop in the same interview that they dropped the existence of the 2019 impending Mac Pro. Bad products won't sell well.
Mac OS and build quality are worth a lot. Are they worth a 20% overcharge? Maybe.
Are they worth a 100% overcharge? Only to four people. If they release a staid eGPU-humping system that can't take advantage of multiple Thunderbolt lanes to get at least 90-95% eGPU performance at a reasonably competitive price...expect *another* failure.
Hopefully AI won't blame Apple pros at that point for avoiding this system like the plague when it offers non-pro performance at an exorbitantly broken price.
Well, I understand what you're saying, however for context: my Vega 64 PCI-E card, at 80% of its capacity is faster than the iMac Pro's version of the Vega 64. Also, the enclosure is about $300-$400.
I'd love to see numbers on that supporting that your PCI-E Vega 64 are over 25% faster than the iMac Pro equivalent.
Moreover, since the eGPU tax increases as the card total Tflops rises, that means your card loses efficacy, the faster it gets. A 580 loses about 20%. Slower cards can sneak by with only a 15% tax. But faster cards can lose up to 30% or more. Even if your Vega 64 is somehow faster than the iMac Pro Vega 64 by over 25%, you're still going to be slower in an eGPU carriage than a PCI-E equivalent.
Finally, since Apple stopped researching the possibility of using multiple thunderbolt 3 ports and having 4+ lanes for eGPU work due to the need for an independent sub controller to handle the data stagger...this eGPU tax will *never* go away.
For a Prosumer machine...getting Vega 56 performance is ok, I guess. But for a Pro that needs 1070 Ti-class performance for heavy graphics work, vid editing, VR or the like...that's not going to cut it.
Finally, every decent carriage I've seen has been 500-600 dollars. Especially the quality/reliable ones. I'm not saying you're a liar. But what brand do you use and what are the overall reviews for that carriage? It's possible the costs for a carriage will decline with time. But I doubt that a Mac will ever cost less than 50% over a similar Wintel system as long as this spiderweb option is pursued.
I don't have my benchmarking results handy, but Barefeats has done a lot of work on it. I said it was faster, I didn't say it was 25% faster, it's more like 10.
You said your Vega, in a carriage, which takes between a 20-30% hit in eGPU tax, is faster than an iMac Pro Vega 64. If you take a carriage 64 operating at 70-80%(I assumed the latter) and you say that it's faster than the iMac Pro 64, that would in fact mean that its full, true speed is 25% faster than the iMac Pro version. Hence why I wanted to see those benchmarks. They don't make a lot of sense to me.
While those carriages are 300-400, when adding a Vega 56 (4-500), you end up paying 2080 prices for 1060 speed. That seems ok to you?
I agree completely with this article except for one point. When the article indicates that eGPUs don't lose a lot and then indicate that eGPU performance is about 80%-85% of the card's natural profile.
I can't imagine anyone that wouldn't balk at having 20% less of something they care about.
20% less salary?
20% value on your home?
20% less value on your life savings?
20% less food given at a restaurant?
20% less product for your money?
20% diminishment is *significant*.
Basically, that means that to get Vega 64 performance externally is:
A. Impossible.
B. You're going to get Vega 56 performance using a Vega 64 plus a 500-600 dollar carriage.
That's outrageous. And that 20% loss of performance *for twice the price* doesn't even factor the crazy markup Apple has been pushing with their newest systems (Read: Mac Mini -- which is marked up 80% already!)
So yeah, I agree that the Mac Pro won't be what we want. I also predict it will be a flop like the *last* Mac Pro. And yes, Apple themselves admitted it was a complete flop in the same interview that they dropped the existence of the 2019 impending Mac Pro. Bad products won't sell well.
Mac OS and build quality are worth a lot. Are they worth a 20% overcharge? Maybe.
Are they worth a 100% overcharge? Only to four people. If they release a staid eGPU-humping system that can't take advantage of multiple Thunderbolt lanes to get at least 90-95% eGPU performance at a reasonably competitive price...expect *another* failure.
Hopefully AI won't blame Apple pros at that point for avoiding this system like the plague when it offers non-pro performance at an exorbitantly broken price.
Well, I understand what you're saying, however for context: my Vega 64 PCI-E card, at 80% of its capacity is faster than the iMac Pro's version of the Vega 64. Also, the enclosure is about $300-$400.
I'd love to see numbers on that supporting that your PCI-E Vega 64 are over 25% faster than the iMac Pro equivalent.
Moreover, since the eGPU tax increases as the card total Tflops rises, that means your card loses efficacy, the faster it gets. A 580 loses about 20%. Slower cards can sneak by with only a 15% tax. But faster cards can lose up to 30% or more. Even if your Vega 64 is somehow faster than the iMac Pro Vega 64 by over 25%, you're still going to be slower in an eGPU carriage than a PCI-E equivalent.
Finally, since Apple stopped researching the possibility of using multiple thunderbolt 3 ports and having 4+ lanes for eGPU work due to the need for an independent sub controller to handle the data stagger...this eGPU tax will *never* go away.
For a Prosumer machine...getting Vega 56 performance is ok, I guess. But for a Pro that needs 1070 Ti-class performance for heavy graphics work, vid editing, VR or the like...that's not going to cut it.
Finally, every decent carriage I've seen has been 500-600 dollars. Especially the quality/reliable ones. I'm not saying you're a liar. But what brand do you use and what are the overall reviews for that carriage? It's possible the costs for a carriage will decline with time. But I doubt that a Mac will ever cost less than 50% over a similar Wintel system as long as this spiderweb option is pursued.
I don't have my benchmarking results handy, but Barefeats has done a lot of work on it. I said it was faster, I didn't say it was 25% faster, it's more like 10.
You said your Vega, in a carriage, which takes between a 20-30% hit in eGPU tax, is faster than an iMac Pro Vega 64. If you take a carriage 64 operating at 70-80%(I assumed the latter) and you say that it's faster than the iMac Pro 64, that would in fact mean that its full, true speed is 25% faster than the iMac Pro version. Hence why I wanted to see those benchmarks. They don't make a lot of sense to me.
While those carriages are 300-400, when adding a Vega 56 (4-500), you end up paying 2080 prices for 1060 speed. That seems ok to you?
I hadn't made a comment on cost-efficiency. We made our opinions clear in the article in regards to what we want to see in a new Mac Pro. To reiterate, we want PCI-E, but we're pretty sure we're not going to get it.
Regarding your math here -- that only works out assuming that the Vega 64 in the Mac Pro is at stock clock speed. But, it is not, we presume for thermal reasons.
From Phil's quote it sounds like they are making it modular so that Apple can produce regular updates rather than saying anything about end-user expansion or lack thereof. Because one problem with the 2013 was the long time between updates.
The biggest problem with the Mac Pros for sometime has been the lack of updated. Specifically updates that actually mean something performance wise when the hardware from Intel exists. With Xeon that is about every two years. The current Mac Pro being the perfect example as hardware for an upgrade does exist. Let’s face it real Xeon updates don’t come every year from Intel.
The second issueis never offering a Mac Pro at a reasonable price point to drive sales volume. Without reasonable volume the Nac Pro will never sell well enough for Apple to give a damn. This means a professional level Mac Pro that starts at less than $1500 dollars. Very few engineers, programmers, doctors or other professionals will even bother to consider Apples high end machines that don’t offer what they need. For one thing most users don’t need dual GPU cards. One good GPU card paired with a decent desktop processor would solve a lot of user needs. The dual GPUs as default never really clicks with me and frankly puts the Mac Pro into a special use category. That category simply is not enough business for any body.
I agree completely with this article except for one point. When the article indicates that eGPUs don't lose a lot and then indicate that eGPU performance is about 80%-85% of the card's natural profile.
I can't imagine anyone that wouldn't balk at having 20% less of something they care about.
20% less salary?
20% value on your home?
20% less value on your life savings?
20% less food given at a restaurant?
20% less product for your money?
20% diminishment is *significant*.
Basically, that means that to get Vega 64 performance externally is:
A. Impossible.
B. You're going to get Vega 56 performance using a Vega 64 plus a 500-600 dollar carriage.
That's outrageous. And that 20% loss of performance *for twice the price* doesn't even factor the crazy markup Apple has been pushing with their newest systems (Read: Mac Mini -- which is marked up 80% already!)
So yeah, I agree that the Mac Pro won't be what we want. I also predict it will be a flop like the *last* Mac Pro. And yes, Apple themselves admitted it was a complete flop in the same interview that they dropped the existence of the 2019 impending Mac Pro. Bad products won't sell well.
Mac OS and build quality are worth a lot. Are they worth a 20% overcharge? Maybe.
Are they worth a 100% overcharge? Only to four people. If they release a staid eGPU-humping system that can't take advantage of multiple Thunderbolt lanes to get at least 90-95% eGPU performance at a reasonably competitive price...expect *another* failure.
Hopefully AI won't blame Apple pros at that point for avoiding this system like the plague when it offers non-pro performance at an exorbitantly broken price.
Well, I understand what you're saying, however for context: my Vega 64 PCI-E card, at 80% of its capacity is faster than the iMac Pro's version of the Vega 64. Also, the enclosure is about $300-$400.
I'd love to see numbers on that supporting that your PCI-E Vega 64 are over 25% faster than the iMac Pro equivalent.
Moreover, since the eGPU tax increases as the card total Tflops rises, that means your card loses efficacy, the faster it gets. A 580 loses about 20%. Slower cards can sneak by with only a 15% tax. But faster cards can lose up to 30% or more. Even if your Vega 64 is somehow faster than the iMac Pro Vega 64 by over 25%, you're still going to be slower in an eGPU carriage than a PCI-E equivalent.
Finally, since Apple stopped researching the possibility of using multiple thunderbolt 3 ports and having 4+ lanes for eGPU work due to the need for an independent sub controller to handle the data stagger...this eGPU tax will *never* go away.
For a Prosumer machine...getting Vega 56 performance is ok, I guess. But for a Pro that needs 1070 Ti-class performance for heavy graphics work, vid editing, VR or the like...that's not going to cut it.
Finally, every decent carriage I've seen has been 500-600 dollars. Especially the quality/reliable ones. I'm not saying you're a liar. But what brand do you use and what are the overall reviews for that carriage? It's possible the costs for a carriage will decline with time. But I doubt that a Mac will ever cost less than 50% over a similar Wintel system as long as this spiderweb option is pursued.
I don't have my benchmarking results handy, but Barefeats has done a lot of work on it. I said it was faster, I didn't say it was 25% faster, it's more like 10.
You said your Vega, in a carriage, which takes between a 20-30% hit in eGPU tax, is faster than an iMac Pro Vega 64. If you take a carriage 64 operating at 70-80%(I assumed the latter) and you say that it's faster than the iMac Pro 64, that would in fact mean that its full, true speed is 25% faster than the iMac Pro version. Hence why I wanted to see those benchmarks. They don't make a lot of sense to me.
While those carriages are 300-400, when adding a Vega 56 (4-500), you end up paying 2080 prices for 1060 speed. That seems ok to you?
I hadn't made a comment on cost-efficiency. We made our opinions clear in the article in regards to what we want to see in a new Mac Pro.
Regarding your previous math, that works out -- assuming that the Vega 64 in the Mac Pro is at stock clock speed. But, it is not, we presume for thermal reasons.
That doesn't answer my question at all. Presumably, (actually no presumption is necessary...benches exist), the iMac Pro's Vega 56 puts out close to 1070-1070 Ti Workflow speed. That justifies the cost and the power.
The point isn't what people are worrying about...which is DIY. The point is that with eGPUs, you pay twice as much for substantially less performance than what most pros truly require. Anything above 1070 performance is currently unattainable with a carriage. Period. Once the informal Mac drivers for the 2080 are out, maybe you get 1080 performance (Vega 64), but by paying so much that you put an iMac Pro's hefty cost to shame.
So my question stands. If the Mac Pro is modular in the respect that it allows eGPU spiderwebbed fusterclucks of nerfed performance, does the prospect of paying 2070-80 prices for 1070-80 performance sound worthwhile? Because I can tell you that most people in this forum and on the street don't think so.
The last Mac Pro was a marvel of engineering but shipping them with dual (not even crossfired...dualed) D700s on the high end, which were just "Pro" rebadged 270xes was a joke. The 2013 iMac brought a 780m that outperformed each card and they rarely worked together as effectively as people liked. For graphical tasks, an imac that cost half as much *literally* was almost as good a workstation for many tasks.
And now we seem to have the same situation. The iMac Pro is horrendously priced. However, the cost is justified. It has an expensive CPU. Expensive RAM. An expensive GPU that, for all benchmarked indications seems to run at about 90-95% of the listed speed, pricey storage and of course that 1000 dollar screen. The system may have been marked up heavily but it was comprised of top shelf parts.
The markup in a SuperMini with eGPU Vegas is hardly worth the same price, when you're eating a far larger hit on performance. The last Mac Pro was a failure. It was literally obsolete within six months of its release. No crossfire support. A midrange GPU in a 3-4000+ system and a difficult to expand system with no real Thunderbolt expandability (No TB 3). We'll see how this one works out.
I agree completely with this article except for one point. When the article indicates that eGPUs don't lose a lot and then indicate that eGPU performance is about 80%-85% of the card's natural profile.
I can't imagine anyone that wouldn't balk at having 20% less of something they care about.
20% less salary?
20% value on your home?
20% less value on your life savings?
20% less food given at a restaurant?
20% less product for your money?
20% diminishment is *significant*.
Basically, that means that to get Vega 64 performance externally is:
A. Impossible.
B. You're going to get Vega 56 performance using a Vega 64 plus a 500-600 dollar carriage.
That's outrageous. And that 20% loss of performance *for twice the price* doesn't even factor the crazy markup Apple has been pushing with their newest systems (Read: Mac Mini -- which is marked up 80% already!)
So yeah, I agree that the Mac Pro won't be what we want. I also predict it will be a flop like the *last* Mac Pro. And yes, Apple themselves admitted it was a complete flop in the same interview that they dropped the existence of the 2019 impending Mac Pro. Bad products won't sell well.
Mac OS and build quality are worth a lot. Are they worth a 20% overcharge? Maybe.
Are they worth a 100% overcharge? Only to four people. If they release a staid eGPU-humping system that can't take advantage of multiple Thunderbolt lanes to get at least 90-95% eGPU performance at a reasonably competitive price...expect *another* failure.
Hopefully AI won't blame Apple pros at that point for avoiding this system like the plague when it offers non-pro performance at an exorbitantly broken price.
Well, I understand what you're saying, however for context: my Vega 64 PCI-E card, at 80% of its capacity is faster than the iMac Pro's version of the Vega 64. Also, the enclosure is about $300-$400.
I'd love to see numbers on that supporting that your PCI-E Vega 64 are over 25% faster than the iMac Pro equivalent.
Moreover, since the eGPU tax increases as the card total Tflops rises, that means your card loses efficacy, the faster it gets. A 580 loses about 20%. Slower cards can sneak by with only a 15% tax. But faster cards can lose up to 30% or more. Even if your Vega 64 is somehow faster than the iMac Pro Vega 64 by over 25%, you're still going to be slower in an eGPU carriage than a PCI-E equivalent.
Finally, since Apple stopped researching the possibility of using multiple thunderbolt 3 ports and having 4+ lanes for eGPU work due to the need for an independent sub controller to handle the data stagger...this eGPU tax will *never* go away.
For a Prosumer machine...getting Vega 56 performance is ok, I guess. But for a Pro that needs 1070 Ti-class performance for heavy graphics work, vid editing, VR or the like...that's not going to cut it.
Finally, every decent carriage I've seen has been 500-600 dollars. Especially the quality/reliable ones. I'm not saying you're a liar. But what brand do you use and what are the overall reviews for that carriage? It's possible the costs for a carriage will decline with time. But I doubt that a Mac will ever cost less than 50% over a similar Wintel system as long as this spiderweb option is pursued.
I don't have my benchmarking results handy, but Barefeats has done a lot of work on it. I said it was faster, I didn't say it was 25% faster, it's more like 10.
You said your Vega, in a carriage, which takes between a 20-30% hit in eGPU tax, is faster than an iMac Pro Vega 64. If you take a carriage 64 operating at 70-80%(I assumed the latter) and you say that it's faster than the iMac Pro 64, that would in fact mean that its full, true speed is 25% faster than the iMac Pro version. Hence why I wanted to see those benchmarks. They don't make a lot of sense to me.
While those carriages are 300-400, when adding a Vega 56 (4-500), you end up paying 2080 prices for 1060 speed. That seems ok to you?
I hadn't made a comment on cost-efficiency. We made our opinions clear in the article in regards to what we want to see in a new Mac Pro.
Regarding your previous math, that works out -- assuming that the Vega 64 in the Mac Pro is at stock clock speed. But, it is not, we presume for thermal reasons.
That doesn't answer my question at all. Presumably, (actually no presumption is necessary...benches exist), the iMac Pro's Vega 56 puts out close to 1070-1070 Ti Workflow speed. That justifies the cost and the power.
The point isn't what people are worrying about...which is DIY. The point is that with eGPUs, you pay twice as much for substantially less performance than what most pros truly require. Anything above 1070 performance is currently unattainable with a carriage. Period. Once the informal Mac drivers for the 2080 are out, maybe you get 1080 performance (Vega 64), but by paying so much that you put an iMac Pro's hefty cost to shame.
So my question stands. If the Mac Pro is modular in the respect that it allows eGPU spiderwebbed fusterclucks of nerfed performance, does the prospect of paying 2070-80 prices for 1070-80 performance sound worthwhile? Because I can tell you that most people in this forum and on the street don't think so.
The last Mac Pro was a marvel of engineering but shipping them with dual (not even crossfired...dualed) D700s on the high end, which were just "Pro" rebadged 270xes was a joke. The 2013 iMac brought a 780m that outperformed each card and they rarely worked together as effectively as people liked. For graphical tasks, an imac that cost half as much *literally* was almost as good a workstation for many tasks.
And now we seem to have the same situation. The iMac Pro is horrendously priced. However, the cost is justified. It has an expensive CPU. Expensive RAM. An expensive GPU that, for all benchmarked indications seems to run at about 90-95% of the listed speed, pricey storage and of course that 1000 dollar screen. The system may have been marked up heavily but it was comprised of top shelf parts.
The markup in a SuperMini with eGPU Vegas is hardly worth the same price, when you're eating a far larger hit on performance. The last Mac Pro was a failure. It was literally obsolete within six months of its release. No crossfire support. A midrange GPU in a 3-4000+ system and a difficult to expand system with no real Thunderbolt expandability (No TB 3). We'll see how this one works out.
There's a larger point here, which we alluded to in the piece. It has been six weeks since Mojave was released. There are no informal Nvidia drivers, and comments by Nvidia suggest that they're just waiting on Apple. And, they've never been explicitly supported in an eGPU at all. Does that sound like a commitment to DIY to you?
The last two times, the web drivers were updated within a week of the OS shipping to customers, and there were releases during the beta period after WWDC. There weren't this time.
I am cognizant of all of your points that you've made so far, and agree with most. We talked about many of them in this article, in fact. Is an eGPU cost-effective? It isn't really -- but that is in no way the point of this post, and Apple doesn't care if they are or not.
People are still not listening (though heaven knows this article is trying to get them to listen), and are too blinded by their own ideas of what the Mac Pro should be like. I think William is exactly right about Apple still preferring external connections to a heavy/bulky internal warehouse of (mostly unused) potential. I like the idea of a stack of Mac mini-like modules that are highly-tuned for different pro needs beyond the base unit, meaning the total machine can start fairly low in price (relative to Mac Pros, I mean) and then you can pick other “modules” tuned for what you need them to do — a “Pro” Mac for an audio engineer is quite a different beast than a “Pro” Mac for a 3D renderer.
I have no idea when TB4 will come out, but at that point I think you finally reach the point where those speeds can’t be beat by PCI-E — and that sort of thinking may be why Apple is so reticent to say stuff like that.
One think I know “for sure” though — you’re not going to get a machine that can be upgraded/changed in a wide variety of ways without paying Apple more money for new “modules,” I’d wager. I do think the company is polling and listening to serious Pro users — and I think this forthcoming system will be priced in a way that reflects the sort of person or company who can make back the cost of that investment in a couple of days of specialized work.
Yes and No, every 'pro' is different... and part of being a 'pro' 'modular' computer is the ability to configure it to the job that you want to use it... and to an extent this includes the ability to repurpose the use of the computer to a different task (some computers are bleeding edge and should be updated/upgraded as soon something changes for the better) -- including the ability to repurpose this to another professional task within the company. Take Apples' most common 'pro' usage... the creative.. in this case the video professional. If you are GPU bound and not processor bound as soon as a new generation of GPUs come out - it would be reasonable to be able to take out the GPUs and replace them (time is money). A pro could be a creative (audio/visual), a pro could be scientific (ability to extend tensor cores etc). So it is not necessarily DIY, but it is reconfiguration... no, I don't need to select the enclosure, tinker with specific power supplies, change the processor, but pretty much everything else falls in the space of configurability to the task. The Mac Pro should really be the catch all for all those that don't fall into a given niche... and thus its ability to be shaped to the task should not be inhibited.
This is a perfectly reasonable definition of what YOU believe a "pro" computer should be. If you gather 10, or maybe 10,000, additional computer buyers off the street and ask them to define what a "pro" computer is you will now have up to 10,000 alternative definitions of what a "pro" computer is. Oh, and one more for Apple's definition of what a "pro" computer is. That one's easy = whatever Apple marketing team decides to embellish with the "Pro" suffix. They have absolute authority by virtue of owning the brand name.
Not trying to be facetious because a Venn diagram of all common characteristics of all 10,000 equally relevant definitions may actually collect up a subset of features, capabilities, and characteristics that more than 50% of computer buyers are comfortable with. Hopefully the result of such an exercise would align with Apple's definition, but Apple would not be bound to it. Apple could say "screw you," retire the Pro moniker entirely, and use the suffix "Extreme" instead.
There are some slivers of technology that attempt to associate names or designators with useful meanings. Most of these are standards, e.g., IEEE 802.11ac. You could even say that the term "managed" has relevance in terms of network equipment. Back in the days of TTL chips the 54xx named chips had a clear and significant difference over their 74xx brethren. Later generations of technology have kind of fallen off the wagon and at-best only make a feeble attempt to conform to expectations, e.g., the original expectations around "4G" cellular platforms were abandoned in favor of vendor marketing perverted definitions and non-conforming implementations. In all likelihood "5G" will suffer the same fate. So even in the face of what looks like a "standard definition" with an industry assigned moniker the infinite power of individual company marketing, buzz-speak, spin, and the ease of pulling the wool over the eyes of oblivious consumers beats all reason, conformity, and expectations to a bloody pulp, with nothing left but a meaningless blob of self obfuscating characters strung on to the tapered end of a product name. Whether Apple decides to call something a Mac Pro, Mac Orb, Mac SS, Mac Gasm, Mac Lite, Mac Max, Mac 52, ... or whatever, it's all the same and the name basically means nothing at all to anyone - except for Apple, and they're not going to tell you anything about their naming formula.
This is really tough. The entire article turns on the word "modular" and what does it mean. External "modules" or are PCI cards also modules?
AppleInsider's guess is as good as anyone's. While I hope you are wrong, generally my best way of predicting what Apple will do is, think, what would really really disappoint me, and that will be pretty darn close.
Lack of storage expansion is inaccessible.
Where I take some issue is with the fuzzy word "pro". Apple it's self mentioned that pro's encompass a very wide number of user types. This article basically seems to focus on pro's as pixar like creative professionals.
A broader umbrella for pro includes many other types. Power users. Enthusiasts. Pro's from very different industries like scientific research that requires MASSIVE upgrading, including video cards, storage, ram, for computation.
In essence, the article suggests that the new Mac Pro will in essence be a repeat of the failed trashcan, but with a faster thunderbolt 3 port this time, which is still pathetically slow compared to PCI. Certainly possible. But it's fundamentally why I think this article will prove false in the end. And if not, it's time to sell the stock, because they'll just keep re-making cubes.
But Apple has already lost a lot of pros (in the broader sense of the word). If it doesn't deliver here, it will lose near the last of those old loyal users. Apple can do just fine without them on iPhones alone. But some day, it may need their enthusiasm, and they wont be there. They will be on windows/linux and some on hackintosh for a while longer.
And to the article's point, it's not like the Cheese Grater's adoration isn't know. So ignoring it at this point is basically an invitation to leave. Sadly, it's very possible.
From Phil's quote it sounds like they are making it modular so that Apple can produce regular updates rather than saying anything about end-user expansion or lack thereof. Because one problem with the 2013 was the long time between updates.
I think you have something there. Car makers do this now - VW has the MQB architecture that defines how the underpinnings go together and what size the car will be, but really has little to do with how the car looks. It standardizes engine mountings regardless for what fuel is used for example. Like you say, Apple might be referring to modularity more for their own purposes and not ours.
Well, if that is the case -- Apple is going to get roasted... since they emphasised over and over again with the gang of 5 to report on ... that basically the Mac Pro 2013 was a failure and they were going back to the drawing board to fix the issue with a new Mac Pro... It has to be modular for the customer as well. It however does not have to be your standard DIY form of a solution (and I don't really expect it) -- as long as they solve the same problem without compromise. (a 1080Ti performant GPU had about a 30% hit in performance - basically paying extra for the Thunderbolt solution then cutting the performance of the card to one equivalent to less than half the price. GPUs are going to become more and more powerful, so the impact for the current version of Thunderbolt is not going to decrease.
I generally agree with this but I see the current Mac Pros failure as one of configuration that could have easily been fixed with in months of debuting. Simply put the starting price is too high on a machine configuration most people would never buy. For one the dual GPU solution was a joke from the beginning. Second deleteing that extra GPU could have freed up room for some PCI-E SSD slots. In other words you could have had machine configurations that served a variety of professionals.
Why Apple waited so Long to fix a machine that couldn’t possibly drive volume is beyond me. The required adjustment could have been done by a work study engineer. Effectively you delete a GPU card and put in a M2 carrier board.
Beyondthat at I cant even begin to understand the lack of an Apple branded storage array.
God forbid they ship a Pro model with the features people are asking for.
There is one time to think different, and another time to listen. They thought different on the Pro already and it was not quite the ticket!
... Real pros don’t have time to fiddle-fart around with slots
Do you even understand how stupid you sound? - Or understand how real Pro systems are used?
It's not stupid. You're confusing DIY with pro. I work pro in enterprise, and in enterprise we pay for new machines, we never ever crack open the case and perform DIY mods ourselves.
DIY and tinkering != professional.
To use your earlier quote: "Your use case != everyone's use case"
Where I work it's a regular occurrence to open up computers to repair/upgrade them. Replace failed/failing drives, power supplies, RAM, etc. But then again, we use PCs which make that a much easier, more feasible proposition than it would if we were using sealed up Macs. If you are an independent or small-scale shop (or use computers that are not upgrade/repair friendly) it probably doesn't make sense to have staff with those skills. But to say it doesn't happen anywhere in enterprise is false.
And where I work (a Fortune 100 company) IT does NOT swap out any components on their PCs. They will buy new machines. Sounds like your IT department is looking for ways to stay employeed.
I got the impression from the first time they said "modular" that this is the kind of approach they would go for. A central processing box, with connectivity to external storage, graphics, etc. I was surprised at how many people thought "modular" meant a cheesegrater-like box. The cheesegrater isn't really modular.
I think their way to accommodate the many different "Professional" users out there is to provide for different external units that plug in over TB3. People who aren't doing video or graphics work (and even some who do) will be fine with the internal GPU, for others there are the eGPUs. Similarly if you're working with audio, plug in an external DSP or hardware encoder.
Then, when there are new processors available, you upgrade the central box, and keep all the peripherals, provided TB3 is available. Yes, that is a risk, but any future version is likely to stay backwards-compatible, and the worst outcome is you buy a (GASP!) dongle. I expect, however, the USB-C connector will stay for a while. Maybe as long as the USB-A connector has.
Comments
Yes and No, every 'pro' is different... and part of being a 'pro' 'modular' computer is the ability to configure it to the job that you want to use it... and to an extent this includes the ability to repurpose the use of the computer to a different task (some computers are bleeding edge and should be updated/upgraded as soon something changes for the better) -- including the ability to repurpose this to another professional task within the company. Take Apples' most common 'pro' usage... the creative.. in this case the video professional. If you are GPU bound and not processor bound as soon as a new generation of GPUs come out - it would be reasonable to be able to take out the GPUs and replace them (time is money). A pro could be a creative (audio/visual), a pro could be scientific (ability to extend tensor cores etc). So it is not necessarily DIY, but it is reconfiguration... no, I don't need to select the enclosure, tinker with specific power supplies, change the processor, but pretty much everything else falls in the space of configurability to the task. The Mac Pro should really be the catch all for all those that don't fall into a given niche... and thus its ability to be shaped to the task should not be inhibited.
I’ll continue to use their iPads, iPhones maybe MacBooks (they’re good, just iPads are a better fit for me in mobile), but for work I may have to make the transition to a hackintosh or god forbid windows.
The reason? Well my current 2012 cMP (granted with some heavy Mods) is running 12- cores at 3.46, 128gb ram, a 1080ti, a highpoint 7101 4x- m2 ssd blade NVMe raid card on 16 lanes, usb C, 16 TB internal storage on Ssd, 4K 10 bit broadcast I/O. It’s gone through several iterations before that.
Anything that can do less than that or have a shorter lifespan potential is a serious downgrade and an absolute no go.
If the new MP ends up just a stack of pricey Apple parts intended to replace a series of PCI slots it really seems like a solution looking for a problem.
I’m thinking of the Black Magic EGPU. It’s cool, but it’s custom design compromises what the original concept was designed to accomplish (modularity) at a extreme price premium, just to optimize the platform. I guess it depends on what you want. Flashy and bespoke or something that works “off the shelf” and comes in about 20-30% less expensive that may or may not be “fit and form” perfect.
A tower about half the size of the cMP with PCI 3.0 seems like a much simpler solution. Apple can design and fabricate it without much hassle (we know they don’t like spending time on these “pro” systems anyway.
If they do stick with a similar look in a smaller package, it would be a clear signal to pro’s Apple is listening and knows what they want. Make it available in space grey and we’ll all have orgasms. “Another classic in the making”. Great tag.
They already have enough to scrap from the cMP (ditch the giant optical drives, the giant heat sink, the ridiculously huge power supply, the 3.5” Hd’s and replace them with SSD and M2. The new iMac pro is a great template in terms the internals. There’s not much to work out on that end. (IMO they are wasting Xeons on that machine).
A nice “core” package with a few upgrade options like processor (single or multiple cpu configs) would make for really nice expandable package/ range or options.
For those that need even more expansion, make a really nice expansion chassis available where the user can add a second GPU, more drives, more expansion cards etc. Maybe they can work out a unique interconnect to take advantage of native pci performance. If you don’t need all of that there are plenty of thunderbolt 3 third party options for lighter weight work, albeit you may have to make several
purchases and may have to share TB3 bandwidth.
I feel like the expansion chassis and how it connects to the “core” are what would define a real “pro” system more so than most of the parts inside. The user wants to add those and they don’t want to not can they rely on Apple for all of those parts. The more they lock it down the fewer third party options would be available.
I couldn’t imagine buying a whole box when all I want to do in the future is a “usb4” or new 9K broadcast monitor card. I would have thrown out my tower at least 3 times in that scenario.
Also whatever chassis they conceive should work with the IMac Pros an the MBP’s for those that need it, thus expanding the range of “pro-ness” to three and not just one of their most respected lines.
A MacBook PRO that care about thinnest more than preformance? A Mac Pro is not really upgradable?
Yes, they are powerful when they are newly released but after a few years. People will want to upgrade. Maybe the REAL pro with deep pockets can buy a new one every year. But not all pro will do that. Just look at how often Apple updates the Mac Pro make me sad.
Different Pro's need different things. While some pro's won't bother doing anything to alter their system, others expect it to be changeable.
The word Modular can be seen as both internal as well as external. The 2013 Mac Pro is mostly a modular system externally (excluding RAM & SSD). Whereas the 2012 and older crates are internally modular. I can also see the hybrid idea of blocks interconnecting as a modular system as well.
The need of a given subgroup of Pro's also needs to be looked at from what has been stated Apple is going to focus the new Mac Pro on image and video editing as well as AR/VR production. Given their tight relationships with Pixar and Disney that makes sense!
Sadly, that leaves a large group of Pro's hanging in the wind like engineering, code & web development and lastly music editing.
So looking through my crystal ball I see a semi-closed architecture similar the the 2013 Mac Pro, where the panels within the system are interchangeable at time of build and not likely afterwards. So you can get different CPU & GPU modular boards, RAM will be still serviceable and boot and application storage (secure enclave) all other storage will be external. The interface between these modules with be PCIe 5.0 as well as the option for InfiniBand interlink between Mac Pro's so they can be ganged into a compute-cluster for still higher computing (small scale super computing) for CGI and animation.
The sign posts are clear! Apple is sticking within the visual fields which is still the most dominate pro space.
Truly, it shouldn't have taken them this long to design and build the next Mac Pro unless they needed to wait for standards as well as the technology to mature.
Will Pro's buy it? Large companies will for sure, small shops might, and the home user likely not. The new Mini is for them.
Now, if Apple would only create a working Pro's MacBook Pro instead of slapping the Pro moniker on what was clearly the next generation MacBook. The Real MacBook Pro would have what the engineering, code & web development and lastly music editing pro's want. Ports (no dongles), MagSafe and performance the i9 could offer if it had the needed cooling the current MacBook Pro's don't have do to the size. As well as a bigger battery option.
Then Apple would have all the bases covered.
Today (and it’s been this way for a long long time), personal computing is done via GUI EVEN KEYBOARD SHORTCUTS display the result of their irrational on a display.
And if you want to drill drill down to the topic at hand, the Mac Pro is a GUI driven computer.
A display is not an accessory. It’s a required component of the computer, regardless of the fact that it’s not built in.
In fact, responses like yours, which neither refute nor add anything to the discussion is why I qualified my comment with “personal” computing.
But alas it’s to no avail on the interwebz.
And guess what one of the use cases that Apple showed that Mac Minis were useful or used in? Yep, you guessed it - headless configuration (sans monitor)... rows and rows of Macs housed to act as a computer/rendor/colo backend. Having a monitor in that configuration would be ... unmanageable. It is still an accessory.
I just bought a new monitor on Sunday (3 hour delivery at 10:20 at night - good service) -- and it was... on the accessories - monitors tab ;o
You said your Vega, in a carriage, which takes between a 20-30% hit in eGPU tax, is faster than an iMac Pro Vega 64. If you take a carriage 64 operating at 70-80%(I assumed the latter) and you say that it's faster than the iMac Pro 64, that would in fact mean that its full, true speed is 25% faster than the iMac Pro version. Hence why I wanted to see those benchmarks. They don't make a lot of sense to me.
While those carriages are 300-400, when adding a Vega 56 (4-500), you end up paying 2080 prices for 1060 speed. That seems ok to you?
Regarding your math here -- that only works out assuming that the Vega 64 in the Mac Pro is at stock clock speed. But, it is not, we presume for thermal reasons.
The second issueis never offering a Mac Pro at a reasonable price point to drive sales volume. Without reasonable volume the Nac Pro will never sell well enough for Apple to give a damn. This means a professional level Mac Pro that starts at less than $1500 dollars. Very few engineers, programmers, doctors or other professionals will even bother to consider Apples high end machines that don’t offer what they need. For one thing most users don’t need dual GPU cards. One good GPU card paired with a decent desktop processor would solve a lot of user needs. The dual GPUs as default never really clicks with me and frankly puts the Mac Pro into a special use category. That category simply is not enough business for any body.
That doesn't answer my question at all. Presumably, (actually no presumption is necessary...benches exist), the iMac Pro's Vega 56 puts out close to 1070-1070 Ti Workflow speed. That justifies the cost and the power.
The point isn't what people are worrying about...which is DIY. The point is that with eGPUs, you pay twice as much for substantially less performance than what most pros truly require. Anything above 1070 performance is currently unattainable with a carriage. Period. Once the informal Mac drivers for the 2080 are out, maybe you get 1080 performance (Vega 64), but by paying so much that you put an iMac Pro's hefty cost to shame.
So my question stands. If the Mac Pro is modular in the respect that it allows eGPU spiderwebbed fusterclucks of nerfed performance, does the prospect of paying 2070-80 prices for 1070-80 performance sound worthwhile? Because I can tell you that most people in this forum and on the street don't think so.
The last Mac Pro was a marvel of engineering but shipping them with dual (not even crossfired...dualed) D700s on the high end, which were just "Pro" rebadged 270xes was a joke. The 2013 iMac brought a 780m that outperformed each card and they rarely worked together as effectively as people liked. For graphical tasks, an imac that cost half as much *literally* was almost as good a workstation for many tasks.
And now we seem to have the same situation. The iMac Pro is horrendously priced. However, the cost is justified. It has an expensive CPU. Expensive RAM. An expensive GPU that, for all benchmarked indications seems to run at about 90-95% of the listed speed, pricey storage and of course that 1000 dollar screen. The system may have been marked up heavily but it was comprised of top shelf parts.
The markup in a SuperMini with eGPU Vegas is hardly worth the same price, when you're eating a far larger hit on performance. The last Mac Pro was a failure. It was literally obsolete within six months of its release. No crossfire support. A midrange GPU in a 3-4000+ system and a difficult to expand system with no real Thunderbolt expandability (No TB 3). We'll see how this one works out.
The last two times, the web drivers were updated within a week of the OS shipping to customers, and there were releases during the beta period after WWDC. There weren't this time.
I have no idea when TB4 will come out, but at that point I think you finally reach the point where those speeds can’t be beat by PCI-E — and that sort of thinking may be why Apple is so reticent to say stuff like that.
One think I know “for sure” though — you’re not going to get a machine that can be upgraded/changed in a wide variety of ways without paying Apple more money for new “modules,” I’d wager. I do think the company is polling and listening to serious Pro users — and I think this forthcoming system will be priced in a way that reflects the sort of person or company who can make back the cost of that investment in a couple of days of specialized work.
Not trying to be facetious because a Venn diagram of all common characteristics of all 10,000 equally relevant definitions may actually collect up a subset of features, capabilities, and characteristics that more than 50% of computer buyers are comfortable with. Hopefully the result of such an exercise would align with Apple's definition, but Apple would not be bound to it. Apple could say "screw you," retire the Pro moniker entirely, and use the suffix "Extreme" instead.
There are some slivers of technology that attempt to associate names or designators with useful meanings. Most of these are standards, e.g., IEEE 802.11ac. You could even say that the term "managed" has relevance in terms of network equipment. Back in the days of TTL chips the 54xx named chips had a clear and significant difference over their 74xx brethren. Later generations of technology have kind of fallen off the wagon and at-best only make a feeble attempt to conform to expectations, e.g., the original expectations around "4G" cellular platforms were abandoned in favor of vendor marketing perverted definitions and non-conforming implementations. In all likelihood "5G" will suffer the same fate. So even in the face of what looks like a "standard definition" with an industry assigned moniker the infinite power of individual company marketing, buzz-speak, spin, and the ease of pulling the wool over the eyes of oblivious consumers beats all reason, conformity, and expectations to a bloody pulp, with nothing left but a meaningless blob of self obfuscating characters strung on to the tapered end of a product name. Whether Apple decides to call something a Mac Pro, Mac Orb, Mac SS, Mac Gasm, Mac Lite, Mac Max, Mac 52, ... or whatever, it's all the same and the name basically means nothing at all to anyone - except for Apple, and they're not going to tell you anything about their naming formula.
AppleInsider's guess is as good as anyone's. While I hope you are wrong, generally my best way of predicting what Apple will do is, think, what would really really disappoint me, and that will be pretty darn close.
Lack of storage expansion is inaccessible.
Where I take some issue is with the fuzzy word "pro". Apple it's self mentioned that pro's encompass a very wide number of user types. This article basically seems to focus on pro's as pixar like creative professionals.
A broader umbrella for pro includes many other types. Power users. Enthusiasts. Pro's from very different industries like scientific research that requires MASSIVE upgrading, including video cards, storage, ram, for computation.
In essence, the article suggests that the new Mac Pro will in essence be a repeat of the failed trashcan, but with a faster thunderbolt 3 port this time, which is still pathetically slow compared to PCI. Certainly possible. But it's fundamentally why I think this article will prove false in the end. And if not, it's time to sell the stock, because they'll just keep re-making cubes.
But Apple has already lost a lot of pros (in the broader sense of the word). If it doesn't deliver here, it will lose near the last of those old loyal users. Apple can do just fine without them on iPhones alone. But some day, it may need their enthusiasm, and they wont be there. They will be on windows/linux and some on hackintosh for a while longer.
And to the article's point, it's not like the Cheese Grater's adoration isn't know. So ignoring it at this point is basically an invitation to leave. Sadly, it's very possible.
Why Apple waited so Long to fix a machine that couldn’t possibly drive volume is beyond me. The required adjustment could have been done by a work study engineer. Effectively you delete a GPU card and put in a M2 carrier board.
Beyondthat at I cant even begin to understand the lack of an Apple branded storage array.
I got the impression from the first time they said "modular" that this is the kind of approach they would go for. A central processing box, with connectivity to external storage, graphics, etc. I was surprised at how many people thought "modular" meant a cheesegrater-like box. The cheesegrater isn't really modular.
I think their way to accommodate the many different "Professional" users out there is to provide for different external units that plug in over TB3. People who aren't doing video or graphics work (and even some who do) will be fine with the internal GPU, for others there are the eGPUs. Similarly if you're working with audio, plug in an external DSP or hardware encoder.
Then, when there are new processors available, you upgrade the central box, and keep all the peripherals, provided TB3 is available. Yes, that is a risk, but any future version is likely to stay backwards-compatible, and the worst outcome is you buy a (GASP!) dongle. I expect, however, the USB-C connector will stay for a while. Maybe as long as the USB-A connector has.