Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Profit-wise AMD would be a better option for Apple. Similarly as MS and Sony ordered a custom chip, payed AMD a lump sum and then just started a copy-machine, Apple could design their custom SoC with AMD. The more SoC's they're able to produce, the less the R&D cost per chip is going to be. With Intel I don't think this is possible even though they brought up a semi-custom chip recently, I don't think Intel would give it away... they want to make it themselves and charge per chip.

If AMD CPU arrives, I'm sure Apple will write all its apps and the os to work fully under HSA concept. That's how they can make a distinct to others.

But for Mac Pro I can't see a point to make a custom SoC just for it. Volumes would be too low. Therefore I think that if AMD is Apples way, mMP will be delivered with an off-the-shelf parts... unless the custom part will end up to iMP 2018 at the end of the year?

And speaking of custom chips, isn't Xeon W also an Apple special version of the Xeon? If so, it would be odd if Apple is not going to use it in other products too, like mMP.
 
Last edited:
Taking the Mac Pro off into a corner where is share no significant parts with any other Mac would be one of the most sure fire ways off killing off the product. The Mac Pro is highly unlikely to sell enough to offset those costs for a better return than other Apple products and then next "what isn't working so it should be terminated" meeting Apple execs hold the Mac Pro would be sitting there with a bullseye painted on it.

Volume is not a concern for Apple at this point .

But would it not kill off all Macs eventually, if there was no flagship model, and one that can compete with all comers ?
Meaning be just like them , only with OSX .
The days of being different have come and gone with the G models, it's about OS only now .

I'll say you can't sell any Mac in the long run without a widely accepted and used workstation line .
Now try selling an overpriced iAnything based solely on the power of iOS and European-ish product design - not gonna happen .

Have your execs explain in that meeting how Apple went Blackberry in a matter of years , after relying on a few geeky laptops and iMacs to carry the torch , and as a result getting kicked out of the entire market in record time .

If I had Apple stock, I'd be watching this space closely .
 
  • Like
Reactions: ssgbryan and -hh
Profit-wise AMD would be a better option for Apple. Similarly as MS and Sony ordered a custom chip, payed AMD a lump sum and then just started a copy-machine, Apple could design their custom SoC with AMD. The more SoC's they're able to produce, the less the R&D cost per chip is going to be. With Intel I don't think this is possible even though they brought up a semi-custom chip recently, I don't think Intel would give it away... they want to make it themselves and charge per chip.

If AMD CPU arrives, I'm sure Apple will write all its apps and the os to work fully under HSA concept. That's how they can make a distinct to others.

But for Mac Pro I can't see a point to make a custom SoC just for it. Volumes would be too low. Therefore I think that if AMD is Apples way, mMP will be delivered with an off-the-shelf parts... unless the custom part will end up to iMP 2018 at the end of the year?

And speaking of custom chips, isn't Xeon W also an Apple special version of the Xeon? If so, it would be odd if Apple is not going to use it in other products too, like mMP.
the one in the IMAC pro are just down clocked CPUS to downgrade the new mac pro like that will be a joke it does not need to look cool it needs power.

Also useing AMD cpu's is just doing drivers more or less.
 
We have no info on the next Purley chips, right? If Apple came out with a new Mac Pro by the end of this year it'd have to be using processors that exist right now.

Does someone know if any of the previous 2009-2012 era Mac Pros had the same or different sockets? Because I can't imagine Apple making two different mobos with sockets for the Mac Pro, in which case they'd either go with Silver/Gold dual socket or (more likely) stay with the -W chips they're already using. Given that the iMac Pros mostly don't throttle but are geared for silence more than anything else and they start at 8 cores, you could have a lower-count buyin with the Mac Pro and still have a more powerful machine at the high end. Only people it won't please are the dual socket peeps.
 
We have no info on the next Purley chips, right? If Apple came out with a new Mac Pro by the end of this year it'd have to be using processors that exist right now.

apple isn't using Purley now so why would they be using it on the next iteration?

There is some info about Cascade X which likely is coupled to a "Cascade W" . Intel W is what Apple is using now ( Skylake W ). A recent Intel timeline puts the "X" version at about Q4 2018 .

https://www.extremetech.com/computi...ew-high-end-cascade-lake-x-debuting-next-year

It is highly likely Cascade roughly corresponds to what Coffee Lake is to the mainstream die design. For the most part a clock speed bump ( 14nm+++ yet another optimization round on same process ) . While the top end Coffee lake processors got a core count bump ( 4 -> 6 bigger die at the same price ), that would be doubtful for the Intel W series which is already a substantially bigger die ( substantial enough that it already effects yield and for highest core count approximately at lithographic limits. ) .


For the SP class more likely getting memory subsystem upgrade to support Optane DIMMs ( https://www.anandtech.com/show/12041/intel-to-launch-3d-xpoint-dimms-in-2h-2018 ) than core count increases. And a bump in clock speeds and thermals.




Does someone know if any of the previous 2009-2012 era Mac Pros had the same or different sockets?

2006-7 same. 2008 is oddball because the "front side bus" design was a dead end ( at 6-8 cores it is firmly on the downward slope of diminishing returns. ); same but only a year long. the whole Xeon E5 sequence was "two cycles on same socket and chipset". There is very high chance that hasn't changed.

When Intel shifts to 10nm+ (and if not then then certainly at 10nm++) there will be a socket change. (https://www.anandtech.com/show/1111...n-core-on-14nm-data-center-first-to-new-nodes . )



Because I can't imagine Apple making two different mobos with sockets for the Mac Pro,

All the more highlighted by the 2009-2012 models going to CPU daughterboard just so could share the same main logic board and the rest of the infrastructure. If there was volume to support two substantially different Mac Pro models they could have done it then. It is even more complicated now because not only the sockets but the PCH chipsets are different. NUMA effects more pronounced and the boot contexts more varied.



Given that the iMac Pros mostly don't throttle but are geared for silence more than anything else and they start at 8 cores, you could have a lower-count buyin with the Mac Pro and still have a more powerful machine at the high end. Only people it won't please are the dual socket peeps.

if throw in 8 DIMMs slots would make a substantial fraction of those dual socket folks happy..... since a fair number of them wanted the 8 slots at least as much as the extra cores. ( Cheaper path to larger memory footprints through "mature tech" DIMMs. ). A Mac Pro moving from a 128GB upper cap on RAM to a 256GB upper cap on RAM is probably "big enough" for most folks deeply hooked on effectively running "RAM disks" inside their applications.

Can't make all of the historic 2 socket folks happy, but Apple can skim off more than a healthy chunk with 18 cores and 256GB of RAM. [ can snag a bit more with support for more cost effective "RAM disk"-like performance with Optane SSD; "low enough to be very useful' latency at about half the cost of RAM. ]
 
(Apple have) systematically made poor design/engineering decisions that alienated their consumer base; that is the premise I put forth to their poor sales, not that a 'Mac Pro customer base just isn't there or evolved away'.

This forum wants a Dell OptiPlex form-factor that they can upgrade at will. Any Mac that does not meet this is considered a failure of Apple design.

To quote Tim, "People Love the iMac" and that is the Mac desktop they have been buying. Since 2000, the Mac Pro has been a very small percentage of Mac desktop sales and since 2010, desktops have been a very small percentage of Mac sales in total.

The Intel cheese-grater was rarely updated under Steve and that was because it was selling so poorly in comparison to the iMac. Now the cylinder has not been upgraded under Tim because it is selling so poorly in comparison to the iMac (25,000 MP sales or less per quarter compared to almost 1,000,000 iMacs sold per quarter in 2017).


Volume is not a concern for Apple at this point. But would it not kill off all Macs eventually, if there was no flagship model, and one that can compete with all comers?

I do not believe so. Since Apple has moved their focus away from only offering high-end PCs (PowerMac and PowerBook), they have seen their sales quintuple (the peak under the Power era was 4 million units in 1995 compared to the current peak of over 20 million).
 
Do you mean the G3, G4, G5? (PowerPC CPUs?)

Yup.
The OG ... ;)
[doublepost=1515434839][/doublepost]
This forum wants a Dell OptiPlex form-factor that they can upgrade at will. Any Mac that does not meet this is considered a failure of Apple design.

That's about sums it up .

To quote Tim, "People Love the iMac" and that is the Mac desktop they have been buying. Since 2000, the Mac Pro has been a very small percentage of Mac desktop sales and since 2010, desktops have been a very small percentage of Mac sales in total.

I was not aware that Tim is considered quotable by some , and still don't think he is .

The Intel cheese-grater was rarely updated under Steve and that was because it was selling so poorly in comparison to the iMac. Now the cylinder has not been upgraded under Tim because it is selling so poorly in comparison to the iMac (25,000 MP sales or less per quarter compared to almost 1,000,000 iMacs sold per quarter in 2017).

Steve always hated the Mac, and loved the iMac ( and iGadgets ).
For good reasons ; it saved the company and is a great design .
It also was and still is no substitute for a tower .

Both Tim and Steve are/were marketing guys , they just want to sell stuff .
Remember Steve declaring the iPad was the future of computing ? Ha, ha and ha .

However, if Steve's PowerMac hadn't gone Intel MacPro - arguably the the most important change along with designing the iPod and iMac - we might all be using Windows by now .
Without a usable MP, there'll be no Macs at all, that simple .
It's the OS .


I do not believe so. Since Apple has moved their focus away from only offering high-end PCs (PowerMac and PowerBook), they have seen their sales quintuple (the peak under the Power era was 4 million units in 1995 compared to the current peak of over 20 million).

Apples to oranges , no pun intented .
Apple extented their footprint in the PC market with the original iMac , and got established with the cMP in 2006 .
They peed away their potential when the cMP was updated only once, OSX lost backward compatibilty with most major programs, and the trashcanMP proved to be a one way street into oblivion .
 
Last edited:
  • Like
Reactions: ssgbryan
I wouldn't understand why the iMP gets a Xeon and the mMP should get a Threadripper instead of an EPYC.

But again, I still don't understand why all of them (am I right?) are missing QuickSynce or any h264/265 hw coding, in the era of multimedia and videoconference.


Apple has a vendor independent video encode/decode frame work. The modern version ( as of macOS 10.8) is

https://developer.apple.com/documentation/videotoolbox

(the older from 10.6.8 era version that is deprecated/obsolete is

https://developer.apple.com/library/content/technotes/tn2267/_index.html )

The video encode/decode fixed function units are typically attached to a GPU's set of fix function units. Usually because some parts (or corner cases ) of the computation will be either done the GPUs computation hardware and/or leverages the GPUs memory streaming subsystem. So for Intel , no iGPU means no QuickSync. I'm pretty sure the same is true from AMD.

That doesn't mean that there isn't a hardware decode/encode subsystem not available. AMD has one in their GPUs. (for example see pages 13-14 here http://radeon.com/_downloads/vega-whitepaper-11.6.17.pdf ) This issue is flushing out the interconnections and interfaces with Apple's framework. ( or developers not bypassing the framework and making specific low level calls directly to QuickSync. )

Decoding relatively close to the video frame buffer makes sense since really want to push the expanded video out to the monitor as quickly as possible. Encoding is often being pushed onto the Internet (or storage disk ) so local to GPU isn't quite so naturally coupled to the output framebuffer. However, encode and decode are highly likley to share lots of the same fixed function logic/circuits.




[doublepost=1515353417][/doublepost]Could it be solved with an ARM co-processor?


If developers are making QuickSync (or specifically tuned for QuickSync ) calls to the hardware then ARM isn't going to help at all. Likewise if it is a poor connection between Apple's framework and the GPU's encode/decode units .... just being ARM isn't particularly going to help. ( the fix function units from iOS devices may presume that the RAM is shared between GPU and CPU and that's would be as equally a disconnect as it is for a dGPU for a "co-processor". )


A corner case would be if the T1/T2/T-next co-processor was handling the Facetime webcam anyway. It could make sense just to compress the stream before even gets to the rest of the system. But that is only one direction for an embedded camera.
 
  • Like
Reactions: askunk
This forum wants a Dell OptiPlex form-factor that they can upgrade at will. Any Mac that does not meet this is considered a failure of Apple design.

To quote Tim, "People Love the iMac" and that is the Mac desktop they have been buying. Since 2000, the Mac Pro has been a very small percentage of Mac desktop sales and since 2010, desktops have been a very small percentage of Mac sales in total.

The Intel cheese-grater was rarely updated under Steve and that was because it was selling so poorly in comparison to the iMac. Now the cylinder has not been upgraded under Tim because it is selling so poorly in comparison to the iMac (25,000 MP sales or less per quarter compared to almost 1,000,000 iMacs sold per quarter in 2017).

Ah I get it. So all these places that have condensed cholesterol in the form of an exotic and artful combination of cow proteines maybe some lettuce, tomator, and a pickle or two, hand assembled to satiate the oddly blue collar yet complicated and discerning palate, and charge 25$ CAD for it, should just give up because they will never sell more than McDonalds, having some poor labourer slap together some vat grown synth product in an unhygienic room while combating a runny nose. In fact the McDonald's synthburger / multicoloured urinal puck should be touted as the paragon of burger design because probably Trillions Served™ by now ...

Again, nothing you say in incompatible with the idea that Apple is missing the bullseye with it's dart, and then claiming it is because the dartboard is in the wrong place, and then claiming the game is unviable.

And please, optiplex is for the plebs. I think we could agree on a Precision level Apple product.

I'll be waiting for that Virginia Tech imac cluster ...
 
Without a usable MP, there'll be no Macs at all, that simple. It's the OS.

You do not need a Mac Pro to run macOS. Out of the 20 million machines Apple sells every year that run macOS, currently 100,000 of them are Mac Pros.

Though as small as that market is, Apple still supports it with new product (iMac Pro) even though one could convincingly argue they don't need to and even should not from a financial return standpoint considering even with the prices they charge, while per unit margins are probably strong, overall program revenue is possibly quite negative when R&D and advertising are factored in.


Apple extended their footprint in the PC market with the original iMac, and got established with the cMP in 2006. They peed away their potential when the cMP was updated only once, OSX lost backward compatibility with most major programs, and the trashcanMP proved to be a one way street into oblivion.

The Intel-powered Mac Pro families likely did little to extend their footprint - that was driven by the iMac, the MacBook Air and the MacBook Pro.


Optiplex? Seriously?!!

OptiPlex. (Lenovo) ThinkCentre. (Hp) Pavilion. Take your pick. Every description of an "xMac" is a desktop or tower where the CPU | RAM | GPU can be replaced and multiple drives can be installed.


Again, nothing you say in incompatible with the idea that Apple is missing the bullseye with it's dart, and then claiming it is because the dartboard is in the wrong place, and then claiming the game is unviable.

It's not unviable, even if it is possibly unprofitable. If it was unviable, Apple would have not even bothered with the iMac Pro, much less a new Mac Pro. The 5K iMac would have been the top system and if you didn't like it, then suck eggs and run Windows or Linux on a Dell Precision or HP Z-series.

And in the game of Darts, the bullseye is not the highest value target. Apple is making more money (scoring more points) by hitting the 51, 54, 57 and 60 point spots on the inner ring rather then just targeting only the bullseye (worth 50 points).
 
Last edited:
And in the game of Darts, the bullseye is not the highest value target. Apple is making more money (scoring more points) by hitting the 51, 54, 57 and 60 point spots on the inner ring rather then just targeting only the bullseye (worth 50 points).

Yes yes... Wwwwooooooonnnn huuuundred aaaahhhhnnndddd eeehhhhhhhhteeeeeeee ( in slightly tipsy welsh accent )
 
The Intel cheese-grater was rarely updated under Steve and that was because it was selling so poorly in comparison to the iMac.

It was updated roughly as often as new Xeons were available, until Sandy Bridge. Just as one might expect.


Now the cylinder has not been upgraded under Tim because it is selling so poorly in comparison to the iMac

Actually, according to Apple executives, it has not been upgraded because of poor decision making during the design phase when they opted to go with a ludicrously low thermal budget for what was claimed to be a pro machine. This resulted in extraordinarily limited update options, thermal throttling, and rampant GPU failures.


(25,000 MP sales or less per quarter compared to almost 1,000,000 iMacs sold per quarter in 2017).

Wasn't aware this has been broken out in Apple 10-K or similar reports? Can you provide a link?
 
OptiPlex. (Lenovo) ThinkCentre. (Hp) Pavilion. Take your pick. Every description of an "xMac" is a desktop or tower where the CPU | RAM | GPU can be replaced and multiple drives can be installed.

You conveniently dodged the hi end workstation options offered by these vendors, because you know Apple doesn’t have an answer for that. You are trying to insist that people are clamoring for a tower Mac Pro and are essentially after upgradability, not performance.

Do you even use Mac pros ? As your main work system ?
 
It was updated roughly as often as new Xeons were available, until Sandy Bridge. Just as one might expect.




Actually, according to Apple executives, it has not been upgraded because of poor decision making during the design phase when they opted to go with a ludicrously low thermal budget for what was claimed to be a pro machine. This resulted in extraordinarily limited update options, thermal throttling, and rampant GPU failures.




Wasn't aware this has been broken out in Apple 10-K or similar reports? Can you provide a link?

I believe he's extrapolating from total known Mac sales, and the fact that while they didn't give hard numbers Apple admitted the Mac Pro was single digit share.
 
You conveniently dodged the hi end workstation options offered by these vendors, because you know Apple doesn’t have an answer for that. You are trying to insist that people are clamoring for a tower Mac Pro and are essentially after upgradability, not performance.

Do you even use Mac pros ? As your main work system ?


HP even has a compellingly Mac vs Z argument.
 
Wasn't aware this has been broken out in Apple 10-K or similar reports? Can you provide a link?

At the April "Mac Pro Symposium" the execs noted that the Mac Pro was in the "low single digits" of Mac desktop sales and that Mac desktops made up 20% of all Mac sales. So for a five million Mac quarter (which is about the average), desktops would be one million and if you take 2-3% of that, you get 25,000-30,000.


You conveniently dodged the hi end workstation options offered by these vendors, because you know Apple doesn’t have an answer for that. You are trying to insist that people are clamoring for a tower Mac Pro and are essentially after upgradability, not performance.

You are confusing the Mac Pro with the "xMac" (eXpandable Mac) - a small desktop / small tower configuration with a Core i7 or Core i9 and the GPU on a PCIe card plugged into a 16x PCIe slot that they can easily upgrade it to the latest (preferably nVidia) option because the majority of folks who want Apple to release one want to be able to play games on it. We're talking effectively a Hackintosh, just made and supported by Apple.

Because Apple will never give them an "xMac", those folks have now pinned their hopes on this new "modular, expandable" Mac Pro because even though they don't need Xeon CPUs and ECC memory and multiple SSDs, they're praying the (hopefully nVidia) GPU is on a PCIe card that plugs into a PCIe x16 slot. Of course, they're also terrified of what all that is going to cost and that is why they keep calling for the "xMac" as it would be thousands of dollars cheaper while still giving them the powerful (nVidia) GPU they want.


Do you even use Mac pros? As your main work system?

Nope. I've used iMacs my entire Apple career (now over a decade). Currently a mid-2017 top-spec 5K as my workload does not require more than four cores and what little Windows and macOS gaming I do runs fine on the Radeon Pro 580 so I don't need to move to an iMac Pro.

Honestly, I'd be happy with a Mac Mini if it had a four-core i5, the current generation of NVme SSDs as in the iMac 5K and TB3/USB-C as I would then connect it to one of those new 34" TB3 monitors coming out while also having a small-form-factor Windows box dedicated to gaming with an nVidia 1070.

But just because I personally don't need or own a Mac Pro, I have educated myself on the roles the Mac Pro currently and historically served as well as where it currently and historically has fit into the Mac ecosystem. That education has shown me there is a place for a high-performance high-core computer in the Mac ecosystem, but it has also shown me that this has not been the primary focus of the Mac ecosystem for almost the past two decades.


[doublepost=1515442273][/doublepost]
HP even has a compellingly Mac vs Z argument.

Are they using Adobe apps for those comparisons? That would certainly favor a Z-series when equipped with nVidia GPUs.

Heck, the main reason workstation users give for having left Mac is not because of the lack of updates to the cylinder Mac Pro, but because the cylinder Mac Pro does not have nVidia GPU options and their workloads are focused on applications that are designed around CUDA.
 
Last edited:
You are obviously not concerned with a Mac Pro that’s for sure.

Without having used or needing the Mac Pro you claim you have educated yourself as regards to its historic viability.

Should have guessed when you said dual socket systems do not run macOS ‘like ever’.

Riot !
[doublepost=1515442891][/doublepost]
You are confusing the Mac Pro with the "xMac" (eXpandable Mac) - a small desktop / small tower configuration with a Core i7 or Core i9 and the GPU on a PCIe card plugged into a 16x PCIe slot that they can easily upgrade it to the latest (preferably nVidia) option because the majority of folks who want Apple to release one want to be able to play games on it. We're talking effectively a Hackintosh, just made and supported by Apple.

Because Apple will never give them an "xMac", those folks have now pinned their hopes on this new "modular, expandable" Mac Pro because even though they don't need Xeon CPUs and ECC memory and multiple SSDs, they're praying the (hopefully nVidia) GPU is on a PCIe card that plugs into a PCIe x16 slot. Of course, they're also terrified of what all that is going to cost and that is why they keep calling for the "xMac" as it would be thousands of dollars cheaper while still giving them the powerful (nVidia) GPU they want.

In short : You.
 
  • Like
Reactions: ssgbryan
the roles the Mac Pro currently and historically served as well as where it currently and historically has fit into the Mac ecosystem.

This is something that I think is important. The Mac Pro was never meant to sit on every desk in every office-but there is a demand for such power. Even if they aren’t selling at MacBook Pro and iMac volumes there is still a need for that kind of power (and more power as so much of technology advances). I don’t think anyone here is really gushing over a “new sexy form factor” as much as “holy **** that things powerful” which translates to “I can get a whole lot more done” or “I can be so much faster”.

The Mac Pro, although a sliver of the overall pie of Mac sales still has a place to be updated regularly for those who need/want a powerful competitive Mac. It certainly doesn’t need to be earth shattering in the mm’s that it shaved off or that it’s now in a glossy/polished space gray finish.

Leaving it on the vine or neglected like the step-child sends a message-and I think you can find in every industry those who can get along with the maxed iMac or MacBook Pro, but then there are those who really want beasts to churn out projects quickly. It is also worth noting that they are reliant on third party parts (intel, AMD, etc.) but 4 years, no exit plan, thermal corner... someone forgot something along the way. Whether it’s video editing, developers, graphics, design, CAD and BIM, among all the others who desire one, a nice beast with modern tech would be greatly appreciated (and even more of it was updated at a minimum of every 3 years, but would certainly like closer to 18/24 month revision cycle).
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.