MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)

[goMac]

I don't believe anybody at Apple has ever stated exactly when any redesign or new Mac Pro/mMP started.

Could be 5 months. Could also be a year.

My impression from what I’ve heard is that they honestly did start around the time they got all those reporters together to talk about the Mac Pro.

The internal conversation was:
- Can we upgrade the trash can? (This was mentioned in the Mac Pro interview)
- Can the iMac Pro replace the Mac Pro?

When both of those came back no, they called the Mac press together and started work. But that also puts the decision late in the iMac Pro’s preproduction phase, which makes sense. The iMac Pro was the original plan b which they realized would not work when they showed it around before WWDC quietly.

I’ve heard through birdies this is how things went. There even was a time early this year when they were looking at the iMac Pro where no Mac Pro work was being done at all.
[doublepost=1508527317][/doublepost]

Absolutely nothing related to the trash can Mac Pro would be usable in the new Mac Pro.

I really believe that this is a start from scratch scenario. As for when they started development vs date of product delivery. It's a pointless argument. We have zero control of apples timeline. Whether they started 5 months ago or 5 years ago. Neither can positively determine the delivery date.

The best prediction to probably make around this is Apple will probably try to reuse parts of the iMac Pro in the Mac Pro. I don't mean that as in something like an AIO design or no slots or under clocked CPUs or anything... More that they're already going to be working with Intel on the iMac Pro, and those engineers are already going to be familiar with building a workstation around Xeon chipsets. Which is another reason I'm not inclined to believe the AMD CPU buzz.

I don't know when this thing is going to ship but I feel pretty safe saying there won't be an announcement this year. Especially in since Apple announced there will be no more events this year.

 

[beaker7]

I'm not sure we should be too worked up about timing yet. They were obviously going to be waiting for Skylake Xeons by the time April 2017 came around. And Intel hasn't even got those shipping in much volume yet. For example, you can't buy an HP Z series with Xeon SP in it yet. Even if this was going to be a cheese grader, there is good reason to think Apple wouldn't have any details on it yet, much less would it be available, given their history.

Dell has been shipping servers with Xeon SP for a couple months, and workstations for a few weeks.

 

[ventricle]

I remember reading that back in the Jobs days, people would regularly be sent to work on projects that would end up being nothing or scrapped. Somewhere in the Apple Vault there is probably a mostly finished Mac Pro that they could work from.

And it could only be released at a desired price : performance ratio vis the competition that would be seriously embarrassing. It’s now 1402 ****ing days since last release, hang the Caution tag.

 

[-hh]

My impression from what I’ve heard is that they honestly did start around the time they got all those reporters together to talk about the Mac Pro.

The internal conversation was:
- Can we upgrade the trash can? (This was mentioned in the Mac Pro interview)
- Can the iMac Pro replace the Mac Pro?

When both of those came back no, they called the Mac press together and started work. But that also puts the decision late in the iMac Pro’s preproduction phase, which makes sense. The iMac Pro was the original plan b which they realized would not work when they showed it around before WWDC quietly.

I’ve heard through birdies this is how things went. There even was a time early this year when they were looking at the iMac Pro where no Mac Pro work was being done at all.

Unfortunately, such rumors are 100% believable in this age-of-Apple, even though it is pretty profoundly incomprehensible from a strategic standpoint to ever have a product line drop to "zero" Engineering support, particularly when you look at how well resourced Apple is ... including the squandering of $millions because the Magic Mouse 2 'click didn't sound right':

https://www.wired.com/2015/10/exclusive-why-apple-is-still-sweating-the-details-on-imac/

FWIW, do note that this Magic Mouse 2 resource expenditure was taking place in late 2015 ... precisely the period when the tcMP was already two years old and thus, should have been scheduled for its hardware refresh design work .

[doublepost=1508527317][/doublepost]

The best prediction to probably make around this is Apple will probably try to reuse parts of the iMac Pro in the Mac Pro. I don't mean that as in something like an AIO design or no slots or under clocked CPUs or anything... More that they're already going to be working with Intel on the iMac Pro, and those engineers are already going to be familiar with building a workstation around Xeon chipsets. Which is another reason I'm not inclined to believe the AMD CPU buzz.

Sure, and Apple has recycled/repurposed quite a bit of their designs over the past decade, such as how the mini & iMac are in large part a repackaged/derivation of their laptops...even including how they've hamstrung user-upgradability. /S

In contemplating more of the "nuts and bolts" of core PC architecture, with a Xeon-based system, I can see some (valid) technical debates over stuff like allocations of PCI lanes ... etc ... such that there may very well be something like Northbridge/Southbridge controller selection (or redesign) which could contribute to cost-schedule-performance.

But what I'm afraid of is the risk here is in "Apple being Apple" just to be different, and taking a six month hit on schedule in order to have Intel's chip designers & foundries fab a custom-for-Apple controller chip. presumably because Apple believes that some piddling 2.7% difference vs Intel's off-the-shelf (available now, & defined price points) is worth more to them than delaying the next-MacPro for another six months (or whatever).

Of course, the cynical view of Apple here is that they see just such a schedule delay as "necessary" (worse: "beneficial") because they: (a) want to market the nMP as even higher performance than the iMac Pro, and (b) see having more separation in release schedules as a way to minimize cannibalization of iMac Pro sales.

You'll note that neither of these speculative considerations are customer-focused.

I don't know when this thing is going to ship but I feel pretty safe saying there won't be an announcement this year. Especially in since Apple announced there will be no more events this year.

Indeed. As vague as the April meeting was, they barely committed to having anything by the end of 2018.


-hh

 

[pat500000]

My impression from what I’ve heard is that they honestly did start around the time they got all those reporters together to talk about the Mac Pro.

The internal conversation was:
- Can we upgrade the trash can? (This was mentioned in the Mac Pro interview)
- Can the iMac Pro replace the Mac Pro?

When both of those came back no, they called the Mac press together and started work. But that also puts the decision late in the iMac Pro’s preproduction phase, which makes sense. The iMac Pro was the original plan b which they realized would not work when they showed it around before WWDC quietly.

I’ve heard through birdies this is how things went. There even was a time early this year when they were looking at the iMac Pro where no Mac Pro work was being done at all.
[doublepost=1508527317][/doublepost]

The best prediction to probably make around this is Apple will probably try to reuse parts of the iMac Pro in the Mac Pro. I don't mean that as in something like an AIO design or no slots or under clocked CPUs or anything... More that they're already going to be working with Intel on the iMac Pro, and those engineers are already going to be familiar with building a workstation around Xeon chipsets. Which is another reason I'm not inclined to believe the AMD CPU buzz.

I don't know when this thing is going to ship but I feel pretty safe saying there won't be an announcement this year. Especially in since Apple announced there will be no more events this year.

Those people need to be replace. Having a conversation about those two questions is nonsense. It sounds like those over there are just plain lazy. Maybe it's time to outsource them and get someone from outside who loves mac pro and design something.

 

[ventricle]

particularly when you look at how well resourced Apple is ... including the squandering of $millions because the Magic Mouse 2 'click didn't sound right':

https://www.wired.com/2015/10/exclusive-why-apple-is-still-sweating-the-details-on-imac/

FWIW, do note that this Magic Mouse 2 resource expenditure was taking place in late 2015 ... precisely the period when the tcMP was already two years old and thus, should have been scheduled for its hardware refresh design work .

The first puck mouse was very uncomfortable to use, its successors have hardly improved. The last time I checked, of eight mice in use in my family at least five are wired (possibly all) and none are designed by Apple. As for keyboards in use, none are the short, wireless, Apple default type. Apple simply can’t see the forest for the trees. Talk about form fetish!

You'll note that neither of these speculative considerations are customer-focused.

No competition: If there were at least two hardware manufacturers that could legally run OS X, this thread would not exist. But even the family jewels (OS) is losing its sheen due to paranoid restrictions.

 

[rjtiedeman]

i really don’t need super expensive desktop. just want one that can use other monitors. i would be happy with the 6.1 with 2017 tech not 2013 tech.

a lot can happen between now next year to kill this project.

 

[goMac]

Those people need to be replace. Having a conversation about those two questions is nonsense. It sounds like those over there are just plain lazy. Maybe it's time to outsource them and get someone from outside who loves mac pro and design something.

My understanding which I talked about last year and which was kind of mentioned in the Mac Pro interview is that they did try engineering new can Mac Pros. So engineering was working with a design that was forced on them.

When it became clear that the pretty looking box wasn't going to work, they were trying to decide if another pretty looking box (that solved the Thunderbolt/discreet GPU problem already) was going to work.

I don't think Ive has been a great influence (which isn't just true of the Mac Pro) but the Thunderbolt/swappable GPU thing I've mentioned is a real problem within Apple, which is why the iMac Pro was one possible out.
[doublepost=1508787806][/doublepost]

FWIW, do note that this Magic Mouse 2 resource expenditure was taking place in late 2015 ... precisely the period when the tcMP was already two years old and thus, should have been scheduled for its hardware refresh design work .

Well, millions sounds like a lot, but in Silicon Valley money that's not much. Each engineer is probably a quarter million a year, senior engineers more. Executives even more (and you know Ive got involved). And that doesn't include materials, sample runs, etc. An 8 person peripheral team working on nothing but keyboards and mice would probably fit into that budget.

If I had to take a wild guess, I'd guess that the new Mac Pro will probably be at least $10 million in R&D.

 

[Zarniwoop]

Apple could choose AMD's Threadripper tech for next MP if they value two things; insanely fast RAID capable SSD drives, and HSA/hUMA memory model. They already have the latter in iOS, but with Intel it has not been possible. Software developers could get much more out from same hardware with hUMA.

What comes to SSD, Apple could build an insane >20GB/s SSD drive with multiple Samsung 960 or similar drives. Each one needs PCIe x4 and that is easy with Threadrippers' 64 PCIe lanes. Can't do that with Intel without extra cost or suffocating the PCIe channels.

To edit raw 8k videos, one needs a fast storage. Current 4GB/s drives (that have been maximum with Intel DMI) won't work. If 8k is Apple's goal, then AMD is their way.

 

[beaker7]

Apple could choose AMD's Threadripper tech for next MP if they value two things; insanely fast RAID capable SSD drives, and HSA/hUMA memory model. They already have the latter in iOS, but with Intel it has not been possible. Software developers could get much more out from same hardware with hUMA.

What comes to SSD, Apple could build an insane >20GB/s SSD drive with multiple Samsung 960 or similar drives. Each one needs PCIe x4 and that is easy with Threadrippers' 64 PCIe lanes. Can't do that with Intel without extra cost or suffocating the PCIe channels.

To edit raw 8k videos, one needs a fast storage. Current 4GB/s drives (that have been maximum with Intel DMI) won't work. If 8k is Apple's goal, then AMD is their way.

Love the fact that TR sports 64 lanes but there is nothing here that cannot be done using Intel VROC.

 

[CWallace]

Love the fact that TR sports 64 lanes but there is nothing here that cannot be done using Intel VROC.

But at a price - must use Intel SSDs if you want to boot from the RAID and if you want more than RAID 0, you need to buy a physical key from Intel to enable those modes.

https://www.pcworld.com/article/319...le-crazy-raid-configurations-for-a-price.html

 

[flat five]

If that were it, we'd have it by now. Something is taking them this long to bake.

maybe a small team is assigned.

------
and hey, maybe a small team of designers attempting to prove themselves.

(but yes, these are some thin air maybes )

 

[Joe The Dragon]

But at a price - must use Intel SSDs if you want to boot from the RAID and if you want more than RAID 0, you need to buy a physical key from Intel to enable those modes.

https://www.pcworld.com/article/319...le-crazy-raid-configurations-for-a-price.html

Intel SSDs only and a F*** raid key.

 

[goMac]

Apple could choose AMD's Threadripper tech for next MP if they value two things; insanely fast RAID capable SSD drives, and HSA/hUMA memory model. They already have the latter in iOS, but with Intel it has not been possible. Software developers could get much more out from same hardware with hUMA.

I think we're likely to be into AMD and Intel's next generation of CPUs before the Mac Pro ships.

HSA/hUMA is not only on iOS, but it's on macOS as part of Metal. No doubt Apple could take advantage of the acceleration Vega would bring to HSA in Metal, but not much to look at for the CPU side.

 

[Jaho101]

maybe a small team is assigned.

------
and hey, maybe a small team of designers attempting to prove themselves.

(but yes, these are some thin air maybes )

Even if the designs are finalized, they've still got a ton of work to do to build all the machines and production lines. The iMac has had the same tooling for a while now, and the MacBooks just got a refresh that took a while because of the new hardware, these Mac Pros are really going to be from scratch hardware wise, and built out of much bigger pieces than something like iPhones (Apple spends millions just making sure iPhone production gets up running quicker than any project of that scale).


EDIT: Switching to AMD is probably out of the question for Apple, the cost of switching things to work in another x86 architecture and then supporting both legacy Intel and AMD on their own nearly legacy platform doesn't make financial sense. The most we can hope for is lower prices from Intel, since they're not that far behind AMD except in the price/performance ratio.

 

[Peace]

Even if the designs are finalized, they've still got a ton of work to do to build all the machines and production lines. The iMac has had the same tooling for a while now, and the MacBooks just got a refresh that took a while because of the new hardware, these Mac Pros are really going to be from scratch hardware wise, and built out of much bigger pieces than something like iPhones (Apple spends millions just making sure iPhone production gets up running quicker than any project of that scale).


EDIT: Switching to AMD is probably out of the question for Apple, the cost of switching things to work in another x86 architecture and then supporting both legacy Intel and AMD on their own nearly legacy platform doesn't make financial sense. The most we can hope for is lower prices from Intel, since they're not that far behind AMD except in the price/performance ratio.

I'm guessing Apple is working more on it's own SoC designs for future Macs more then they are working on switching to AMD.

 

[Zarniwoop]

I think we're likely to be into AMD and Intel's next generation of CPUs before the Mac Pro ships.

HSA/hUMA is not only on iOS, but it's on macOS as part of Metal. No doubt Apple could take advantage of the acceleration Vega would bring to HSA in Metal, but not much to look at for the CPU side.

hUMA requires 100% access to system RAM for co-processors such as GPU. Intel does not allow this, it is limited.

I'm guessing Apple is working more on it's own SoC designs for future Macs more then they are working on switching to AMD.

With AMD Apple could design a custom APU that as Zen cores and Axx ARM cpu on the same SoC. Unless Intel grants Apple x86 license, there's no other way to do it. And still Apple would need x86-64 license from AMD.
[doublepost=1508836151][/doublepost]

Intel SSDs only and a F*** raid key.

And can't have dual GPU and this together without hitting the PCIe wall.

 

[-hh]

My understanding which I talked about last year and which was kind of mentioned in the Mac Pro interview is that they did try engineering new can Mac Pros. So engineering was working with a design that was forced on them.

Failure to 'bake in' the capabilities for future changes is a major strategic failure of the original 2013 design team and their leadership in particular. Period, full stop.

When it became clear that the pretty looking box wasn't going to work...

Well, with a bit of searching of my old posts on MR, I find this post from 20 Dec 2013:

...Once again, the question will come back to what their trade-off decisions were...personally, I don't like the engineering implications of a hard constraint on future capability growth --> it invariably becomes a cost driver.

I don't want to sound too smug, but saw it and called it before the tcMP even shipped...

[doublepost=1508787806][/doublepost]

Well, millions sounds like a lot, but in Silicon Valley money that's not much. Each engineer is probably a quarter million a year, senior engineers more. Executives even more (and you know Ive got involved). And that doesn't include materials, sample runs, etc. An 8 person peripheral team working on nothing but keyboards and mice would probably fit into that budget.

Agreed; my point was merely that they clearly had "adequate" Engineering staff who was expending resources on the sound of a mouse click ... but apparently not on this. As such, all of the apologetically-angled claims that Apple can only do "so much" (etc) are demonstrated as utterly invalid.

And that's without actually running any math. For example, Apple made $8.7B in net profits last quarter, and if they decided to reinvest only 1% of this is as a bump for their R&D teams, the mathematical implications are that they can thus "afford" to hire +4 new Engineers **per day**, every day, for the entire quarter.

If I had to take a wild guess, I'd guess that the new Mac Pro will probably be at least $10 million in R&D.

Similarly, using this "1% of Net Profits" paradigm to find the way to pay for a $10M project means that it requires two weeks worth of money out of that quarter ... assuming, of course, that the $10M is given an extra 40% ($4M) in management reserves for risk reduction, schedule acceleration, etc.

More seriously, there is the facilitation of a (presumably new) production line to pay for too. Figure a $500/unit "tax" on the retail price, times 100K units and you have a $50M budget to cover your NRE here...and you'll only need to exceed that sales volume by +20K units to "pay back" the $10M R&D cost.

-hh

 

[ventricle]

I don't want to sound too smug, but saw it and called it before the tcMP even shipped...

You and many others, and rightly so! One very important group missed it though, the OEM. You’d think they might have avoided that by something very basic, like asking the market, or product testing. The hubris is incredible. It’s not their only miss too, which is why the platform will always be second in market share. Very profitable though, no one can argue with that. So far…

 

[William Payne]

Market share comparisons between Apple and Microsoft is a bit awkward to do. Apple have always been a computer company. They build computers. (Hardware)

Microsoft weren't a hardware company, they until recently were never a computer hardware company. They just made operating systems. Microsoft needed to sell their OS and not having any computers of their own. They had to make it available to the hardware makers.

 

[vailr]

Apple's system designers don't really have any urgent need to create something completely "out of the box" for the new 2018 MacPro. Instead, they need to focus on "making the box better".
Again: Why re-invent the wheel? HP's latest Z8 workstation would be a completely adequate design, and no need for Mr. Ive to spend millions of dollars on "beautification" research.
HP could start cranking out 2018 MacPros before the end of 2017, from the same production line that currently makes the HP Z8 workstation.
Apple's peculiar EFI bios firmware being the only minor design decision to contend with.
[doublepost=1508869166][/doublepost]

Microsoft weren't a hardware company, they until recently were never a computer hardware company. They just made operating systems.

Since before Windows 95 was released, Microsoft has always offered their own brand of mice and keyboards, among other items. And there's MSN for dialup internet, which is also something other than an "operating system".

 

[AidenShaw]

Apple's peculiar EFI bios firmware being the only minor design decision to contend with.

Apple just needs to adopt UEFI....

 

[Joe The Dragon]

Even if the designs are finalized, they've still got a ton of work to do to build all the machines and production lines. The iMac has had the same tooling for a while now, and the MacBooks just got a refresh that took a while because of the new hardware, these Mac Pros are really going to be from scratch hardware wise, and built out of much bigger pieces than something like iPhones (Apple spends millions just making sure iPhone production gets up running quicker than any project of that scale).


EDIT: Switching to AMD is probably out of the question for Apple, the cost of switching things to work in another x86 architecture and then supporting both legacy Intel and AMD on their own nearly legacy platform doesn't make financial sense. The most we can hope for is lower prices from Intel, since they're not that far behind AMD except in the price/performance ratio.

AMD just needs the drivers. Windows and linux both do amd and intel with the same installers and base kernels

 

[fuchsdh]

Apple just needs to adopt UEFI....

If they were, I imagine it would have happened with High Sierra and eGPU support.

 

[goMac]

Apple just needs to adopt UEFI....

I was under the impression that newer Macs have adopted the GOP EFI protocol which should be able to do boot screens on PC cards...

The problem is there is no new Mac Pro with a PCIe slot.
[doublepost=1508883100][/doublepost]

More seriously, there is the facilitation of a (presumably new) production line to pay for too. Figure a $500/unit "tax" on the retail price, times 100K units and you have a $50M budget to cover your NRE here...and you'll only need to exceed that sales volume by +20K units to "pay back" the $10M R&D cost.

The old Mac Pro also paid for itself easily because they could go 2-3 years on a board layout. The upgrades from a 4,1->5,1 and 1,1->2,1 were basically free. The 3,1 wasn't even an elaborate change over the 2,1.

When Apple talks about modular, that's a big thing they mean. They can do upgrades quickly without putting much investment in.

(Also why the cMP was continually stuck on SATA2.)