Concept: Mac Mini stacks, the 2013 Mac Pro

[Umbongo]

I'm someone who would love an xMac. I would not expect it to be the same cost as building it myself, just somewhat cheaper than the iMac as it doesn't have the display costs. The reasons I like the idea are:

• Much better thermal constraints so that desktop components can be used throughout
• Upgradability after purchase
• Ability to easily replace failed components
• Choice of monitors, which was very important when the iMacs had extremely glossy screens, but is hopefully not a problem with the new model
• Be able to keep monitors for use with replacement computers

The Mac Pro is an excellent machine, but it is also very expensive. I don't need server grade components.

Right, but all you've described is the same as having a cheaper Mac Pro. The level of hardware the base model Mac Pro has can be found elsewhere for roughly half the price, just usually with a slightly weaker PSU and cheaper enclosure.

 

[adam9c1]

When they said they are working on a completely different new product I thought this:

Stackable
Modular
However without external cables (all boxes have a dock on the bottom and on top to pass through data and power)

Power supply is modular as well, external, you add modules like add batteries to big ups.

 

[xgman]

When they said they are working on a completely different new product I thought this:

.

No one actually said anything to that effect.

 

[TableSyrup]

I'll take two of these, rooms, please

 

[xav8tor]

I'll take two of these, rooms, please

Me too.

OP...that concept is NUTS for serious workstation users.

 

[scottrichardson]

I gather from the discussion, and my own logical deduction, that stacking CPU modules just isn't going to happen - even if they double, triple, or quadruple the bandwidth of Thunderbolt (not all that unrealistic), there's the distance between the modules and the introduced latency to consider.

What I think would be better is YES, have a modular system, however the CPU/RAM module is ALWAYS contained within the one unit.

Let's pretend Thunderbolt 2.0 is announced doubling the bandwidth, allowing for fully fledged GPU's.

So you can buy:

- Logic Board Module (custom interconnect with the following CPU/RAM modules)

- CPU/RAM Module with one of the following..
- 1 x Quad Core CPU, 1 x empty CPU slot
- 2 x Quad Core CPU
- 1 x Hex Core CPU, 1 x empty CPU slot
- 2 x Hex Core CPU
- 1 x Eight Core CPU, 1 x empty CPU slot
- 2 x Eight Core CPU

- PCI Expansion Module with one of the following...
- 1 x x16 PCI Slot
- 2 x x16 PCI Slots
- 2 x x8 PCI Slots
- 3 x x8 PCI Slots
- 1 x x16 PCI Slot, 3 x x8 PCI Slots

- Storage Module with one of the following...
- 2 x 2.5" Drive Bays
- 2 x 2.5" Drive Bays, 2 x 3.5" Drive Bays
- 2 x 2.5" Drive Bays, 4 x 3.5" Drive Bays
- 4 x 2.5" Drive Bays, including hardware RAID controller
- 4 x 3.5" Drive Bays, including hardware RAID controller
- etc

- Connectivity Module with one of the following...
- 4 x USB 3, 2 x gigabit Ethernet, 2 x Firewire 800
- 4 x USB 3, 2 x gigabit Ethernet, 2 x eSATA
- 6 x USB 3, 2 x 10GigE Ethernet, 2 x eSATA

Etc etc etc..

May even be able to combine the logic board module with the RAM/CPU one etc. EVERY module would have TWO thunderbolt 2.0 connectors (or more).

Thoughts?

----------

Further to my post - imagine the 3rd party accessories market? If there was a consistent size/shape for modules, 3rd parties would go nuts making their own modules.

 

[FluJunkie]

Probably not so far fetched. All they need to do is enable thunderbolt for the grid computing stuff built into OS X.

"The grid computing stuff built into OS X" is something Apple has categorically abandoned.

it is possible to cluster compute with mac minis:
https://discussions.apple.com/thread/3976865?start=0&tstart=0

It's possible to cluster compute with Raspberry Pis. It doesn't mean its a good idea. Beyond that, cluster computing is not a be-all, end-all substitute for workstations - they are for example, terrible for high RAM, clock-intensive serial jobs.

 

[Neodym]

"The grid computing stuff built into OS X" is something Apple has categorically abandoned.

Source?

It's possible to cluster compute with Raspberry Pis. It doesn't mean its a good idea. Beyond that, cluster computing is not a be-all, end-all substitute for workstations - they are for example, terrible for high RAM, clock-intensive serial jobs.

Question is whether Apple considers the "classic workstation" a dwindling market and thus maybe something that can be left to competitors like e.g. Dell or HP.

Apple has a good track record in innovating whole market segments and consumer products. Before the iPad people thought that a tablet would never work in the mass market - as some companies had already tried and failed. Similar the iPhone - touch interface? No way!

They might be able to do something similar to the workstation market, in whatever way. Perhaps by finding a way to automatically parallelize tasks considered purely "serial jobs" today. Or by offering so much performance advantages for massively parallelized software that programmers may feel compelled to re-think their code and earn the benefits from the massively multi-core/CPU architecture.

Hardware-wise it has been proven that you can build a high-performance machine out of inexpensive standard parts - think of the famous "ASCI Red".

If Apple would bring such a concept to the desktop, then ARM would probably the CPU of choice: Development there is currently much faster than on the x86 platform, Apple has the expertise in-house (and is independent from external suppliers and their roadmaps) and can leverage the huge gains from economies of scale, as those CPU's are already produced in the millions for their iOS devices.

 

[ScottishCaptain]

Source?

https://forums.macrumors.com/threads/1412900/

CLANG/LLVM doesn't even support OpenMPI. There is absolutely no "grid computing stuff" built into Xcode or Mac OS X anymore, Apple has abandoned all of it. If you want that sort of stuff, you'll be looking at using a third party solution for the platform AND having to install and maintain your own installation of GCC or another compiler.

-SC

 

[FluJunkie]

Source?

Already answered below.

Question is whether Apple considers the "classic workstation" a dwindling market and thus maybe something that can be left to competitors like e.g. Dell or HP.

I'd put the odds of them thinking the classic workstation market - with its strongholds in video and image production dead - and abandoning a product line where most of the heavy lifting is done by Intel with very little needed R&D budget while simultaneously launching a "desktop cluster" mini stack that would require the raw development of several new technologies to be...pretty damned slim.

They might be able to do something similar to the workstation market, in whatever way. Perhaps by finding a way to automatically parallelize tasks considered purely "serial jobs" today. Or by offering so much performance advantages for massively parallelized software that programmers may feel compelled to re-think their code and earn the benefits from the massively multi-core/CPU architecture.

The problem is not that there are some programs where the programmers have decided not to parallelize. There are simply some tasks that are inherently serial - and many of these tasks are what people use workstations for.

Hardware-wise it has been proven that you can build a high-performance machine out of inexpensive standard parts - think of the famous "ASCI Red".

If Apple would bring such a concept to the desktop, then ARM would probably the CPU of choice: Development there is currently much faster than on the x86 platform, Apple has the expertise in-house (and is independent from external suppliers and their roadmaps) and can leverage the huge gains from economies of scale, as those CPU's are already produced in the millions for their iOS devices.

I very much doubt Apple's economies of scale in producing iPhone chips isn't one that's matched by Intel producing x86 chips. Beyond that, while ARM chips are outstanding for low-power, moderate-performance applications like tablets and smart phones, trying to put an ARM chip in an industrial grade workstation is actively deciding to go with inferior parts. And having to deal with myriad incompatibilities.

They just left the lingering effects of switching processor families (from PPC to Intel) behind. I can't imagine them being eager to go for Round 2.

 

[ActionableMango]

Thoughts?

Shoot me now, before I ever have to see such a thing.

 

[Neodym]

Judging from the latest rumor I may be right with my speculation about Apple considering ARM chips for desktops as well as for mobile devices. Perhaps some people could be in for an unwanted surprise when the "great something" Tim Cook talked about is NOT simply an upgraded MP with Sandy or Ivy...

Could become an interesting year 2013.

 

[ActionableMango]

Judging from the latest rumor I may be right with my speculation about Apple considering ARM chips for desktops as well as for mobile devices. Perhaps some people could be in for an unwanted surprise when the "great something" Tim Cook talked about is NOT simply an upgraded MP with Sandy or Ivy...

Could become an interesting year 2013.

I have to think that's a ways off. There weren't even 64-bit ARM chips until a few days ago.

 

[ScottishCaptain]

Why would anyone want an ARM based Macintosh?

Do people not realize that x86 extends beyond the CPU socket? There's ACPI, a whole variety of BIOS systems (legacy, EFI/UEFI, CoreBoot, OpenBoot, etc). Everything is standardized in such a way that a wide variety of software can run on a slightly different platforms without an issue.

ARM has none of that following. The spec ends at the CPU and ISA. There are no platform configuration systems, there is no common firmware model. There is nothing but proprietary crap implemented by the manufacture of that particular device. Why is this a good thing?

Your "ARM" based Mac system won't boot anything other then OS X. It probably won't load any applications other then those bought from the Mac App Store. Everything is going to be signed and encrypted, so good luck changing the things Apple doesn't think you should be able to change. Hell, I doubt an ARM version of OS X would even come with Terminal.app.

Nothing good can come from an ARM based Macintosh. If Apple were to produce such a system, it wouldn't be because ARM is better- it's not. They'd be doing it because it might be profitable, and for no other reason (as if they need more money). ARM belongs in the palm of your hand where it makes sense, and nowhere else.

-SC

 

[scottrichardson]

Shoot me now, before I ever have to see such a thing.

Yeah, in all seriousness, I really wouldn't go for this myself LOL, but it was nice to hypothesise.

 

[Neodym]

Why would anyone want an ARM based Macintosh?

Apple would want it to
* have better control over the whole package and be independent of other's roadmaps plus allow for a better security
* distinguish themselves from being "just another Intel box at higher prices"
* bring together iOS and OSX to have only one hardware platform to develop for
* leverage economies of scale to either bring down prices and/or net a better profit from their products

Customers would want it to
* have lower prices
* be on the forefront of technology
* distinguish themselves from the Wintel boxes out there
* get access to latest software optimized for ARM
* not get left behind when Intel loses importance as even Windows is going ARM

Do people not realize that x86 extends beyond the CPU socket? There's ACPI, a whole variety of BIOS systems (legacy, EFI/UEFI, CoreBoot, OpenBoot, etc). Everything is standardized in such a way that a wide variety of software can run on a slightly different platforms without an issue.

If everything would be as standardized as you say, then why is there such a "whole variety of BIOS systems"? Besides - things like ACPI can be adapted to another CPU platform such as ARM.

ARM has none of that following.

Maybe not yet, but it surely is already under evaluation, if not development.

The spec ends at the CPU and ISA. There are no platform configuration systems, there is no common firmware model. There is nothing but proprietary crap implemented by the manufacture of that particular device. Why is this a good thing?

First of all - x86 is as proprietary as ARM. Only that x86 is still dominating the major part of the desktop and mobile computing market. Currently ARM is still perceived in the mobile niche, but that can change. Easily!

Development on ARM is going strong, while x86 with all its inherent shortcomings sees a harder time to bring improvements over the predecessor. Frequency is nearly topped by now around 3-4GHz - one reason to go multicore. And going multicore requires a new thinking and new tools anyway. In that case, why not make the jump and go completely ARM? The user base is strong with millions and millions of mobile devices already using ARM.

Such a move would also offer chances to improve on limitations x86 is currently having.

And most of all: If the market is about to swing to another major hardware platform (and all signs seem to indicate that such a swing may be imminent) any company will try to be the one to define the results and set the standards, as it would be able to gain huge benefits from that (see what Apple did in the mobile music player, smartphone and tablet market).

Your "ARM" based Mac system won't boot anything other then OS X.

That's fully in Apple interest. You may not want it as user, but as a company you want to position yourself with as little competition as possible.

It probably won't load any applications other then those bought from the Mac App Store.

That's got nothing to do with the underlying hardware, but with the policies introduced with the OS. Apple can do this within OSX on Intel as well as on ARM.

Everything is going to be signed and encrypted, so good luck changing the things Apple doesn't think you should be able to change. Hell, I doubt an ARM version of OS X would even come with Terminal.app.

That would be absolutely in line with Apple's policy over the recent years --> "walled garden".

Nothing good can come from an ARM based Macintosh.

And you are the one to be able to judge this exactly why?

If Apple were to produce such a system, it wouldn't be because ARM is better- it's not.

How would you know? Are you working in Apple's labs? Do you know how an ARM cluster of - say - 4-8 A6(X) chips clustered would perform against a core i5/i7 CPU of today?

Even a single chip in the current iPhones and iPads offers a performance that was available in iMacs only a couple of years ago. And that design is already 1-2 years old...

They'd be doing it because it might be profitable, and for no other reason (as if they need more money).

Of course profitabilty is one major goal. Apple is a company, not a charity! They got bills to pay, stockholders to please and stand their ground in a market environment constantly growing more difficult.

ARM belongs in the palm of your hand where it makes sense, and nowhere else.

Yeah right. All the engineers and managers at Apple and Microsoft have no clue about where the market is going to and what performance an ARM-based Macintosh would be able to deliver. Good for us that you know better and have good arguments to prove it. Oh wait...

 

[ssgbryan]

Neodym, your post is so full of fail, I don't know where to begin. But I will start here:

Apple would want it to
* have better control over the whole package and be independent of other's roadmaps plus allow for a better security
* distinguish themselves from being "just another Intel box at higher prices"
* bring together iOS and OSX to have only one hardware platform to develop for
* leverage economies of scale to either bring down prices and/or net a better profit from their products

Customers would want it to
* have lower prices
* be on the forefront of technology
* distinguish themselves from the Wintel boxes out there
* get access to latest software optimized for ARM
* not get left behind when Intel loses importance as even Windows is going ARM

Better control? Do you have any idea how much a fab plant costs? It is still cheaper to contract that out - Adjust the NDA and Apple can save a few billion dollars.

distinguish themselves from the Wintel boxes out there? You mean like having no software available from the get-go - they saw that already when Apple was powered by Motorola - Been there, done that. Apple computers are already distinguished - "laptop on a stick"

We can always depend on Sir Idiot-Boy do design a super-slim box that cooks the internal components and pushes them to premature failure due to heat (As I glance over to a pile of apple kit that cooked itself to death.)

Merge iOS & OSX - When did they become separate operating systems? - iOS was always a subset of OSX - the GUI's are what are different.

Lower prices - HAHAHAHHAAHAHAHAHAHAHA - Tim Cook will have the choice between lowering the price or pocketing the additional profits - which do YOU think he will do.

Latest software optimized for ARM? - And just how many versions of software (and years of waiting) do you think that will take?

Can you name ANY piece of software that jumped architectures, while at the same time fully integrated the capabilities of the new architecture and was just as fast as the other version?

Good god, did you not see how long it took to move from PPC to Intel? In 2012, I am just NOW getting software that is feature compatible with the Win86 versions. And we are STILL not leveraging all of the capabilities of the X86 chip.

For that matter, I am still running 10.6.8 because I have software that needs Rosetta - And no, there isn't any alternate software available to take the place of those apps. If there was, I would have bought it. And the software developer has made it clear they aren't porting it to intel (or anything else for that matter - not enough sales).

Depending on the developers who gave us the iFart isn't a winning strategy.

be on the forefront of technology i.e. go buy all your software again - the fanbois will line up for that, but most people won't, unless all of the software is in the $4.99 category. Do you think anyone that buys an iMac or Mini is worried about being on the forefront of technology - put the crack pipe down.

even Windows is going ARM If you were around during the Windows NT days (like I was), you would know that windows was available on several platforms (MIPS, Alpha, PowerPC, ARM, Itanium) They all went under without a trace - what has changed since then?

You are betting that everyone will seamlessly convert their software, no muss, no fuss -

It didn't happen when we went from real-mode operating systems to protected mode operating systems.

It didn't happen when we went from 16-bit to 32-bit computing.

It isn't happening NOW, as we move from 32-bit computing to 64-bit computing.

It didn't happen when Apple jumped from the 6800 to the PowerPC chip and it certainly didn't happen going from PowerPC to Intel.

It didn't happen when we went from single-threaded apps to multi-threaded apps - software developers are still working on this. Multi-threaded apps were possible when the 386 came out - OS/2 had that capability - hell, I had more fully multi-threaded apps 20 years ago than I have today.

We have a better chance of seeing the Rapture than seeing a seamless transition to ARM.

 

[Neodym]

Neodym, your post is so full of fail, I don't know where to begin. But I will start here:

I wonder if you even bothered to actually _read_ my post before projecting your Apple-based hate on me...

Better control? Do you have any idea how much a fab plant costs? It is still cheaper to contract that out - Adjust the NDA and Apple can save a few billion dollars.

Where did i write that Apple needs its own fab? And even you have to agree that it makes a difference between ordering exactly the piece of silicon that you want vs. having to accept what the supplier is offering to you (not counting small tweaks like early die-shrink as happened in the original MBA).

distinguish themselves from the Wintel boxes out there? You mean like having no software available from the get-go - they saw that already when Apple was powered by Motorola - Been there, done that. Apple computers are already distinguished - "laptop on a stick"

In case you didn't notice (go ahead and browse through e.g. the MacPro forum here at MacRumors) - Apple received a lot of flak (not only lately) for its high prices of basically standard PC hardware with a different OS and a nice casing. People are not willing anymore to pay the Apple tax (which increases by the day as Apple refuses to lower prices as hardware grows older) when they can easily compare offers with similar or even identical hardware from e.g. Dell or HP.

We can always depend on Sir Idiot-Boy do design a super-slim box that cooks the internal components and pushes them to premature failure due to heat (As I glance over to a pile of apple kit that cooked itself to death.)

Help me out here - how is that related to the question of a possible hardware platform migration to ARM?

Merge iOS & OSX - When did they become separate operating systems? - iOS was always a subset of OSX - the GUI's are what are different.

How do you think the code is working on two different hardware platforms like ARM (iOS) and Intel (OSX)? A developer may only need to set a checkmark in the compiler, but underneath there is a lot of programming work to make it seem to work automagically, when in fact it's not!

Lower prices - HAHAHAHHAAHAHAHAHAHAHA - Tim Cook will have the choice between lowering the price or pocketing the additional profits - which do YOU think he will do.

That's why i wrote _both_ possibilities. But even Apple isn't living alone on the island of happiness and has to consider competition for pricing its own products. And demanding a higher price is easier if the customer can not directly cross-check the hardware prices when purchased from other sources than Apple.

Latest software optimized for ARM? - And just how many versions of software (and years of waiting) do you think that will take?

How many PPC-optimized (or at least PPC-compatible) programs did you come across recently? If the mass market transitions away from Intel towards e.g. ARM, sooner or later you will see software offered for ARM only. Perhaps the new architecture will offer certain benefits simply not available on x86 (like e.g. easy massively-multicore aware applications), which would increase the probability and lessen the time needed for this situation to appear.

Can you name ANY piece of software that jumped architectures, while at the same time fully integrated the capabilities of the new architecture and was just as fast as the other version?

What for? I don't claim such things. Surely it won't happen overnight, but once people get used to the new platform they will never look back (except a few power-users and die-hards). Especially if the new platform should offer some significant advantages over what we have today: x86 with its inherent shortcomings, low core-count and slow development when compared to ARM.

Good god, did you not see how long it took to move from PPC to Intel? In 2012, I am just NOW getting software that is feature compatible with the Win86 versions. And we are STILL not leveraging all of the capabilities of the X86 chip.

When Apple transitioned away from PPC, many people were angry just like you and predicted the end of Apple and the world as we know it.

Nowadays the very same people claim the end of Apple and the world as we know it if Apple should leave the once hated Intel platform, because this is absolutely the best since the invention of sliced bread.

Same procedure as last time... YAWN!

For that matter, I am still running 10.6.8 because I have software that needs Rosetta - And no, there isn't any alternate software available to take the place of those apps. If there was, I would have bought it. And the software developer has made it clear they aren't porting it to intel (or anything else for that matter - not enough sales).

Let me get this straight - a small software developer (at least in comparison to Apple) refuses to port his product due to the lack of sales, but you expect Apple with its VASTLY bigger user base to ignore the world going on and to support a few die-hards instead? Errr - let me borrow a small part of your own post: "HAHAHAHHAAHAHAHAHAHAHA"!

Depending on the developers who gave us the iFart isn't a winning strategy.

I've got news for you: You're an absolute minority and not the benchmark for Apple's business decisions...

be on the forefront of technology i.e. go buy all your software again - the fanbois will line up for that, but most people won't, unless all of the software is in the $4.99 category. Do you think anyone that buys an iMac or Mini is worried about being on the forefront of technology - put the crack pipe down.

Running out of arguments that you have to resort to personal attacks already? Outside of your little hate-filled world not only fanbois make conscious decisions to either stay with their current setup (both hard- and software) or to accept the fact that the world keeps turning and new options arise.

No one forces you to upgrade. Especially the Mac Pro is a very solid product being able to last for years. Stick with what you have and be happy. If you feel the performance is not sufficient anymore, you have to invest - simple as that! No matter which hardware, which OS or which workflow.

even Windows is going ARM If you were around during the Windows NT days (like I was),

I have been around long before that poor excuse of a proper OS called Windows NT.

you would know that windows was available on several platforms (MIPS, Alpha, PowerPC, ARM, Itanium) They all went under without a trace - what has changed since then?

I don't know how that is related to the original topic - maybe you could explain it to me to allow me a proper answer.

You are betting that everyone will seamlessly convert their software, no muss, no fuss -

I'm never gambling. I don't know where you read this nonsense into my post.

We have a better chance of seeing the Rapture than seeing a seamless transition to ARM.

I don't know which posting you are referring to, but i never predicted a _seamless_ transition! A transition like going from Intel to ARM will never be seamless - it is inherently disruptive!

 

[robogobo]

With Apple's "skate to where the puck is going" approach, it's doubtful they're looking to the current needs (read: wants) of "pros" in terms of PCI expandability and dozens-core processors contained within a huge enclosure. The majority of Mac Pro purchasers are buying more machine than they need (graphic designers and freelance photographers, I'm looking at you) perhaps for status, but also due to a huge performance gap between the current MP and the iMac, which lacks expandability, doesn't handle custom monitor configurations very well and, up until recently didn't offer the CPU and GPU power to run pro applications. These people are using these workstations for five years or longer for production tasks, which is well beyond their intended service life. They're an amazing value in those cases, but that's not at all what Apple thinks about when building the next generation. They'd much rather make 3 $1500 sales over six years than a single $3000 sale. Not only will they net more cash, but they'll also lower the frequency of complaints over dropped OS support for their aging speedy dinosaur.

The big picture suggests that the puck is going toward a modular storage scenario, but I don't think cluster desktop computing is quite ready for prime time. It would be better to redesign the case to send storage and peripherals external via thunderbolt enclosures and offer a range of CPU, GPU and RAM configurations in a smaller package to fill the gap for pros and wannabe pros who think an iMac is beneath them for whatever reason. I suppose this is the xMac concept.

If Apple redesigns the case, all bets are off. Will they continue to feed the thin and beautiful beast that brought us the new generation of iMac and MacBook Pro? Most likely. Look for hybrid drives and thunderbolt accessories, dropped support for old tech and a handful of pissed off customers - namely those who really, really needed the Mac Pros of yesterday. They're the ones who will be left out in the cold, and run off to join the rest of the industry who have already found warm, cozy solace in the arms of Dell and the like. The future for production houses will be custom black box workstations with all the expandability they could ever desire.

 

[calaverasgrande]

I think it could work if you had a base with some kind of peripheral interconnect that carried power as well as 4 or more TB ports. Possibly the optical spec of TB may yield higher data rate?
But then I think it is more likely the Apple will go with the simplest looking design with the least visible ports and jack. EG: why does the new imac/Mac Mini lose the optical, and put the SD card slot on back?

 

[MacSince1990]

Your "ARM" based Mac system won't boot anything other then OS X. It probably won't load any applications other then those bought from the Mac App Store. Everything is going to be signed and encrypted, so good luck changing the things Apple doesn't think you should be able to change. Hell, I doubt an ARM version of OS X would even come with Terminal.app.

Nothing good can come from an ARM based Macintosh. If Apple were to produce such a system, it wouldn't be because ARM is better- it's not. They'd be doing it because it might be profitable, and for no other reason (as if they need more money). ARM belongs in the palm of your hand where it makes sense, and nowhere else.

-SC

You've just described exactly why it might happen..