ARM-based Macs seen as 'inevitable,' but Apple unlikely to switch anytime soon - Page 3

I can see it in a small OSX powered tablet device. Oh wait. iPad with a different GUI layer. Never Mind!

 

Well I'm stumped, who on earth would want an ARM Macintosh device? Even the most bubble-headed of users require some form on Intel x86 application; even if its something as trivial as Microsoft Word or Skype.

 

Plus, Apple machines can run full-fat Windows and Apple probably sell quite a bit of hardware with that as one of the selling points. I know WindowsRT is the ARM version of Windows8 but its stored in ROM and cannot be purchased seperatly; so it begs the question: why would apple do this?

 

In a simmilar thread I said something along the lines of using the ARM chip as a co-processor would be a great idea. All background services run on that and leave as much CPU time as possible on the main x86 chip.

The 650m has 384 cores.

http://www.geforce.com/hardware/desktop-gpus/geforce-gt-650m/specifications

Huh? Some of us don't get out much.

 

Apple transitioned the Mac from Motorola/Freescale's 68k processor family to the IBM/Motorola/Freescale PowerPC. Macs running both architectures ran Windows in emulation. PPC-based Macs ran 68k transparently. Apple sold a Power Macintosh/SoftWindows bundle for those who wanted to run Windows on their Macs. The PPC G3 ran 68k-based code in emulation faster than the most recent 68k-based Macs ran it natively. Eventually, VirtualPC replaced SoftWindows as the x86 emulation software of choice.

 

During Apple's most recent transition--the one from the PPC to Intel x86, again Intel-based Macs ran PPC code transparently. Intel-based Macs run Windows software in a virtual environment like Parallels natively with the aid of Boot Camp. With WINE or CrossOver, many Windows titles run as Mac applications.

 

And Windows? Without question, there are people who require Windows. However, this is not everybody. The Windows of today is not Windows 3.1 or even Windows 7. The Windows of today is Windows 8. Windows 8 is on a pace to make Windows Me look like a roaring success. Apple appears to have decided that the ability to boot my Mac into Windows and run my MacBook Pro as Yet Another Intel Laptop (YAIL) is just not that important anymore.

 

Times change. Apple may be wrong, but I doubt it.

Processor emulation was (and still is) pig dog slow compared to real machines with regards to recent developments - processors from over a decade ago we still cannot emulator correctly (EG: Sony Emotion Engine from the PS2). Getting the full feature set of an x86 chip would be just as difficult if not more so. Dual Booting is a far better option and x86 based Macintosh computers offer this, and this is crucial for a lot of people.

 

Although older versions of OSX had Rosetta (a dynamic binary translator), the instructions it had to translate were from older, less resource intensive software and ran on far more technologically advanced and faster systems than what they were originally designed for (PPC vs Intel Core). But you have to remember that Rosetta was not as compatibale with older software as the 68k emulator was and many software packages did not run correctly. Also, Intel processors include technologies such as SSE4, Intel VT-x and AES. On an ARM system, merely translating x86 binary is not enough and many software packages would fail to run or have a significant performance loss. Virtualisation software like VirtualBox, for example, would not run on ARM processors at all as ARM processors do not have any virtualisation technology, at least right now.

 

WINE and CrossOver NEED an x86 processor to function. They are not binary translators but an API layer where Windows API calls have been reverse engineered and translated on the fly to native OS calls. The actual binary is still x86 and will not run on anything else.

 

From every aspect, it looks like a bad idea in any way, shape, or form; especially when you remember that Macintoshes are considered high performance machines that reflect their price tag.

 

Chips like the AMD C-60 (1.3GHz dual core) are two years old; outdated in technology terms, but run faster than the brand new A6X released last month along with better graphics and all the tech features mentioned above like SSE and on-chip virtualisation.

 

So not only are ARM chips lacking in onchip features, they are one to two years behind the x86 crowd in terms of performance. TDP is currently ARM's only strong point when compared to x86 processors but that may not be the case for much longer; in Q2 2011 the single core (with HT) Atom Z670 was released, has a TDP of 3w, and performed better than the A5X released later in the year by Apple in the iPad 3 (not by much, admittedly - but it would've crippled it if a second physical core had gotten involved).

 

Once the ARM processors match Celerons in performance, only then would I consider them ready for use in a very cheap, bottom end laptop around the £299 price point.

Edited by benanderson89 - 11/8/12 at 7:43am

If I see it correctly the Atom Z670 needs another separate chip SM35 to have feature parity with the A5X or A6X (and I exclude advanced audio processing of the A chips).
Of course no TDP of this chip is given. Another point is that the graphics processor of the Z670 is in fact a PowerVR GPU, so Intel tries to beat the A(n) cores by being an A(n) for the part that dissipates the most.
Nice.
Also, TDP isn't officially defined and it wouldn't surprise me if Intel 'interpreted' it in a way that others will call cheating.
Another nice fact is that the Intel chip set is 10 times as expensive as a comparable one chip ARM solution.

J.

 

Where to begin? Despite having transitioned the Macintosh to new processors about 20 years ago and again in 2006, your position appears to be that it is not possible to do so again. Suffice it to say, I disagree. I will repeat a statement made above: Power Mac G3s ran MC68040 code faster in emulation than Macintosh Quadras ran it natively.

 

As for Intel code, the emulator does not have to emulate every aspect of the Intel processor. It need only emulate every aspect of the processor required by Macs. Lest you forget--and it appears that you have forgotten--MacOS X does not require the special features of the Intel x86 or any other processor.

 

It is unclear why you bother to mention the fact that Rosetta does not run 68k code. This fact undercuts your argument--badly. Only Classic required 68k code emulation. When Apple made the decision to switch its computers to Intel x86, it finalized its decision to drop Classic. A small fraction of the customer base still required Classic, but Apple was willing to accept any lost sales from this group in favor of the potential upside. The upside drove Apple to the No. 1 position among the Planet's corporations.

 

A point about DarWINE. DarWINE was the project to port WINE to PPC-based Macs prior to the Intel transition. The project planned to integrate the Qemu x86-emulator into DarWINE. This work was abandoned after Apple switched to x86.

 

If Apple transitions the Mac line to a non-Intel processor, there will be no issue with with processor emulation for WINE. WINE will use the Mac's built-in x86 emulator. The issue with WINE the new non-Intel Macs will be what is on all supported platforms. WINE is a clone of a subset of Windows APIs and cannot run all Windows applications.

 

Can a Mac based on a non-Intel processor run Windows? Of course. Can it run Windows as well as an Intel-based Mac? Of course not. If Apple drops Intel, then it means that Apple has decided that Windows is just not that important anymore. Therein lies your objection. For Apple predicate its future on the assumption that Windows is passé is too much for some people to bear.

You should have stopped there. Anything is possible, I don't think anybody would disagree with that. The problem is in the results. Actually you are totally wrong here, the emulator has to emulate the processor precisely for user apps. That is where the problem lies, i86 is not clean to emulate at all. Emulation of 68K and PPC was relatively easy but more importantly the processor doing the emulation in each case was significantly faster. Nothing ARM based at this point is significantly faster than Intels offering. Err there is no cause and effect here to justify your position above. Yes and what happened to that bit of software. When it comes right down to it software emulation is a waste of time. All those man hours would have been more productively used writing apps. Effectively DarWine was a big waste of time. Wishful thinking. Wishful in the sense that such an emulator would produce any sort of respectable results. Currently the best emulator of i86 on ARM can only get 40% of the performance of a native ARM app and that is under ideal conditions. It is important to realize that that is relative to a native ARM app.

I have no doubt that Apple is looking at ARM as a possible evolutionary path for the computers it sells but I really don't see emulation as the path they will follow here. They will simply give software authors 60 days or so to get their apps to run native or pull them from the App Store. The alternative would be compiling to LLVM byte code with the final translation on the device. Emulation is really technology from the last century.
No it really has nothing to do with that. It has to do with people solving real problems with the Windows environment with legacy software that doesn't go away just because Apple wants it to. However there is another issue, running an emulator for user land Mac apps is one thing, it is far far more difficult to emulate an Intel chip in way that is suitable for an operating system and the wide range of software that is supported at the OS level.

I'm not saying it is impossible, I'm just saying the results would be so bad on ARM that nobody would want it thus it is a waste of time.

I believe that Apple will release a large screen, the next "iMac", (20in.-27in.) product that will have an ARM based chip.  It will have I/O and limited storage.   This device will connect to an Apple Home Server (also ARM based) via 802.11.ac as well as keyboard and mouse/track-pad via BT.  Data will flow to and from this HomeServer to all devices on the network.  It will back up to iCloud in the background seamlessly. 

 

The platform will utilize next Apple SoC, A7, it will be 64 bit and have the Imagination Series 6 "Rogue" GPU. The A7 will be a "family" of CPUs.  A7 iPhone 2 cores,  iPad 4 cores, ATV 2-4 cores,  iMac and MacBookAirs 8 cores.   The new HomeServer chip would have 8 or 16 cores.   The Apple ARM based platform will satisfy 95% of the needs of the computing world. Apple will however continue to produce Intel based MacPros and MacBookPros (trucks) for content creation.

 

"PCs are going to be like trucks..."

 

—SJ.

@oilerch50 ,

See wizard69's post above.

Much of what you stated is certainly possible but I think very unlikely.

 

ARM can't do desktop processing, Apple is out of the server game, and what's the point of this device? What purpose does it serve that isn't by existing products?

 

 Many here on AI use their MacMini as a server, but I was talking about a Home Server for "mere mortals".  A home server hosting your iTunes library, iPhoto library, with massive storage and Gigabit wireless connection to various screens of varying sizes (iOS/ARM powered devices: iPhone, iPad, ATV, iPadGrande .  I am guessing that most people are reluctant to trust iCloud with all of their data without a local backup.  

 

 

 

 

The paradigm of computing is changing.   95% of consumer computing NEEDS  in the near future could be met by Cortex A50 series CPU with an Imagination Series 6 "Rogue" GPU.  As I said,  Apple will continue to manufacture Intel based Pro computers for professional content producers.  Most households already have a heterogenous computing environment with both Intel and ARM products.  In the future, most (not all) PCs will be replaced with  one of several ARM based displays/devices (thin clients) that are more than capable "computers" for email, surfing the web, simple word processing, social networking, iPhoto and iTunes.  Heavy lifting will be done in the Cloud.  Think XenDesktop. Perhaps DesktopX, if Apple wants to roll its own virtualization solution.  

 

SJ at D8 from CNET news:

 

 

Edited by oilerch50 - 11/26/12 at 9:34pm

What is the compelling reason for an ARM based iMac? That is the question you have to ask and frankly I can think of nothing. Limited storage would kill the machine. Actually I've wondered why Apple doesn't produce a home server of sorts. However I have highly mixed feelings about iCloud. Frankly iCloud sucks worst than iMaps yet everyone complains about a mapping program that works well for me. In the end a media server would be nice if combined with a refactoring of Apples I services. Now there might be a compelling argument for an ARM chip in a laptop, however calling it a Mac would be a huge mistake. They have a rather high hurdle to jump over to call an ARM based laptop a Mac. It isn't an issue of ARM being able to solve user needs, iPad shows that it can for many. It is rather an issue of how you manage user expectations.
For an ARM based device to run Mac OS it would need to be able to reasonably compete with current Mac offerings and more importantly Apple would need a software solution. Software is a huge issue that can't be underestimated.

This is an interesting evolution of computing technology, that is the so called cloud services. Frankly I think the train has left the tracks with all the horrors that implies. I have many issues with iCloud so frankly I've been thinking a bit about a home server. It really looks like Apple has basically said that they are not about to play in that sand box. So it looks like DIY will be the order of the day if you want such a system. Actually it can. The issues here are far less technical and more marketing. How would Apple go about marketing a lower performance computer that isn't i86 compatible? ARM on iPad works because it isn't being marketed as a Mac. This is very significant. Remember Jobs was a marketing man. This seems to cause people all sorts of distress but a salesman can only sell you what he has available today. Jobs was very very good at selling what he had. The post PC era is more about Apple than the industry as a whole. I can make an argument that for most consumers the PC was never the ideal solution for their needs anyways.

He was legendary for convincing an audience that anything without a fruit logo was a piece of crap. It amused me. I really like a number of Apple's products, but the instilled mindset always seemed so silly. I used to imagine their headquarters as being filled with people wearing black turtlenecks.

Funny I never imagined Apples headquarters that way, rather I imagined a counter culture atmosphere. Steve could pull the proverbial wool over the publics eyes. I think his best sakes job came from the introduction of the G5 which was an extremely slow machine doing anything that wasn't supported by Alt-Vec or the units float performance. People lapped it up though.

Don't get me wrong most companies need a good salesman. But even at this late date you still find people that take things a bit too seriously. IPad Mini being a perfect example of people misunderstanding what Steve was saying and more importantly selling.

Well regardless of how you interpreted his statement on 7" tablets, he would have most likely worded it differently if the ipad mini was about to be released. I still find the idea of buildings populated by a sea of turtlenecks to be extremely funny. Sometimes I just imagine things in ways that amuse me.