Apple wants to move Macs away from Intel chips - report - Page 5

Seems like we're going to have the UltraMac: superthin, 13", 15" & 17" multitouch retina screen, A10 perhaps hahaha.

Moving from Intel to AMD doesn't make sense unless Apple licences the AMD tech in the way it does ARM chips. Apple is able to take the ARM chips (Apple once owned a percentage of ARM) and change/improve on the design through it own in-house chip-design facility which employs some brilliant minds (these people must be frustrated knowing that they can produce better silicon than what is out there).

 

Intel and Apple do not share the same roadmap. Just like Flash allowed Adobe some control over Mac OS, Intel has control over what CPU goes into Macs. And Apple (or Steve Jobs) hate being held back by the non-development of better and faster solutions.

 

Apple ported OSX way before the move to Intel chips, and I'm sure a lot of the rumours then were actually leaked by Apple to test the waters. The biggest challenge Apple faced at the time was how to prevent a quick and easy installation of OSX on cheap WinPCs (hackintosh is too much hassle for the average computer user) -- a move to in-house designed CPUs would allow Apple to breathe easier.

Good point. When I said that there's no way Apple would replace Intel with an alternative, I was referring specifically to ARM which is the subject of this rumor. That's not going to happen - no way will they take the massive performance hit.

It is, as you point out, possible that they could switch to AMD for some or all of their systems without the downside of switching to ARM. I'm not familiar enough with AMD's current offerings to know how plausible that is or what sacrifices might need to be made, but it's far more plausible than switching to ARM.

The platform was a 3-way alliance as a co-development. It did have roots in the POWER architecture though.

Motorola made the G3 and G4 series.
IBM made the G5 series.
I think they both shared designs in the 60x series.

Both of these companies dropped the ball and that held back the Mac platform. Motorola didn't want to make the performance chips and IBM couldn't or wouldn't make a G5 efficient enough for notebook use. I think PA Semi probably came along a bit too late. They may have had some other drawback that wasn't well-publicized.

Intel has a great line of future processors... if apple is really thinking of getting rid of them they will have to make their own processors for their macs. Why waste time and money into something that's already being built by someone else... just laser an apple logo on the processor. And if they think that intel processors are crappy, it might be that apple wants to stop computer failures due to the fact that they don't exactly know how intel processors work like they do with their own custom A chips on iDevices.

Still incorrect ...

IBM also made G3 and G4 processors. At the early days of the G4 Moto couldn't make enough processors, so IBM helped out.

And its also a unknown secret that Apple wouldn't pay for a POWER5L und POWER6L and 6UL design, so they went to Intel.

At these days IBM is the king of the hill in design and manufacturing and always was.

Almost everyone seems to be ignoring that the article itself says "isn't' imminent" on this, which is tech-speak for "not in the next five years", given that this is how far out they look.

 

It's not happening next year or the year after. ARM sucks right now. It can't possibly provide what Intel does in the computer realm. 

 

 

...aren't AMD getting into 'low' power in a more serious way?  *(Seeing as they can't hurt Intel, at the moment in a 'head to head' performance battle?)

 

Maybe becoming an ARM cpu maker..?

 

The phone and tablet market are...huge...

 

Lemon Bon Bon.

 

Of course it's not happening tomorrow...or next year or the year after...

 

Not in the next five years?  I'll guess we'll see.  But there's a lot of pieces lining up on the chess board.

 

Arm doesn't suck for phones or tablets though.  And that's why Apple are using them?  Designing their own chips based on them..?

 

In the way that laptops have chased down desktops in terms of performance...tablets will chase down laptops in terms of performance too.

 

How far behind?  And for what apps?  How many years will it take?  1?  2?  5?  10?

 

Lemon Bon Bon.

Lots of things make me doubt this happening in the next decade. The investment it would take to build a high performance chip like Intels would make the A6 seem tiny, desktop and laptop chips are still over an order of magnitude ahead of ARM chips. And Apple cannot use x86 instructions in their own chips anyways, the licence is exclusive to Intel, AMD, and Via. Apart from that, the patent portfolios of AMD and Intel make it nearly impossible to build an x86 of such performance even if they COULD use the ISA. Apple has a lot of sway with Intel, the relationship is already very beneficial to both, I cannot see this happening. It's not like the move from PowerPC to x86, as PowerPC was falling behind on efficiency at the time and had no viable mobile CPUs in the G5 architecture, while Intels roadmap continues to be more mobile oriented with Haswell and future chips. Calling it now, not gonna happen.

On the other hand, if Apple bought out AMD...That could be interesting. Although I'd hope they would let AMD continue to sell chips in the non-mac space so Intel doesn't become a monopoly.
Edited by tipoo - 10/4/12 at 9:00am

Depends on how you see things, I mean it takes two Cortex A9 cores with good useage on both to beat a single Atom core clock for clock, and that's ignoring dual core Atoms and the upcoming architecture change.

Even the link from the article shows the single core smartphone variant of Atom coming in second for Javascript, although that's also a test of software.


Edited by tipoo - 10/4/12 at 9:07am

Thunderbolt alternative, which isn't really Thunderbolt at all.

If it's just an incompatible offshoot, then it has even less chance of getting traction than Thunderbolt has.

Since someone mentioned POWER chips earlier I have a question:

 

Has Apple compiled OS X to run on POWER chips and if so how did it perform?

 

Gene King

 

I disagree. It could be a very very great idea. 

 

but not at this time. Those A5, A6 etc chips that 'sources' are claiming would be how Apple wants to go are far from even close to ready for such a move. They are great in iPhones and iPads but unless Apple really intends to drop all possible usefulness for Pro users, cancel Final Cut etc, they aren't strong enough for computers by a long shot. Maybe in 8-10 years but not really any sooner. 

 

But even then I don't think they would put them into computers. I think at that point they would be looking at an iPad/iPhone that would be powerful enough to be THE computer for many basic users that need to be portable. No more getting a Macbook Air or 13 inch MBP. Your iPad can do it all, the power would be there, the apps would be there. Folks that were getting iMacs for checking email, playing online bridge etc would dump those for the iPad. Or perhaps keep the computer as a base station (maybe moving to a mac mini on the living room tv for when you need to see something on a display. or perhaps at that point the ipad could have an app that lets it be the display)

Edited by charlituna - 10/4/12 at 12:29pm

 

I don't know a single consumer that really cared about Windows on their Mac. 

 

Now geeks, developers etc (who are maybe 10% of Mac users) sure. 

 

And I'm not so sure about your market share numbers. I think it all depends on how you define a computer and its uses. Depending on that defination, the ipad could qualify and that puts a big dent in Microsoft's market share. Add Android tablets into that and Microsoft is the vast minority. 

 

POWER is just the successor to PowerPC. They're not moving back to IBM.

PowerPC is more of a derivative of POWER. But it's complicated.

I don't see why Apple would go back after closing out that chapter a few years ago anyway.

 

I do.  I was surprised that anyone outside a pro setting would want it, but apparently some do.  My wife is one example.  She wants access to Windows because a lot of software aimed at her hobby is not available for Mac.  I think more people would install it if the cost of a stand-alone version of Windows wasn't so high.

Everyone is moving onto ARM.  I don't know if you guys follow open-source news, but Dell and Ubuntu have already released ARM based Ubuntu servers, and alot of the sub 200 dollar computers already are running various forms of Linux on ARM processors.  Microsoft is also moving Windows 8 onto ARM (although they are still in Intel's camp also).  

 

No surprise that Apple would be considering this.  The fact is, that ARM SoCs are finally powerful enough to run a real computer, and much more power efficient than x86.  Not to mention cheaper.  Unless Intel can bring their power consumption and price down in a huge way, they're SOL...  

Hmm... Intel is too late.

Apple already did a "Samsung" on Sharp Actius MM10/20 and Sony 505.

Let's get this straight - Apple didn't invent thin and light category despite of what they trumpet.

 

No, but they created the 'ultrabook' concept, despite what Intel is claiming.

The thing is ARM is moving onto x86. Cortex 15 architecture is has an auto-of-order decoder - just like all of x86 chips besides Atom.

Problem with current ARM architecture is multi-tasking... it's just not wide enough to multi-task fast and power efficiently simply because the underlying architecture is that of a DSP controller.

It will have to significantly redesigned to compete with Intel in modern multitasking.

Edited by skatman - 10/5/12 at 4:43pm

What AMD can offer is significantly better graphics. Beyond that AMD does have issues with performance of its CPUs. What AMD imagines though is a future move to heterogeneous computing where a leadership position in GPUs can be leveraged to their advantage. It is a vision that jells well with Apples direction.

The thermals on AMDs chips are interesting to say the least. In some cases their systems can actually beat Intels offerings in power usage if the GPU is applied to advantage. Overall they are bigger power users but that comes with GPU performance that is anywhere from 20% to 60% faster than Intels chips.

Mind you this is all with a 64 bit architecture that is very similar to Intels. Intel in fact licensed the AMD 64 bit extensions.

In any event I always saw AMDs best opportunity was in the Mini where chips like Llano and Trinity offer a surprisingly good fit.

Where would Apple and PPC be today if they had not gone Intel? I venture to guess nowhere close to where Apple is now and IBM would still be struggling with development and production costs for fabbing what amounts an exclusive high performance chip for a tiny market share.

Now that Apple has deep pockets as a result of marketing successes in part because moving to Intel, do they want to go back to *how it was*? Even if there is some performance gain of the moment in technical terms?

The economics of scale factor heavily with computer chips. It's been proven time and again, buying and marketing is not always based on what performs best. Smart phones may be computers, but they aren't the same market. Whatever technical advantages *of the moment* ARM processors may have.

It raises the same issues with compatibility, marketing and productions costs as they faced with PPC. The "fallback" to Windows option is a like a security blanket for cautious or conservative buyers. And AMD, like IBM with PPC, has had moments where they were ahead of Intel on the performance curve. Those moments never lasted long.
Edited by Gary54 - 10/4/12 at 11:42pm

Not to mention what it puts their customers through. Myself, starting with System 6, I have gone through the transitions from Motorola 68k to PPC to Intel and from Classic to OSX. Every single one of them was a royal pain in the kazoo. I have files which are important to me that I cannot open and programs I cannot run without jumping through tons of hoops. If it can be done at all.

Making the transition to ARM will do it all over again. There has to be a major technical advantage involved to justify all the pain and aggravation for the customers.

Why do people always want to buy a cow when they only want milk?

It needs to be a 64-bit chip but it wouldn't have to be in a server. It could go in an 11" Air. The CPU in the entry 11" Air costs $225. If they made their own chip, they could probably cut the cost by up to $200 factoring in the GPU and motherboard and hit $799. It should be easy enough to get all Mac App Store apps compatible with it.

Some people assume that ARM would be slow but obviously the chips in iOS devices are slow because they are fanless devices. They'd crank up the clock speed in machines with better cooling:

http://news.cnet.com/8301-1001_3-57427681-92/heads-up-intel-tsmc-cranks-up-arm-chip-to-3ghz/

They can have dual-core in phones/tablets, quad-core in Airs, 8/16-cores in MBP/Mini/iMac and 50-core in MP.

It's always about performance per watt. Every machine has a set power profile. If the performance per watt is better than Intel, ARM will outperform them in the same power limit.

Not quite. The software would have to be entirely different. It changes from virtualisation to emulation. When we were on PPC, we had that great company Connectix who made software that emulated chip architectures. They even emulated a PSOne nearly perfectly with the Virtual Game Station.

The XBox has to use emulation to run original XBox games because it used an Intel CPU and the 360 uses PPC. Microsoft actually bought Connectix so they might be using some of the software for the XBox.

Apple could probably do the whole Rosetta binary translation for x86 Mac apps but Windows would have to be emulated. Both come with a significant performance hit but if we are talking about 5-10 years down the line, processors will be 6-30x or more faster so even an order of magnitude hit would still run up to 3x faster than what we have natively now.

Windows 8 RT can be dismissed just now but how many pro apps and games will be ported to it in a few years? What I see a lot of these days is people not liking monopolistic dependency. It's not good for Apple to be dependent on just Intel for the long term.

Imagine if they depended on Intel for iPhones. They couldn't be in control of yields, architecture design, heat output, power draw, delivery times, price. Plus, in the end, Intel gives them the same chips everyone else gets.

It does but purely optical connections are the way to go and they should be easy to convert. If you are just sending a light signal, you can translate it. Think of a connector like the following:



It has power with a mag-safe style interface. It is the size of a headphone jack and the wire would be tiny. It can easily have an adaptor box to switch an optical Thunderbolt signal.

Every port would share a 1Tbit+ connection and connect to the motherboard with a single wire, sending display data along with it.

Think way down the line, say 10 years. CPUs will be 30x faster, maybe more if they switch to optical transistors or qubits or just use graphene instead of silicon. With that power, you can do pretty much anything you want. You can emulate Windows and every app that has existed before. If we get ReRAM, computers can have 60GB+ RAM as standard and over 1TB of SSD. No matter how it is running, you won't be hitting a bottleneck even with an emulator.

Apple has shown what it can do with the A6 - outperform everyone else and ship 5 million in a weekend. If the benefits outweigh any negatives over the next decade, there's no reason this couldn't happen. As we've seen from the last decade, a whole lot of things can happen we don't expect.

I think it was helpful to ease the concerns of the uninitiated into the platform. If you buy a Mac and don't like the Apple platform, at least you have a nice computer that can run Windows.

POWER is a RISC-based chipset. The original PowerPC was a single-chip implementation of POWER. Apparently latter versions of PowerPC and POWER cross-pollinated each other. POWER3 was based on PowerPC 620. POWER3 was originally intended to be the PowerPC 630. The last PowerPC family, the PowerPC 970 aka PPC G5, was based on the POWER4 chipset. The current POWER chip set is the POWER7.

Or if you need Windows for a specific bit of software...  While OSX support is great, there is some obscure software (whose value is questionable, but never the less...) that is Windows only...  Oh, and lots of games are still Windows only (don't know why, OSX or Linux and OpenGL is much quicker than Windows and DirectX).   

MacOS X v10.4 runs very well on POWER5 boxes

A10 Thunderbolt perhaps.

No, that was done my Sharp, Sony, Fujitsu... maybe others LONG before apple.
 

 

Mac has 50 installed base and climbing rapidly with 5 million sales per quarter.

 

The kind of sales any Window straggling software vendors will find hard to ignore in time.  Games...CAD...whatever the apps.

 

It's madness to see 50 million in potential sales sitting on the shelf.  It's not the kind of money you walk away from.  Apple's making money out of it... :)

 

Open GL has always been great in my view.  I remember Unreal Tourney back in the day.  And I felt GL always looked richer than Direct X.

 

Lemon Bon Bon.

 

Great post, Marv'.  At the end of the day.  If Apple can put a 64 bit ARM chip in an Air, they well.  When?  That's for Apple to decide.  Maybe they'll head in that direction at some point.

 

You have ARM chips doing amazing things pushing amazing amount of pixels in an iPad 3 and the a6 chip with improved GPU ...will make a great iPad 4.

 

As we move to SoC, Apple are in complete control with ipods, iphones and iPads.  How long before you can dock a bigger iPad on a stand and call it 'iMac' of sorts?

 

Not in the next year or so.

 

But I feel we're in the process of moving to ARM.  50 million Apple 'A' class Arm chips vs 5 million Intel Mac chips.  Look at the profits and level of control Apple has.  Intel is in the Mac driving seat for now...  It maybe won't change in the forseeable future.

 

But Mac OS X is iOS.  It's basically the same thing.  iOS runs on ARM?  'X' is portable.  It's compact.

 

Apple can run it on what chips it wants.

 

Lemon Bon Bon.

...Apple like to control their own destiny.  It's not a crazy idea that they want to move away from Intel.

 

As soon as ARM has something that is 'good enough'...how long that will be..?

 

Lemon Bon Bon.

please can you elaborate this?

There was a chart posted on CultofMac:

http://www.cultofmac.com/193673/why-the-pc-is-dead-five-years-of-iphone-benchmarks-chart/



It shows a dramatic performance curve. The following video shows the iPhone 5 vs the newest iPad and 4S:

http://www.youtube.com/watch?feature=player_detailpage&v=QF9IzH1c7Dk#t=100s

The iPad 4 can be clocked even higher than the iPhone 5. The current iPad is 20% faster than the 4S so the next iPad might even crack 2k on Geekbench.

The entry MBA gets about 7k. The iPhone/iPad CPU will have a TDP of around 2-5W and the MBA 17W. If they double the clock and double the cores, it would already be competitive with Intel's option.

It wouldn't be compelling performance-wise as it's just a side-step but if Apple follows that performance curve and doubles performance every year, they will surpass Intel quite quickly who are only doubling performance every couple of years. Faster, $200 cheaper, cooler, longer battery life and all you lose is native x86 Windows and Thunderbolt compatibility but at the same time gain native iOS and Windows RT compatibility and access to hundreds of thousands of apps. x86 Windows can be emulated and Thunderbolt will be replaced eventually anyway.