Questionable rumor claims Apple's next-gen 'A10' processor could switch to six cores

123468

Comments

  • Reply 101 of 159

    I'm pretty sure Apple has been working on an ARM version of OSX, added to that we know that MS have been working on an ARM version of Windows 10.  It seems sensible that both Apple and MS create ARM based versions of their OS's.

     

    Added to that we are starting to see ARM catching up with Intel-based chipsets.  The new A9 chip in the iPhone is comparable with the Intel Core M chip in the new Macbook and it looks like the A9X chip in the iPad Pro will produce a Geekbench Multi-core score of between 6000-8000 (most likely closer to 7000).  With a score of around 7000, this is comparable with the current top spec MacBook Pro 13" (i5-2.9Ghz, Dual-core).

     

    So the big question is next year.  Realistically we could be seeing an A10X (3-core) with a Geekbench Multi-core of 10,000.  If they manage to create a 4-core version of the A10 chip, then we could be seeing a mobile chipset comparable with the current i7 MacBook Pro's which retail for $2000-$2500.

     

    Forget putting ARM in desktops at the moment, but I see no reason "power wise" why Apple cannot put their custom chips into all their MacBooks within the next 2 years; unless they don't have an ARM version of OSX.

  • Reply 102 of 159
    I would only give this a consideration if it was about the A10X.

    A six-core CPU and a 16-core GPU for the iPad Air 4, iPad Pro 2, and, for the first time, selected MacBook Airs.

    The A9 is 2 CPU cores and 6 GPU cores, I don't see the A10 going above 4 CPU cores and 8 GPU cores to be honest. It'll still be a market leader.

    (In fact I suspect this rumour is a massive mutation of the fact the A9 has a six-core GPU).
  • Reply 103 of 159
    Quote:

    Originally Posted by melgross View Post



    Then, there's the little problem that these third party apps would need to run in emulation. Forget it!

     

    Apps sold in the Mac OS App Store may be able to one day take advantage of the App Store features that Apple has recently enabled for iOS.

     

    The main one is bitcode, rather than native code. Basically an intermediary language (IL), with the final, optimising native compile, occurring on the App Store itself.

     

    Right now, XCode is compiling to x86-64 binaries for Mac OS X apps. However we know that Mac OS X supports fat binaries, so it is possible for XCode to add bitcode support easily here in the future.

     

    I'm sure app slimming will also be viable for Mac OS X apps.

     

    Maybe even the App Store will de/recompile x86-64 Mac OS X binaries into bitcode/ARM/ The result might not be super-optimal, but it will remove the need for emulation. And it could be done without anyone of us knowing, right up until the announcement of the A10X MacBook.

  • Reply 104 of 159
    Quote:

    Originally Posted by konqerror View Post

     

     

    Absolutely wrong. DRAM has to be continually refreshed. Every 25-100 ms (depending on temperature, die quality, etc) the entire contents of RAM have to be read out and rewritten. Sitting idle, your iPhone is burning 40 GB/s of internal memory bandwidth. 


     

    DRAM refresh is done internally within the memory chip. It doesn't burn memory bandwidth between the memory and the CPU. It uses special refresh cycles that occur between read/write cycles, and basically reads an entire row of memory and rewrites it back, all in the background, all automatically, all very low power. If you look at that table, the refresh is measured in micro-amps. Not milli-amps. Refresh power use is 1/10th of the overall DRAM power use, at most.

     

    It is the raising of signal strength to the external bus that uses significant power, hence DRAM power use scales with DRAM memory bandwidth utilised.

  • Reply 105 of 159
    Quote:

    Originally Posted by Misa View Post



    Because they are still very far apart. The A9's geekbench scores only put it within reach of a 7-year old desktop or a 4 year old subnotebook. That's barely impressive.

     

    Barely impressive? Sheesh.

     

    A9: ~2W - 4W max - $50 max SoC cost

    7-year old desktop: 95W + 50W GPU

    4-year-old macbook: 25W

    2-year-old macbook: 15W - $200+ cost

     

    Guess what? That A9 is running at 1.8GHz dual core, 450MHz GPU for a reason - to keep it under 2W in normal use. It is competitive apparently with a low-end Core M (4.5W in marketing, measured use is higher, turbo speed is 2.6GHz which is what benchmarks will run at until throttling kicks in).

     

    The A9X in a tablet can stretch to 6W - 8W max use, hence it has three or four cores, and something around 12 GPU cores - 2x the A9.

     

    A MacBook can employ even better cooling, so the chip can be clocked higher. I see no issues with an A10X in a laptop setting being able to run faster or have more cores. The six core rumour could be 2 fast cores and 4 slower cores for example - similar to ARM's big.LITTLE, which does work quite well for reducing power consumption. Those fast cores in a laptop could be 3GHz. Core M is left in the dust.

     

    But this is all theoretical until Apple release things.

  • Reply 106 of 159
    hattig wrote: »
    melgross wrote: »
    Then, there's the little problem that these third party apps would need to run in emulation. Forget it!

    Apps sold in the Mac OS App Store may be able to one day take advantage of the App Store features that Apple has recently enabled for iOS.

    The main one is bitcode, rather than native code. Basically an intermediary language (IL), with the final, optimising native compile, occurring on the App Store itself.

    Right now, XCode is compiling to x86-64 binaries for Mac OS X apps. However we know that Mac OS X supports fat binaries, so it is possible for XCode to add bitcode support easily here in the future.

    I'm sure app slimming will also be viable for Mac OS X apps.

    Maybe even the App Store will de/recompile x86-64 Mac OS X binaries into bitcode/ARM/ The result might not be super-optimal, but it will remove the need for emulation. And it could be done without anyone of us knowing, right up until the announcement of the A10X MacBook.


    I don't understand all I know about this ...

    But, I think Apple's (Chris Lattner's) Swift Language and LLVM may play a major roll in some of the platform independence you describe.

    As I understand it, the LLVM has a JIT compiler similar to Java. Additionally, LLVM has several Intermediate Formats (IFs) that provide machine independence, optimized code and deferred compilation.

    And Swift has a natural affinity with these LLVM IFs.


    Apple appears to be wrapping/reimplementing its current Objective-C, C, C++ Code/APIs in Swift to take advantage of the Swift/LLVM affinity.

    Lastly, Apple considers Swift to be a System Programming Language -- To me, that means that Swift can/will be used for everything from writing compilers, the LLVM, Developer Tools, Apps, an App programing language, a high-level and low-level Scripting Language ... the whole magilla.



    Overview and description[edit]
    LLVM can provide the middle layers of a complete compiler system, taking intermediate form (IF) code from a compiler and emitting an optimized IF. This new IF can then be converted and linked into machine-dependent assembly code for a target platform. LLVM can accept the IF from the GCC toolchain, allowing it to be used with a wide array of extant compilers written for that project.

    LLVM can also generate relocatable machine code at compile-time or link-time or even binary machine code at run-time.

    LLVM supports a language-independent instruction set and type system.[12] Each instruction is in static single assignment form (SSA), meaning that each variable (called a typed register) is assigned once and is frozen. This helps simplify the analysis of dependencies among variables. LLVM allows code to be compiled statically, as it is under the traditional GCC system, or left for late-compiling from the IF to machine code in a just-in-time (JIT) compiler fashion similar to Java. The type system consists of basic types such as integers or floats and five derived types: pointers, arrays, vectors, structures, and functions. A type construct in a concrete language can be represented by combining these basic types in LLVM. For example, a class in C++ can be represented by a combination of structures, functions and arrays of function pointers.

    The LLVM JIT compiler can optimize unneeded static branches out of a program at runtime, and thus is useful for partial evaluation in cases where a program has many options, most of which can easily be determined unneeded in a specific environment. This feature is used in the OpenGL pipeline of Mac OS X Leopard (v10.5) to provide support for missing hardware features.[13] Graphics code within the OpenGL stack was left in intermediate form, and then compiled when run on the target machine. On systems with high-end GPUs, the resulting code was quite thin, passing the instructions onto the GPU with minimal changes. On systems with low-end GPUs, LLVM would compile optional procedures that run on the local central processing unit (CPU) that emulate instructions that the GPU cannot run internally. LLVM improved performance on low-end machines using Intel GMA chipsets. A similar system was developed under the Gallium3D LLVMpipe, and incorporated into the GNOME shell to allow it to run without a proper 3D hardware driver loaded.[14]

    When it comes to the run-time performance of the compiled programs, GCC previously outperformed LLVM by about 10% on average.[15][16] Newer results do indicate, however, that LLVM has now caught up with GCC in this area, and is now compiling binaries of approximately equal performance, except for programs using OpenMP.[17]

    https://en.wikipedia.org/wiki/LLVM
  • Reply 107 of 159
    Quote:

    Originally Posted by melgross View Post



    $949 for a Pro with 4GB RAM and 128GB storage vs a Macbook Air with 4GB RAM and 128GB storage for $999. But you need to buy a keyboard for the Pro, and Apple's is $169, and it looks as though third parties keyboards will be about the same. So the Pro with keyboard is $1130. That's a lot more, and it's not really good as a laptop.

     

    OTOH the iPad Pro has a massively superior touch display than the MacBook Air.

     

    But the Air: proper SSD storage, lots of I/O, keyboard, touchpad.

     

    I do think that iPads are overpriced for what they offer, a side effect of miniaturisation and Apple's margins.

     

    An ARM MacBook Air would likely also incorporate the better display (but maybe not touch) however, and possibly try to be slimmer and lighter, so it would be more expensive. But that's not down to the cost of the "A10X" versus the Core i5...

     

    More likely Apple will use the threat of ARM to get even cheaper Core i5s from Intel in 2016, and the first MacBooks with ARM will be with A11X in 2017...

  • Reply 108 of 159

    Yes, LLVM is going to be a core aspect of the App Store's recompilation mechanism. Apple's bitcode is probably very close to an existing IL/IF.

     

    Swift as a systems language is going to be interesting to see in practice, although I think mass use of it will be years down the line.

  • Reply 109 of 159
    Ahh ...

    Didn't now that!

    LOL!

    t seems that there's a bug going around attacking... um... senior citizens that's causing us to "tweak" (not twerk!) folks in our comments.

    I think it's called "Gettin' Cheeky With It"... but you may have another name for it stateside(?). :D
  • Reply 110 of 159
    There are so many poor comments made in this article I cannot deal with them all...bottom line is who cares? There are so many ways Apple improves the true performance of its devices. Sure going to quad or higher core configurations but at this point Apple only knows the true path it will take. My personal opinion is they will eventually move to 4 unless there is some reason going to more would make sense??? Apple will not bypass 4 and go to 6 just because of performance. They are already outperforming Octa configs with dual core.
  • Reply 111 of 159
    wood1208wood1208 Posts: 2,917member
    Moving to Quad core is possibility. Going to hex or octa is not what Apple have done as increment. But, you never know about Apple. Logic is if Apple can achieve so good performance with dual core than just think with quad core. Moreover, chip die is not shrinking to 10nm until 2017 so may be move to higher core can get pushed to 2017 A11 processor.
  • Reply 112 of 159
    wood1208wood1208 Posts: 2,917member
    Quote:
    Originally Posted by davygee View Post

     

    I'm pretty sure Apple has been working on an ARM version of OSX, added to that we know that MS have been working on an ARM version of Windows 10.  It seems sensible that both Apple and MS create ARM based versions of their OS's.

     

    Added to that we are starting to see ARM catching up with Intel-based chipsets.  The new A9 chip in the iPhone is comparable with the Intel Core M chip in the new Macbook and it looks like the A9X chip in the iPad Pro will produce a Geekbench Multi-core score of between 6000-8000 (most likely closer to 7000).  With a score of around 7000, this is comparable with the current top spec MacBook Pro 13" (i5-2.9Ghz, Dual-core).

     

    So the big question is next year.  Realistically we could be seeing an A10X (3-core) with a Geekbench Multi-core of 10,000.  If they manage to create a 4-core version of the A10 chip, then we could be seeing a mobile chipset comparable with the current i7 MacBook Pro's which retail for $2000-$2500.

     

    Forget putting ARM in desktops at the moment, but I see no reason "power wise" why Apple cannot put their custom chips into all their MacBooks within the next 2 years; unless they don't have an ARM version of OSX.




    Apple already tested OSX on ARM but due to performance issue stayed with intel. Apple may have intention to move to it's own chip for all of their devices, question is if Intel keep reducing TDP and improving performance of their mobile chip than decision becomes difficult. Though, the way A-series improving in performance and multi-core is a trend than it won't take much time to catch up and surpass intel's x86 architecture in performance and TDP, battery life. In 2017, A-series will move to die shrink 10nm which gives more real estate on chip to squeeze more semiconductor. ARM is also pursuing it's architecture/chip in server business with HP.

    Apple can repeat performance of dual core with ios (both in-house design) with multi-core with OSX(both in-house design). Due to lower ARM TDP, it will allow Apple to make it thinner laptops with longer battery life. That is the moto of Apple in it's all design.

  • Reply 113 of 159
    pmzpmz Posts: 3,433member
    Quote:

    Originally Posted by SolipsismY View Post



    The biggest issue I see is a logistical one on how to work the OS to make it work with 3rd-party apps: Open installs or Mac App Store-only?

    I think you know the answer to that.

     

    The move to Apple chips in Macs is the move to Mac App Store only.

  • Reply 114 of 159
    pmzpmz Posts: 3,433member
    Quote:
    Originally Posted by Hattig View Post


    I do think that iPads are overpriced for what they offer

    :no: some people. What sort of insane delusion of reality do you live in that could even possibly motivate such a thought, let alone a post?

     

    Are you people really this disenchanted with technology already, you have no appreciation for what these devices actually are? The iPad is still years ahead of its time.

  • Reply 115 of 159
    Hexacore? Not. I believe Apple will eventually add more cores, but for now, why? Their dual-core chips spank the competition's octocore chips, so it's unnecessary. Better to keep the design simpler and more power efficient.

    Since Apple does their own chip design, there is no reason they can't do something like move to a triple core design. They've already done that with the GPU portion of the X chips.
  • Reply 116 of 159
    jbdragonjbdragon Posts: 2,311member
    Quote:

    Originally Posted by Red Oak View Post



    The A10 is going to be a beast. The Apple chip development team is on Fire. Its shocking how quickly they are catching up to Intel in raw performance

     

    Apple is only using 2 or 3 Cores right now.  I just don't see any reason why Apple would jump to 6 cores at this point.

     

    I also don't see Apple switching to the A processor from Intel any time soon.  Intel is not just sitting still letting Apple try to catch p, but more then that, may people duel boot their Mac's so that they can also run Windows.  You would lose that ability going to a A processor.   Maybe it could still work under emulation, but now you're greatly killing off speed to do it.

  • Reply 117 of 159
    Originally Posted by asdasd View Post

    Because the mouse and keyboard are the future of desktop.



    Just wake up from a coma? It’s not 1984.

  • Reply 118 of 159
    Quote:

    Originally Posted by JBDragon View Post

     

     

    Apple is only using 2 or 3 Cores right now.  I just don't see any reason why Apple would jump to 6 cores at this point.

     

    I also don't see Apple switching to the A processor from Intel any time soon.  Intel is not just sitting still letting Apple try to catch p, but more then that, may people duel boot their Mac's so that they can also run Windows.  You would lose that ability going to a A processor.   Maybe it could still work under emulation, but now you're greatly killing off speed to do it.




    I agree, I don't see why Apple would increase to 6 cores straight away from 3 core max they have at the moment.

     

    I also agree about the dual boot, but there are two ways to solve this issue, firstly use an ARM-based version of Windows 10 (when it becomes available, possibly next year) or use the power/speed of the chip to emulate intel-based Windows.

     

    As you were saying, the Apple's A chips are performing well in single core performance and next year we could be seeing a 3500 GkB3 single core score and I see no reason why Apple couldn't push a quad-core A10X or whatever they want to call it into a MacBook.  We should see a 10,000-11,000 GkB3 Multi-core score.  Compare that with the current Intel Core M chip in the MacBook 12" which is only just over 4000 at the moment and even the new Core M3,5,7 chips from Intel won't push the performance up by that much compared to the current Core M.  Talking about a 20% performance increase and a 40% GFX performance increase.

     

    Then there is the cost of the chips as well.  Market prices of $280+ for the new Core M chips compared to under $50 for Apple's.

  • Reply 119 of 159
    asdasdasdasd Posts: 5,686member

    Just wake up from a coma? It’s not 1984.

    It's not 1880 and I'm typing this on a qwerty keyboard. Albeit a virtual one.

    A desktop class device has a mouse input. Tablets are driven by touch. Notice that Apple is selling a keyboard but not a mouse with the iPad pro.
  • Reply 120 of 159
    asdasdasdasd Posts: 5,686member
    pmz wrote: »
    I think you know the answer to that.

    The move to Apple chips in Macs is the move to Mac App Store only.

    Not necessarily. They may disallow apps downloaded from the web from launching regardless of user preferences but that will kill the market for a power Mac or a Mac Pro for power users.

    Apple wasnt doing anything extraordinary when it locked down the iPhone. The pre existing "smart phones" were locked down for apps. Invitation only in fact. Apple drove a truck through that wall and got dogs abuse for a "walled garden" when Google propagandised their open alternative. But that was later.

    Since then people seem to believe that Apple wants to lock down the Mac despite no real effort years later to do so. It's like assuming that a walled garden in the Xbox inevitably leads to a closed Windows machine.
Sign In or Register to comment.