The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
At this point in time, I wonder how many MacBook users even care about Windows. Seems like it would be easy to bifurcate the line such that an ARM MacBook analog running iOS Plus (for lack of a better term) would be the way to go for both cost and performance reasons.
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
Apple has the A-series, S-series, W-series, T-series, and even the M-series co-processor. The first 3 run distinct product lines from Apple. I'm not sure they'd call their own silicon running a low-end Mac an A-series. I'd think that the amount of RAM, GPU, and other features, like the potential for x86_64 emulation (which I doubt would be included), would fall in line with what Apple uses for the iPhone or iPad.
I see what you are saying... except that the others are speciality chips and/or reduced function/size ships. The A series have a much broader potential. Actually, there already are 2 A series offerings the A and the AX -- where the AX has even broader potential than The A -- with fewer space, power, heat dissipation limitations.
So, maybe A11X chips running macOS in certain MacBooks and iPad models.
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
At this point in time, I wonder how many MacBook users even care about Windows. Seems like it would be easy to bifurcate the line such that an ARM MacBook analog running iOS Plus (for lack of a better term) would be the way to go for both cost and performance reasons.
I have to assume Apple has fairly solid numbers for how many users use Boot Camp or VMs. My guess is that it's not as large as many here believe, and if my guess is correct about an ARM-based Mac coming in at the low-end of the product line, I'd wager that running Windows on a low-end Mac becomes an even less likely scenario.
On top of that, we need to consider how important that is to Apple. Apple wasn't the first to offer a dual boot solution, and today I think that need is even less important for Apple's success than it was a decade ago.
PS: Why not call it macOS since it's the same OS, just as it was still macOS hen they added support for x86 along with PPC.
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
At this point in time, I wonder how many MacBook users even care about Windows. Seems like it would be easy to bifurcate the line such that an ARM MacBook analog running iOS Plus (for lack of a better term) would be the way to go for both cost and performance reasons.
I have to assume Apple has fairly solid numbers for how many users use Boot Camp or VMs. My guess is that it's not as large as many here believe, and if my guess is correct about an ARM-based Mac coming in at the low-end of the product line, I'd wager that running Windows on a low-end Mac becomes an even less likely scenario.
On top of that, we need to consider how important that is to Apple. Apple wasn't the first to offer a dual boot solution, and today I think that need is even less important for Apple's success than it was a decade ago.
PS: Why not call it macOS since it's the same OS, just as it was still macOS hen they added support for x86 along with PPC.
I am guessing that in the enterprise market that they have made some strides into with the likes of IBM etc... the use of VMs with x86 is higher - at the higher end of the performance line (i.e. MacBook Pro). I personally do use it, but not for Windows. I have a VMs running Linux with either an Oracle database running or mirroring a server deployment (yes I know I am not typical). I think you are probably underestimating the number of users using VMs - there has to be enough to support VMWare selling Fusion, Parallels and those using the open source one... if it were really small - VMWare would never have bothered.
The number of people running VMs on MacBook 12... or MacBook Air... probably insignificant.
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
If Apple goes ARM in its Mac line up (they should) it will lkkely be with an ARM chip optimized for the task. Doing so leads to a more efficient chip and allows deleting anything i devices specific.
Why? So that they can add Mac specific hardware and optimizations. For example a mac destined chip would need uSB-C/TB I/O. In some Macs they would need to support dual channel RAM. Also dont forget graphics as somme machines would need a PCI Express port. Adding this stuff isnt much different than Intel offer various models of i86 chips all built around the same generation hardware. In Apples case the general CPU archetecture would be either A11 generation or A12 with the rest of the chip optimized for the Mac platform.
About the only place that A11 could be used directly is in the Mac Book where there is little I/O demand. Even then you would likely need an A11X for external RAM support. I say that because i dont believe it is possible yet to stack 8GB in package. Even that may be wrong as TSMC has announced many new processes and improvements recently that might lead to much more RAM in package.
So in simple terms dont expect to see a chip that exactly matches A11, A11X or newr iPhone chips in most Macs.
1) I wonder what the price bump is for an extra 16GiB of RAM.
That's easy. Go to macsales.com and look up the price of 16GB LPDDR. Then apply the same formula used for converting Fahrenheit to Celsius: Double it and add 30.
1) I wonder what the price bump is for an extra 16GiB of RAM.
That's easy. Go to macsales.com and look up the price of 16GB LPDDR. Then apply the same formula used for converting Fahrenheit to Celsius: Double it and add 30.
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
If Apple goes ARM in its Mac line up (they should) it will lkkely be with an ARM chip optimized for the task. Doing so leads to a more efficient chip and allows deleting anything i devices specific.
Why? So that they can add Mac specific hardware and optimizations. For example a mac destined chip would need uSB-C/TB I/O. In some Macs they would need to support dual channel RAM. Also dont forget graphics as somme machines would need a PCI Express port. Adding this stuff isnt much different than Intel offer various models of i86 chips all built around the same generation hardware. In Apples case the general CPU archetecture would be either A11 generation or A12 with the rest of the chip optimized for the Mac platform.
About the only place that A11 could be used directly is in the Mac Book where there is little I/O demand. Even then you would likely need an A11X for external RAM support. I say that because i dont believe it is possible yet to stack 8GB in package. Even that may be wrong as TSMC has announced many new processes and improvements recently that might lead to much more RAM in package.
So in simple terms dont expect to see a chip that exactly matches A11, A11X or newr iPhone chips in most Macs.
Still, don't know about TB on any chip but ARM. Apple transferred all rights to Intel I believe with regards to the joint work, as such, I am not sure using it with another CPU architecture is possible. Of course they could pivot and go with OCuLink-2 instead of Thunderbolt (PCIe 4 standard - equivalent to TB )
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
If Apple goes ARM in its Mac line up (they should) it will lkkely be with an ARM chip optimized for the task. Doing so leads to a more efficient chip and allows deleting anything i devices specific.
Why? So that they can add Mac specific hardware and optimizations. For example a mac destined chip would need uSB-C/TB I/O. In some Macs they would need to support dual channel RAM. Also dont forget graphics as somme machines would need a PCI Express port. Adding this stuff isnt much different than Intel offer various models of i86 chips all built around the same generation hardware. In Apples case the general CPU archetecture would be either A11 generation or A12 with the rest of the chip optimized for the Mac platform.
About the only place that A11 could be used directly is in the Mac Book where there is little I/O demand. Even then you would likely need an A11X for external RAM support. I say that because i dont believe it is possible yet to stack 8GB in package. Even that may be wrong as TSMC has announced many new processes and improvements recently that might lead to much more RAM in package.
So in simple terms dont expect to see a chip that exactly matches A11, A11X or newr iPhone chips in most Macs.
Ahh... Thanks, that explains a lot.
Currently we have 2 loaded (~$3,100 ea) 2017 13 inch MBPs in our household -- both need Windows capability (Parallels). My youngest grandson will start Community College in August... I was hoping for a less-expensive MBP for him -- but that would mean ARM running X86... Sigh!
It appears the first models can't compete on base and boost clocks, there's the problem with milking things to 14nm+++, suddenly the newer node can't compare.
Looks like the first parts will be pipe cleaners for 10nm, alas.
Wonder if Apple will just wait it out, or else has Kaby Lake G parts planned for WWDC.
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
At this point in time, I wonder how many MacBook users even care about Windows. Seems like it would be easy to bifurcate the line such that an ARM MacBook analog running iOS Plus (for lack of a better term) would be the way to go for both cost and performance reasons.
In my SW company we are using MBP for the development of Cloud applications. All Cloud providers (Google, Amazon, Microsoft, Digital Ocean, ...) are only using Intel CPUs. For any cloud development Intel CPUs are a necessity. The day that Apple would stop selling Intel based Macbooka is the day that I'll have to stop buying Macbooks. But I don't think Apple would be that stupid
It appears the first models can't compete on base and boost clocks, there's the problem with milking things to 14nm+++, suddenly the newer node can't compare.
Looks like the first parts will be pipe cleaners for 10nm, alas.
Wonder if Apple will just wait it out, or else has Kaby Lake G parts planned for WWDC.
Based on the complaints on forums about the lack of 32GiB RAM it sounds like there's an interim market to get people to buy new Mac notebooks.
It appears the first models can't compete on base and boost clocks, there's the problem with milking things to 14nm+++, suddenly the newer node can't compare.
Looks like the first parts will be pipe cleaners for 10nm, alas.
Wonder if Apple will just wait it out, or else has Kaby Lake G parts planned for WWDC.
MBP13 has U-series Coffee Lake options, namely 4-core options MBP15 has H-series whatever Coffee Lake options, namely 6-core. iMac has Kaby Lake G options, but really isn’t necessary. Coffee Lake w/discrete GPUs sounds fine if not better, unless they have a new industrial design.
I think this is all but inevitable for WWDC or a little later.
The MB12? The only option is to move up the next Kaby Lake Y series speed grades over this summer. Or put in a fan with a Coffee Lake U chip, maybe a fanless U series chip with cTDP down to 7 W. Maybe.
I wish they would hurry up, I need my 32GB now The ultimate MBP would be
* 32GB of LPDDR4 * The new 6 core 45-watt i7 * 2TB Flash * 15" screen (retina) with 100Hz at 100% adobe RGB gamut (or whatever it is,lol) * Nvidia 1080MTi ( i know not even invented yet) * 4 USB-C ports
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
Why emulate when they could have a low power x86 in the machine as a coprocessor? the i3 in the article would work for MacBook even MacBook Pro with maybe a beefier A12x.
Most of the shipping code Mac is only a compiler away from testing if it runs ARM. So an app could split out parts that need x86 build them as services on coprocessor and run user interface and some tasks on ARM or GPU
dick applebaum said: Isn't the X86 emulator a legal (licensing) issue rather than a tech issue?
There are ways of getting around that as long as it is legally reverse engineered much like AMD did. With USB-C we really don't need Thunderbolt.
Yeah, apparently there are problems with legally reverse engineered code too... I read that Intel was all over AMD -- and finally granted them a non-transferrable license to use the code. I guess, if Apple did a legal reverse engineering of the code, they could include it in their chips as a fait accompli -- and let their lawyers earn their keep.
The rationale to stick with Intel in Macs is getting weaker.
Probably getting close. Apple's A11 Bionic already has nearly twice as many transistors as the Intel quad core i7. Once they can get an X86 emulator to run Windows with minimal performance hit there should be no reason not to switch to ARM.
Isn't the X86 emulator a legal (licensing) issue rather than a tech issue? If so, Apple is in a pretty good negotiating position:
Macs are a small percent of the total computers that use Intel chips
iDevices provide a very large percentage of mobile devices using modems -- potentially Intel modem chips
Is it possible that lower volumes of A11s could be used in MacBooks and iPads soon -- earlier than the expected new iPhones in the Fall-Winter?
If Apple goes ARM in its Mac line up (they should) it will lkkely be with an ARM chip optimized for the task. Doing so leads to a more efficient chip and allows deleting anything i devices specific.
Why? So that they can add Mac specific hardware and optimizations. For example a mac destined chip would need uSB-C/TB I/O. In some Macs they would need to support dual channel RAM. Also dont forget graphics as somme machines would need a PCI Express port. Adding this stuff isnt much different than Intel offer various models of i86 chips all built around the same generation hardware. In Apples case the general CPU archetecture would be either A11 generation or A12 with the rest of the chip optimized for the Mac platform.
About the only place that A11 could be used directly is in the Mac Book where there is little I/O demand. Even then you would likely need an A11X for external RAM support. I say that because i dont believe it is possible yet to stack 8GB in package. Even that may be wrong as TSMC has announced many new processes and improvements recently that might lead to much more RAM in package.
So in simple terms dont expect to see a chip that exactly matches A11, A11X or newr iPhone chips in most Macs.
Still, don't know about TB on any chip but ARM. Apple transferred all rights to Intel I believe with regards to the joint work, as such, I am not sure using it with another CPU architecture is possible. Of course they could pivot and go with OCuLink-2 instead of Thunderbolt (PCIe 4 standard - equivalent to TB )
I really don't know what happened between Intel and Apple with respect to TB. It likely doesn't matter at this point as intel has moved to open up the interface.
In any event, it is just a point to support the example, If ARM goes into Macs it will likely need more external ports, industry standards ports too. The other really interesting thing here is the GPU which Apple could support in multiple ways. A beefed up GPU in an A11X for example might be good enough for a Mac Book right now.
One other thing that people might not consider is that TSMC actually has various tunings to their processes. The current A11 is running on a process optimized for very low power, Apple can easily move the current chips to higher performance processes enabling higher clock rates.
From my perspective the only thing that keeps Apple from moving to ARM is the will to do so. They have a good core to implement in a new chip so in a sense the chip is half done already.
dick applebaum said: Isn't the X86 emulator a legal (licensing) issue rather than a tech issue?
There are ways of getting around that as long as it is legally reverse engineered much like AMD did. With USB-C we really don't need Thunderbolt.
USB-C directly supports TB, it is one and the same port. The advantage TB gives us is greatly increased speed. TB is mandatory for desktops and makes a lot of sense if you want to dock a laptop. In fact I see TB as a modern day docking port.
The AMD / Intel arrangement is more complex than one might think. In any event I really think people are way off the mark with thinking that i86 is even required for most entry level users.
Comments
So, maybe A11X chips running macOS in certain MacBooks and iPad models.
On top of that, we need to consider how important that is to Apple. Apple wasn't the first to offer a dual boot solution, and today I think that need is even less important for Apple's success than it was a decade ago.
PS: Why not call it macOS since it's the same OS, just as it was still macOS hen they added support for x86 along with PPC.
The number of people running VMs on MacBook 12... or MacBook Air... probably insignificant.
Why? So that they can add Mac specific hardware and optimizations. For example a mac destined chip would need uSB-C/TB I/O. In some Macs they would need to support dual channel RAM. Also dont forget graphics as somme machines would need a PCI Express port. Adding this stuff isnt much different than Intel offer various models of i86 chips all built around the same generation hardware. In Apples case the general CPU archetecture would be either A11 generation or A12 with the rest of the chip optimized for the Mac platform.
About the only place that A11 could be used directly is in the Mac Book where there is little I/O demand. Even then you would likely need an A11X for external RAM support. I say that because i dont believe it is possible yet to stack 8GB in package. Even that may be wrong as TSMC has announced many new processes and improvements recently that might lead to much more RAM in package.
So in simple terms dont expect to see a chip that exactly matches A11, A11X or newr iPhone chips in most Macs.
Currently we have 2 loaded (~$3,100 ea) 2017 13 inch MBPs in our household -- both need Windows capability (Parallels). My youngest grandson will start Community College in August... I was hoping for a less-expensive MBP for him -- but that would mean ARM running X86... Sigh!
Looks like the first parts will be pipe cleaners for 10nm, alas.
Wonder if Apple will just wait it out, or else has Kaby Lake G parts planned for WWDC.
MBP15 has H-series whatever Coffee Lake options, namely 6-core.
iMac has Kaby Lake G options, but really isn’t necessary. Coffee Lake w/discrete GPUs sounds fine if not better, unless they have a new industrial design.
I think this is all but inevitable for WWDC or a little later.
The MB12? The only option is to move up the next Kaby Lake Y series speed grades over this summer. Or put in a fan with a Coffee Lake U chip, maybe a fanless U series chip with cTDP down to 7 W. Maybe.
The ultimate MBP would be
* 32GB of LPDDR4
* The new 6 core 45-watt i7
* 2TB Flash
* 15" screen (retina) with 100Hz at 100% adobe RGB gamut (or whatever it is,lol)
* Nvidia 1080MTi ( i know not even invented yet)
* 4 USB-C ports
That machine would say bye bye PC industry
the i3 in the article would work for MacBook even MacBook Pro with maybe a beefier A12x.
Most of the shipping code Mac is only a compiler away from testing if it runs ARM. So an app could split out parts that need x86 build them as services on coprocessor and run user interface and some tasks on ARM or GPU