I still think Apple needs to buy or build its own fab. This seems to be the weak link in its chain. It's need for silicon is not going to go away. As Kennedy said: "we don't do this because it's easy, but because it's hard." I'd have rather seen this than more stores and a big headquarters.
very costly (1) in Billion dollar adventure and not short time solution - long time commitment, when you deal with 10 and 7 nm, tooling and testing are all need to be developed, commissioned, round robin standardized, spc run chart established...(2) not easy to get the set of skills in Semi, unless you try to buy a whole team from some outfit - you can't improve process what you don't really know, large FAB if not consider carefully, you might end up magnify the problem many magnitude ... better partner with someone. it is a learning curve for 10 nm technology...should stay with 14 nm. until it is matured... IMHO.
I am not chip expert but I would agree. It would be strange when Apple would make A10 in 14nm and A10X in 10nm. A11 is expected to be 10 nm.
IDK, Since Apple is reportedly going to use a 10nm A11in next years iPhone, it may be worthwhile to shake out the 10 nm manufacturing process with a lower volume chip like the A10X.
Or, the could even be building an A11X at 10nm for a much more robust iPad Pro.
Then there's this: If the A11X chip is a superset of the A11 chip (same footprint, more cores, faster clocking, etc.) -- they could target A11X chips, and retest any rejects at lower A11 specs.
I don't buy the details; either it's an A10X at 14 nm or it's an A11X at 10 nm since Apple has the resources to optimize the architecture to the 10 nm node.
Even if Apple has rumored plans for a 10.2 inch reduced bezel iPad Pro requiring a SOC at 10nm, why would they scale the existing architecture to 10nm? Apple would opt for an optimized architecture, especially when it comes to the GPU.
I would expect an A11 or A11X, but we have evidence of Apple using an older design with a new process node in the iPad 2,4.
An A11X would seem make sense for a new, large iPad Pro. If they eliminate the bezels on the 12.9" iPad Pro, they would get ~14.3" diagonal in the same form factor.
I don't buy the details; either it's an A10X at 14 nm or it's an A11X at 10 nm since Apple has the resources to optimize the architecture to the 10 nm node.
Even if Apple has rumored plans for a 10.2 inch reduced bezel iPad Pro requiring a SOC at 10nm, why would they scale the existing architecture to 10nm? Apple would opt for an optimized architecture, especially when it comes to the GPU.
I would expect an A11 or A11X, but we have evidence of Apple using an older design with a new process node in the iPad 2,4.
Still, Apple has the resources so that they don't have to do a die shrink.
How can you say you need less resources to do a die shrink. The die shrink is an advance and is made possible because the state of the art with fabricators like Intel, TMSC, and Samsung, and is feasible for Apple to jump on sooner than other vendors BECAUSE of their resources. This is why the iPad 2,4 moved got the smaller process mode, not because Apple didn't have enough resources 4 years ago to keep a larger process node. It was a field test for the future of that process node in upcoming devices.
It can be whatever processor. It just has to have 4G of RAM, to be "not sluggish" for 5 years into the future. Also, if they are going to change the connector away from lightning, now would be the right time to do it. NOT in the model after i buy one.
That 10.5 inch iPad might need a cooler chip though. Just like the special iPhone, previous experience may not be very predictive.
if they are delayed, it might help my wallet because I'm in the market for an iPad Pro and a Nintendo Switch ... or two! Not sure how that'll work out. :P
I suspect that if there is any truth to this report Apple is trying to get a massive performance boost for the iPad so that it can rationally compete with any laptop out there. The problem of course is that Apple doesn't often get this far onto the bleeding edge.
However these sorts of reports do get the imagination a working real hard as to what a process shrunk iPad might be like next year. I could see 4 CPU's with a max clock rate hitting 3 GHz, larger internal caches and of course a newer GPU. This would make for a very fast iPad especially if the hardware is coupled with more advanced memory sub systems.
14nm vs 10nm isn't going to 'save space' in an iPad. Manufacturing process has nothing to do with physical chip size.
Actually it can save space. You are right in one sense that the smaller processes don't necessarily mean a smaller die though it can. What is does mean is more space on die so there is less need for external hardware. Ultimately I expect Apple to integrate the radio technology on chip but I'd be surprised if they can get it ready for this revision.
If Apple isn't after more performance the other thing that they might be after is far lower power usage which can lead to a thinner battery or more power for the display (OLED). I'm a bit up in the air as to the truth in this report, I do know from other sources that TSMC is basically going all in to get 10 nm out the door in 2017. iPad is a good place to introduce such tech as it is far lower volume thus far lower risk and a good place to ramp up technology.
I don't buy the details; either it's an A10X at 14 nm or it's an A11X at 10 nm since Apple has the resources to optimize the architecture to the 10 nm node.
Even if Apple has rumored plans for a 10.2 inch reduced bezel iPad Pro requiring a SOC at 10nm, why would they scale the existing architecture to 10nm? Apple would opt for an optimized architecture, especially when it comes to the GPU.
Each X model has had minor tweaks to get to its X level of performance but in general still of the same generation. Shrinking the processor doesn't mean that it is a new generation per say. The thing here is that a shrink to 10 nm could be a huge win for Apple and give them the several options in the optimization direction they take.
As for why they would scale the existing architecture to 10 nm, there are lots of good reasons to do so. First the existing architecture is pretty impressive, depending upon which way they go to optimize it for a 10X, Apple could see a huge boost in performance. A shrink, that allows for faster operation, would boost performance especially if at the same time they could increase the size of internal caches and the speed of external memory interfaces.
Lets face it every X variant has had enhancements to one extent or another over the non X variant. A shrink not only offers Apple the speed increase for the watt expended it also offers Apple considerably more die space. Die space means more hardware. More hardware means more specialized function blocks. A quick list of potential die space use beyond processor improvements:
Enhanced GPUs. Basically more execution units.
Enhanced video decoders and encoders. Here we can support more codecs in hardware and increase quality.
Enhanced ports (eventually iPad will need USB-C class ports).
Support for AI in hardware. Lots of interesting approaches here but this could be integrated into either the GPU or the CPU complexes. Or Apple could implement an entirely separate processor core for AI acceleration.
Move the RF processing on board the die.
Much larger caches.
New memory interfaces to high performance memory.
Tha tis seven off the top of my head. I'm sure others can add to the list. The point is there is a lot Apple could do with the die space that has little to do with the CPU itself. Personally I think the two big ones that Apple absolutely has to be looking at are RF processing on die and new AI hardware. RF processing just gets rid of an external chip and saves power and space. AI tech (really machine learning tech which I don't consider artificial intelligence), is coming on strong with AMD, Nvidia and Intel all taking their own approaches to the problem of AI acceleration. Apple needs an AI solutions that they can sheared along with the own development vectors, so I wouldn't be surprised to see something tacked onto an A10 to make it an A10X even though big step like this is more A11. The problem is Apple can't afford to get too far behind here.
Any way for us readers to automatically opt out of any DigiTimes related posts in the future? It's an embarrassment that this DigiTimes drivel continue to be treated as anything other than clickbait.
You really need to read Digitimes yourself. It is the fault of sites like Appleinsider that they leave people to believe these are rumors with solid underpinnings. Digitimes has a place for this marginal reporting along with a rather large volume of credible material. They leave it up to the reader to make good use of the legitimate articles.
I am not chip expert but I would agree. It would be strange when Apple would make A10 in 14nm and A10X in 10nm. A11 is expected to be 10 nm.
IDK, Since Apple is reportedly going to use a 10nm A11in next years iPhone, it may be worthwhile to shake out the 10 nm manufacturing process with a lower volume chip like the A10X.
Or, the could even be building an A11X at 10nm for a much more robust iPad Pro.
Then there's this: If the A11X chip is a superset of the A11 chip (same footprint, more cores, faster clocking, etc.) -- they could target A11X chips, and retest any rejects at lower A11 specs.
A process shake out on a low volume chip makes lots of sense. However Apple wouldn't go there unless there was a significant advantage to doing so.
As for a more robust iPad, there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development. The obvious reason being that iPad simply requires far more performance. This seems to have been put on the back burner probably because the process advances have been so quick and so huge that both hardware platforms could be supported by one architecture. With process shrinks hitting the wall we could see split with the processors going in a different direction.
I don't buy the details; either it's an A10X at 14 nm or it's an A11X at 10 nm since Apple has the resources to optimize the architecture to the 10 nm node.
Even if Apple has rumored plans for a 10.2 inch reduced bezel iPad Pro requiring a SOC at 10nm, why would they scale the existing architecture to 10nm? Apple would opt for an optimized architecture, especially when it comes to the GPU.
Each X model has had minor tweaks to get to its X level of performance but in general still of the same generation. Shrinking the processor doesn't mean that it is a new generation per say. The thing here is that a shrink to 10 nm could be a huge win for Apple and give them the several options in the optimization direction they take.
As for why they would scale the existing architecture to 10 nm, there are lots of good reasons to do so. First the existing architecture is pretty impressive, depending upon which way they go to optimize it for a 10X, Apple could see a huge boost in performance. A shrink, that allows for faster operation, would boost performance especially if at the same time they could increase the size of internal caches and the speed of external memory interfaces.
Lets face it every X variant has had enhancements to one extent or another over the non X variant. A shrink not only offers Apple the speed increase for the watt expended it also offers Apple considerably more die space. Die space means more hardware. More hardware means more specialized function blocks. A quick list of potential die space use beyond processor improvements:
Enhanced GPUs. Basically more execution units.
Enhanced video decoders and encoders. Here we can support more codecs in hardware and increase quality.
Enhanced ports (eventually iPad will need USB-C class ports).
Support for AI in hardware. Lots of interesting approaches here but this could be integrated into either the GPU or the CPU complexes. Or Apple could implement an entirely separate processor core for AI acceleration.
Move the RF processing on board the die.
Much larger caches.
New memory interfaces to high performance memory.
Tha tis seven off the top of my head. I'm sure others can add to the list. The point is there is a lot Apple could do with the die space that has little to do with the CPU itself. Personally I think the two big ones that Apple absolutely has to be looking at are RF processing on die and new AI hardware. RF processing just gets rid of an external chip and saves power and space. AI tech (really machine learning tech which I don't consider artificial intelligence), is coming on strong with AMD, Nvidia and Intel all taking their own approaches to the problem of AI acceleration. Apple needs an AI solutions that they can sheared along with the own development vectors, so I wouldn't be surprised to see something tacked onto an A10 to make it an A10X even though big step like this is more A11. The problem is Apple can't afford to get too far behind here.
I'm guessing that I'm the only one that considers a die shrink this'
"The term die shrink (sometimes optical shrink or process shrink) refers to a simple semiconductor scaling of semiconductor devices, mainly transistors. The act of shrinking a die is to create a somewhat identical circuitry using a more advanced fabrication process, usually involving an advance of lithographicnode. This reduces overall costs for a chip company, as the absence of major architectural changes to the processor lowers research and development costs, while at the same time allowing more processor dies to be manufactured on the same piece of silicon wafer, resulting in less cost per product sold."
Why would Apple do a simple die shrink when they have the resources to re-architect the SOC for even greater performance? Apple certainly has the volume to justify that.
I am not chip expert but I would agree. It would be strange when Apple would make A10 in 14nm and A10X in 10nm. A11 is expected to be 10 nm.
IDK, Since Apple is reportedly going to use a 10nm A11in next years iPhone, it may be worthwhile to shake out the 10 nm manufacturing process with a lower volume chip like the A10X.
Or, the could even be building an A11X at 10nm for a much more robust iPad Pro.
Then there's this: If the A11X chip is a superset of the A11 chip (same footprint, more cores, faster clocking, etc.) -- they could target A11X chips, and retest any rejects at lower A11 specs.
A process shake out on a low volume chip makes lots of sense. However Apple wouldn't go there unless there was a significant advantage to doing so.
As for a more robust iPad, there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development. The obvious reason being that iPad simply requires far more performance. This seems to have been put on the back burner probably because the process advances have been so quick and so huge that both hardware platforms could be supported by one architecture. With process shrinks hitting the wall we could see split with the processors going in a different direction.
The whole point of using 10nm on iPad is that any quirks with 10nm would be iron out by the time A11 hits. 10nm is completely new for both Apple and TSMC. And someone has to risk / put in money to test it.
The current iPad Pro is still on A9X. And the A10 ( by the way it is 16nm, not 14nm ) in iPhone 7 is already faster then A9X in many benchmarks. An A10x, even if it is a die shrink of A10, will be pretty damn fast, because the iPad has much higher TDP to work with, along with the optimization and gain from 10nm.
It is basically like Intel's Tick Tock, except Apple is doing it in a MUCH faster cycle. ( Note: Intel's Tock is a full node transition, TSMC is smaller incremental gains half node.)
I am not chip expert but I would agree. It would be strange when Apple would make A10 in 14nm and A10X in 10nm. A11 is expected to be 10 nm.
IDK, Since Apple is reportedly going to use a 10nm A11in next years iPhone, it may be worthwhile to shake out the 10 nm manufacturing process with a lower volume chip like the A10X.
Or, the could even be building an A11X at 10nm for a much more robust iPad Pro.
Then there's this: If the A11X chip is a superset of the A11 chip (same footprint, more cores, faster clocking, etc.) -- they could target A11X chips, and retest any rejects at lower A11 specs.
A process shake out on a low volume chip makes lots of sense. However Apple wouldn't go there unless there was a significant advantage to doing so.
As for a more robust iPad, there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development. The obvious reason being that iPad simply requires far more performance. This seems to have been put on the back burner probably because the process advances have been so quick and so huge that both hardware platforms could be supported by one architecture. With process shrinks hitting the wall we could see split with the processors going in a different direction.
The main reason for doing this would be to leave the A10 with current process to a new iteration of the regular non pro Ipad (right now on A8X) and put the much lower volume PRO one process ahead to further differentiate it. Could put a A11X in the Pro, the A11 in the Iphone 8 and next year the A11X winds up in the regular Ipad
Got the feeling they'll make the regular focus on thin, very light, rigid, say 4-5mm and slightly bigger screen (because of less bezel).
I am not chip expert but I would agree. It would be strange when Apple would make A10 in 14nm and A10X in 10nm. A11 is expected to be 10 nm.
IDK, Since Apple is reportedly going to use a 10nm A11in next years iPhone, it may be worthwhile to shake out the 10 nm manufacturing process with a lower volume chip like the A10X.
Or, the could even be building an A11X at 10nm for a much more robust iPad Pro.
Then there's this: If the A11X chip is a superset of the A11 chip (same footprint, more cores, faster clocking, etc.) -- they could target A11X chips, and retest any rejects at lower A11 specs.
A process shake out on a low volume chip makes lots of sense. However Apple wouldn't go there unless there was a significant advantage to doing so.
As for a more robust iPad, there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development. The obvious reason being that iPad simply requires far more performance. This seems to have been put on the back burner probably because the process advances have been so quick and so huge that both hardware platforms could be supported by one architecture. With process shrinks hitting the wall we could see split with the processors going in a different direction.
The main reason for doing this would be to leave the A10 with current process to a new iteration of the regular non pro Ipad (right now on A8X) and put the much lower volume PRO one process ahead to further differentiate it. Could put a A11X in the Pro, the A11 in the Iphone 8 and next year the A11X winds up in the regular Ipad
Got the feeling they'll make the regular focus on thin, very light, rigid, say 4-5mm and slightly bigger screen (because of less bezel).
Damned if you do...
Let's assume Apple has the capability to build [at least 1 model] of the [early] 2017 iPad Pro with a 10nm A11X -- and that the resulting iPad Pro has RAM, speed, bandwidth, connectivity, etc. that is comparable to the 2016 MacBooks.
What OS will it run?
Isn't iOS too limited for such a beast? Isn't macOS too legacy/bloated with Mac baggage?
As @wizard69 said "there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development."
If true, wouldn't Apple also have investigated taking iPad OS development in a different direction than iPhone OS [iOS] development?
After all we have specialized watchOS and tvOS -- why not iPadOS, or even better proOS?
Wouldn't it make sense to take the best features of iOS and macOS and create a new lean and mean OS especially for the 2017 [and beyond] Pro users of the iPad?
Here's the Damned if you do...
Wouldn't these new Pro iPad/proOS be hot sellers for all kinds of creation activities, e.g. development, AV editing, graphics, drafting/CAD, writing, etc.?
I'd be willing to bet that IBM would be all over [and supplied a lot of input to] these for its MobileFirst and Cloud Services offerings.
Would Apple have the chutzpah to do this?
If they did, could they manufacture enough A11Xs to meet the demand?
I am not chip expert but I would agree. It would be strange when Apple would make A10 in 14nm and A10X in 10nm. A11 is expected to be 10 nm.
IDK, Since Apple is reportedly going to use a 10nm A11in next years iPhone, it may be worthwhile to shake out the 10 nm manufacturing process with a lower volume chip like the A10X.
Or, the could even be building an A11X at 10nm for a much more robust iPad Pro.
Then there's this: If the A11X chip is a superset of the A11 chip (same footprint, more cores, faster clocking, etc.) -- they could target A11X chips, and retest any rejects at lower A11 specs.
A process shake out on a low volume chip makes lots of sense. However Apple wouldn't go there unless there was a significant advantage to doing so.
As for a more robust iPad, there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development. The obvious reason being that iPad simply requires far more performance. This seems to have been put on the back burner probably because the process advances have been so quick and so huge that both hardware platforms could be supported by one architecture. With process shrinks hitting the wall we could see split with the processors going in a different direction.
The main reason for doing this would be to leave the A10 with current process to a new iteration of the regular non pro Ipad (right now on A8X) and put the much lower volume PRO one process ahead to further differentiate it. Could put a A11X in the Pro, the A11 in the Iphone 8 and next year the A11X winds up in the regular Ipad
Got the feeling they'll make the regular focus on thin, very light, rigid, say 4-5mm and slightly bigger screen (because of less bezel).
Damned if you do...
Let's assume Apple has the capability to build [at least 1 model] of the [early] 2017 iPad Pro with a 10nm A11X -- and that the resulting iPad Pro has RAM, speed, bandwidth, connectivity, etc. that is comparable to the 2016 MacBooks.
What OS will it run?
Isn't iOS too limited for such a beast? Isn't macOS too legacy/bloated with Mac baggage?
As @wizard69 said "there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development."
If true, wouldn't Apple also have investigated taking iPad OS development in a different direction than iPhone OS [iOS] development?
After all we have specialized watchOS and tvOS -- why not iPadOS, or even better proOS?
Wouldn't it make sense to take the best features of iOS and macOS and create a new lean and mean OS especially for the 2017 [and beyond] Pro users of the iPad?
Here's the Damned if you do...
Wouldn't these new Pro iPad/proOS be hot sellers for all kinds of creation activities, e.g. development, AV editing, graphics, drafting/CAD, writing, etc.?
I'd be willing to bet that IBM would be all over [and supplied a lot of input] to these for its MobileFirst and Cloud Services offerings.
Would Apple have the chutzpah to do this?
If they did, could they manufacture enough A11Xs to meat the demand?
I'm of the opinion that even an A11X at 10 nm still isn't going to have the performance that some, or even most, creatives are waiting for, so iOS is fine for the time being. What does happen when the performance improves closer to an MBP, is that developers that have big plans for iPad with desktop class/cloud based applications, will be stuck without a traditional desktop file management system. So an iPad OS would need to be a superset of iOS, with some desktop underpinnings, but then you run into the issue of the pointer when using the current iPad keyboard, making the addition of a trackpad and touchbar a possible solution (similar in requirement to future iMac's).
iCloud will have to be beefed up with more server side development tools as well.
I expect that Apple is and has been working on solutions, but they are overly cautious about mixing iOS and Mac OS, so a couple of years seems the earliest. I would also expect at that time, some larger iPad Pro's with large die SOC's capable of higher TDP's approaching MBP performance, at least under external power.
I'm of the opinion that even an A11X at 10 nm still isn't going to have the performance that some, or even most, creatives are waiting for, so iOS is fine for the time being.
You may be right -- likely, it depends on number cores/GHz/GPUs, memory bandwidth, amount of RAM, SSD, connectivity.
What does happen when the performance improves closer to an MBP, is that developers that have big plans for iPad with desktop class/cloud based applications, will be stuck without a traditional desktop file management system. So an iPad OS would need to be a superset of iOS, with some desktop underpinnings, but then you run into the issue of the pointer when using the current iPad keyboard, making the addition of a trackpad and touchbar a possible solution (similar in requirement to future iMac's).
Agreed! Definitely need a KB/Cursor -- don't know about a touch bar.
iCloud will have to be beefed up with more server side development tools as well.
Yes, definitely!
I expect that Apple is and has been working on solutions, but they are overly cautious about mixing iOS and Mac OS, so a couple of years seems the earliest. I would also expect at that time, some larger iPad Pro's with large die SOC's capable of higher TDP's approaching MBP performance, at least under external power.
Good points! But a new/different iPad OS (proOS) mitigates the issuesabout mixing iOS and Mac OS -- and points to a future wherethe OS matches/adjusts to the uses/users of the device -- rather than the other way around.
IMO, the timing of this offering is everything... hoping that in early 2017 the necessary pieces will converge.
Good points! But a new/different iPad OS (proOS) mitigates the issuesabout mixing iOS and Mac OS -- and points to a future wherethe OS matches/adjusts to the uses/users of the device -- rather than the other way around.
IMO, the timing of this offering is everything... hoping that in early 2017 the necessary pieces will converge.
I like that about the OS adjusting to the user.
I'm a daily user of MCAD and CAM software, and Autodesk is about 6 months away from delivering a browser based Fusion 360. It's kind of a big deal, as it opens up iOS and the iPad Pro as first class design and manufacturing tools. I don't think that Apple is going to give up the pace of advance and their advantage in the tablet space, and expect that there won't be a lack of resources in this respect. I'm hopeful that the A11X at 10nm does arrive in iPad Pro's this spring to mark another milestone for iPad and that we gain some of the file management features at WWDC.
Good points! But a new/different iPad OS (proOS) mitigates the issuesabout mixing iOS and Mac OS -- and points to a future wherethe OS matches/adjusts to the uses/users of the device -- rather than the other way around.
IMO, the timing of this offering is everything... hoping that in early 2017 the necessary pieces will converge.
I like that about the OS adjusting to the user.
Yeah, maybe they should call it myOS or adaptiveOS... but seriously, we have a different mixture of devices than we had when OS X (2000) or iOS (2007) were introduced -- likely, it is time to rethink what/how/when the OS provides capabilities. iOS was revolutionary because:
it was intuitive
it got out of the way (between the user and what he wanted to accomplish)
But, today, the iPad user wants to do so much more -- and Apple doesn't seem to be able to enhance iOS fast enough to satisfy [Prosumer or Pro] iPad users without detracting from other iDevice needs.
Also, today, we have a [mostly] stable Swift system programming language which could be used to deliver modular capability to myOS -- that, in fact would allow it to adapt to the user.
I'm unsure if it would be to Apple's advantage to open-source myOS or adaptiveOS.
I'm a daily user of MCAD and CAM software, and Autodesk is about 6 months away from delivering a browser based Fusion 360. It's kind of a big deal, as it opens up iOS and the iPad Pro as first class design and manufacturing tools. I don't think that Apple is going to give up the pace of advance and their advantage in the tablet space, and expect that there won't be a lack of resources in this respect. I'm hopeful that the A11X at 10nm does arrive in iPad Pro's this spring to mark another milestone for iPad and that we gain some of the file management features at WWDC.
Exactly! I think that Apple's new File System and use of FoundationDB in iCloud are steps in this direction.
Also, who says an An APU has to be released before the corresponding AnX APU. As previously discussed, there may be technical and manufacturing advantages to releasing the AnX APU first. Also, it allows Apple to release products like iPads in the Spring and set the stage for iPhones in the Fall.
Comments
very costly (1) in Billion dollar adventure and not short time solution - long time commitment, when you deal with 10 and 7 nm, tooling and testing are all need to be developed, commissioned, round robin standardized, spc run chart established...(2) not easy to get the set of skills in Semi, unless you try to buy a whole team from some outfit - you can't improve process what you don't really know, large FAB if not consider carefully, you might end up magnify the problem many magnitude ... better partner with someone. it is a learning curve for 10 nm technology...should stay with 14 nm. until it is matured... IMHO.
Or, the could even be building an A11X at 10nm for a much more robust iPad Pro.
Then there's this: If the A11X chip is a superset of the A11 chip (same footprint, more cores, faster clocking, etc.) -- they could target A11X chips, and retest any rejects at lower A11 specs.
I give this rumor 10 "mehs" out of a possible 10.
If Apple isn't after more performance the other thing that they might be after is far lower power usage which can lead to a thinner battery or more power for the display (OLED). I'm a bit up in the air as to the truth in this report, I do know from other sources that TSMC is basically going all in to get 10 nm out the door in 2017. iPad is a good place to introduce such tech as it is far lower volume thus far lower risk and a good place to ramp up technology.
As for why they would scale the existing architecture to 10 nm, there are lots of good reasons to do so. First the existing architecture is pretty impressive, depending upon which way they go to optimize it for a 10X, Apple could see a huge boost in performance. A shrink, that allows for faster operation, would boost performance especially if at the same time they could increase the size of internal caches and the speed of external memory interfaces.
Lets face it every X variant has had enhancements to one extent or another over the non X variant. A shrink not only offers Apple the speed increase for the watt expended it also offers Apple considerably more die space. Die space means more hardware. More hardware means more specialized function blocks. A quick list of potential die space use beyond processor improvements:
- Enhanced GPUs. Basically more execution units.
- Enhanced video decoders and encoders. Here we can support more codecs in hardware and increase quality.
- Enhanced ports (eventually iPad will need USB-C class ports).
- Support for AI in hardware. Lots of interesting approaches here but this could be integrated into either the GPU or the CPU complexes. Or Apple could implement an entirely separate processor core for AI acceleration.
- Move the RF processing on board the die.
- Much larger caches.
- New memory interfaces to high performance memory.
Tha tis seven off the top of my head. I'm sure others can add to the list. The point is there is a lot Apple could do with the die space that has little to do with the CPU itself. Personally I think the two big ones that Apple absolutely has to be looking at are RF processing on die and new AI hardware. RF processing just gets rid of an external chip and saves power and space. AI tech (really machine learning tech which I don't consider artificial intelligence), is coming on strong with AMD, Nvidia and Intel all taking their own approaches to the problem of AI acceleration. Apple needs an AI solutions that they can sheared along with the own development vectors, so I wouldn't be surprised to see something tacked onto an A10 to make it an A10X even though big step like this is more A11. The problem is Apple can't afford to get too far behind here.As for a more robust iPad, there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development. The obvious reason being that iPad simply requires far more performance. This seems to have been put on the back burner probably because the process advances have been so quick and so huge that both hardware platforms could be supported by one architecture. With process shrinks hitting the wall we could see split with the processors going in a different direction.
https://en.wikipedia.org/wiki/Die_shrink
"The term die shrink (sometimes optical shrink or process shrink) refers to a simple semiconductor scaling of semiconductor devices, mainly transistors. The act of shrinking a die is to create a somewhat identical circuitry using a more advanced fabrication process, usually involving an advance of lithographic node. This reduces overall costs for a chip company, as the absence of major architectural changes to the processor lowers research and development costs, while at the same time allowing more processor dies to be manufactured on the same piece of silicon wafer, resulting in less cost per product sold."
Why would Apple do a simple die shrink when they have the resources to re-architect the SOC for even greater performance? Apple certainly has the volume to justify that.
The current iPad Pro is still on A9X. And the A10 ( by the way it is 16nm, not 14nm ) in iPhone 7 is already faster then A9X in many benchmarks. An A10x, even if it is a die shrink of A10, will be pretty damn fast, because the iPad has much higher TDP to work with, along with the optimization and gain from 10nm.
It is basically like Intel's Tick Tock, except Apple is doing it in a MUCH faster cycle. ( Note: Intel's Tock is a full node transition, TSMC is smaller incremental gains half node.)
Got the feeling they'll make the regular focus on thin, very light, rigid, say 4-5mm and slightly bigger screen (because of less bezel).
Damned if you do...
Let's assume Apple has the capability to build [at least 1 model] of the [early] 2017 iPad Pro with a 10nm A11X -- and that the resulting iPad Pro has RAM, speed, bandwidth, connectivity, etc. that is comparable to the 2016 MacBooks.
What OS will it run?
Isn't iOS too limited for such a beast? Isn't macOS too legacy/bloated with Mac baggage?
As @wizard69 said "there was a rumor a couple of years ago that Apple was going to take iPhone chip development in a different direction than iPad chip development."
If true, wouldn't Apple also have investigated taking iPad OS development in a different direction than iPhone OS [iOS] development?
After all we have specialized watchOS and tvOS -- why not iPadOS, or even better proOS?
Wouldn't it make sense to take the best features of iOS and macOS and create a new lean and mean OS especially for the 2017 [and beyond] Pro users of the iPad?
Here's the Damned if you do...
Wouldn't these new Pro iPad/proOS be hot sellers for all kinds of creation activities, e.g. development, AV editing, graphics, drafting/CAD, writing, etc.?
I'd be willing to bet that IBM would be all over [and supplied a lot of input to] these for its MobileFirst and Cloud Services offerings.
Would Apple have the chutzpah to do this?
If they did, could they manufacture enough A11Xs to meet the demand?
iCloud will have to be beefed up with more server side development tools as well.
I expect that Apple is and has been working on solutions, but they are overly cautious about mixing iOS and Mac OS, so a couple of years seems the earliest. I would also expect at that time, some larger iPad Pro's with large die SOC's capable of higher TDP's approaching MBP performance, at least under external power.
You may be right -- likely, it depends on number cores/GHz/GPUs, memory bandwidth, amount of RAM, SSD, connectivity.
Here's an interesting comparison:
http://www.laptopmag.com/articles/microsoft-surface-pro-4-vs-ipad-pro
Agreed! Definitely need a KB/Cursor -- don't know about a touch bar.Yes, definitely!
Good points! But a new/different iPad OS (proOS) mitigates the issues about mixing iOS and Mac OS -- and points to a future where the OS matches/adjusts to the uses/users of the device -- rather than the other way around.
IMO, the timing of this offering is everything... hoping that in early 2017 the necessary pieces will converge.
I'm a daily user of MCAD and CAM software, and Autodesk is about 6 months away from delivering a browser based Fusion 360. It's kind of a big deal, as it opens up iOS and the iPad Pro as first class design and manufacturing tools. I don't think that Apple is going to give up the pace of advance and their advantage in the tablet space, and expect that there won't be a lack of resources in this respect. I'm hopeful that the A11X at 10nm does arrive in iPad Pro's this spring to mark another milestone for iPad and that we gain some of the file management features at WWDC.
But, today, the iPad user wants to do so much more -- and Apple doesn't seem to be able to enhance iOS fast enough to satisfy [Prosumer or Pro] iPad users without detracting from other iDevice needs.
Also, today, we have a [mostly] stable Swift system programming language which could be used to deliver modular capability to myOS -- that, in fact would allow it to adapt to the user.
I'm unsure if it would be to Apple's advantage to open-source myOS or adaptiveOS.
Exactly! I think that Apple's new File System and use of FoundationDB in iCloud are steps in this direction.
Also, who says an An APU has to be released before the corresponding AnX APU. As previously discussed, there may be technical and manufacturing advantages to releasing the AnX APU first. Also, it allows Apple to release products like iPads in the Spring and set the stage for iPhones in the Fall.