Apple A9X the new mobile SoC king

[witeken]

"Lies, damned lies, and statistics".

Apparently, TSMC lost the N16 battle in 2014.

http://semiengineering.com/whos-winning-the-finfet-foundry-race/

Intel lost the 14nm battle only half a year or so before HVM was supposed to begin.

 

[carop]

That's an old article that put Global Foundries in the FinFet lead. So that tells you how accurate it was. TSMC went the more conservative route to FinFet and are on their second generation.

I provided a link to the article because the battle methaphor is from Morris Chang himself:

“The looming battle is really not 16nm,” said Morris Chang, chairman of TSMC, in a recent conference call. “In my mind, the 16nm battle has already been fought… My thought (is) primarily on 10nm, not so much on 16nm.”

I know some people on first term basis who ported their N28 planar bulk libraries to the TSMC N16 FinFET flow. So, I have some idea as to why TSMC N16 FinFET is in its second generation.

The following statement was also made by Morris Chang in July 2014:

Morris Chang, chairman of foundry Taiwan Semiconductor Manufacturing Co. Ltd. (Hsinchu, Taiwan), has told analysts that his company would likely fall behind a "major competitor" in foundry market share for 16nm/14nm FinFET node in 2015, but would then go on to be market leader at 16nm/14nm in 2016 and subsequent years.

http://electronics360.globalspec.com/article/4380/tsmc-to-fall-behind-rivals-in-finfet-market-share

Yes, it is a July 2014 article. And yet, it is important because I am not sure if you realize that the N10 TSMC and Samsung battle has already begun. A Skirmish here a skirmish there...Should be raging next year. By the time you look back in, say 2017, the battle will long be over.

 

[catinthefurnace]

All statistics has to assume a random sample, otherwise any and all data is useless.

As far as the sample being biased, it is a possibility, but unless there is actually any evidence of it, the normal thing would be to assume that it is not.

Do you care you revise your statements regarding statistics now that the results have completely flipped?

http://demo.hiraku.tw/CPUIdentifier/

You were so sure you had enough of a sample to make a definitive conclusion.

The problems I pointed out are very real, and rather than argue the specifics of statistical theory with you two, I decided to wait. Statistics can be useful, but too often ignorance of the math behind statistics leave people unable to identify its limitations.

Idontcare acknowledged that the share might change over time as Samsung's yields get better, but I think we can all agree that the phones being tested at this time are all still "launch phones", and not enough time has passed for changes in yields to likely make any difference in distribution. We'll know more in a month or two as the 2nd and third waves of shipments arrive.

I agree with Idontcare that none of this really matters much, but I just get annoyed when people overestimate the powers of statistics. Statistics can add to certainty, but the do not create certainty. Especially with a small sample size where no information is known about the distribution.

 

[Nothingness]

And now a test that shows that Samsung A9 has a better battery life than TSMC A9.

http://www.tomshardware.com/news/iphone-6s-a9-samsung-vs-tsmc,30306.html

 

[Rakehellion]

Do you care you revise your statements regarding statistics now that the results have completely flipped?

http://demo.hiraku.tw/CPUIdentifier/

You were so sure you had enough of a sample to make a definitive conclusion.

The problems I pointed out are very real, and rather than argue the specifics of statistical theory with you two, I decided to wait. Statistics can be useful, but too often ignorance of the math behind statistics leave people unable to identify its limitations.

Idontcare acknowledged that the share might change over time as Samsung's yields get better, but I think we can all agree that the phones being tested at this time are all still "launch phones", and not enough time has passed for changes in yields to likely make any difference in distribution. We'll know more in a month or two as the 2nd and third waves of shipments arrive.

I agree with Idontcare that none of this really matters much, but I just get annoyed when people overestimate the powers of statistics. Statistics can add to certainty, but the do not create certainty. Especially with a small sample size where no information is known about the distribution.

The iPhone came out a month ago, so it's possible we're on a completely different batch. The numbers will likely change over time. In a few more months, both manufacturers might be on a completely different process.

 

[Rakehellion]

And now a test that shows that Samsung A9 has a better battery life than TSMC A9.

http://www.tomshardware.com/news/iphone-6s-a9-samsung-vs-tsmc,30306.html

I think all of those tests are a little shaky. A better idea would be to compare 5 TSMC iPhones with 5 Samsung iPhones and average the results. Any individual manufactured good is susceptible to variations in performance.

 

[stingerman]

I provided a link to the article because the battle methaphor is from Morris Chang himself:

I appreciate the link for the sake of the quotes. But, the quotes were in the context of competitor claims that did not materialize. We have the advantage of hindsight to see that TSMC was unwittingly conservative.

 

[catinthefurnace]

The iPhone came out a month ago, so it's possible we're on a completely different batch. The numbers will likely change over time. In a few more months, both manufacturers might be on a completely different process.

Excuse me but the iPhone 6s came out just 2.5 weeks ago, not a month.

 

[Rakehellion]

Excuse me but the iPhone 6s came out just 2.5 weeks ago, not a month.

And that invalidates my point how?

 

[joshhedge]

I think all of those tests are a little shaky. A better idea would be to compare 5 TSMC iPhones with 5 Samsung iPhones and average the results. Any individual manufactured good is susceptible to variations in performance.

That would still be a statistically useless test. Look at it this way, you're trying to find a correlation between battery life and SoC manufacturer with a sample size of 10 for a phone that sells tens of millions of units. ANY battery life comparisons that are made by reviewers or otherwise would be unreliable unless a vastly greater sample size was used. Apple are liable to be sued should battery life for the average user deviate significantly from the press release they issued. It would be a dreadful business decision for that to happen. I would implore you and others to take Apples word as they are the only ones with the ability to implement methods of testing with a meaningful sample size.

 

[ThatBuzzkiller]

This is not solid evidence but I don't think TSMC is producing Apple A9 processors or at the very least in significant numbers ...

I just checked TSMC's recent Q3 report and strangely enough they lumped in the revenue of the 16nm node with the 20nm node ...

The revenue share for 20/16nm only increased by a single percent compared to the previous quarter and their NET REVENUE only increased by 3%! These are fairly stable numbers but what's more is that last quarter was more profitable this quarter!

These numbers seem to suggest that TSMC is not a big supplier of the Apple A9 processors or even a supplier at all as it's reflected on their quarterly financial report ...

This is supposed to be relatively the most profitable time when chip manufacturers are producing parts for high volume products like iPhones but if they really are making those Apple A9's where exactly is the rest of revenue or profits ?

Now, I know I haven't accounted for the lost revenue for 20nm to 16nm revenue but this is even more rich going by their monthly revenue report. For this September they reported a 13.8% decrease compared to last September when it came to revenue so it's as if TSMC didn't get ANY order from Apple!

 

[MongGrel]

NM, if a mobile thing I'm not getting involved.

My bad.

 

[Nothingness]

This is not solid evidence but I don't think TSMC is producing Apple A9 processors or at the very least in significant numbers ...

I just checked TSMC's recent Q3 report and strangely enough they lumped in the revenue of the 16nm node with the 20nm node ...

The revenue share for 20/16nm only increased by a single percent compared to the previous quarter and their NET REVENUE only increased by 3%! These are fairly stable numbers but what's more is that last quarter was more profitable this quarter!

These numbers seem to suggest that TSMC is not a big supplier of the Apple A9 processors or even a supplier at all as it's reflected on their quarterly financial report ...

This is supposed to be relatively the most profitable time when chip manufacturers are producing parts for high volume products like iPhones but if they really are making those Apple A9's where exactly is the rest of revenue or profits ?

Now, I know I haven't accounted for the lost revenue for 20nm to 16nm revenue but this is even more rich going by their monthly revenue report. For this September they reported a 13.8% decrease compared to last September when it came to revenue so it's as if TSMC didn't get ANY order from Apple!

Are you really trying to say the data collected by the TSMC vs Samsung is completely wrong?

From the earning call transcript:

Donald Lu - Goldman Sachs - Analyst
Okay. My second question is on 16-nanometer ramp. Since you talk about it's going to be faster than 20-nanometer, would that imply it will be
over 22% I believe in Q1 next year of your total revenue?
C.C. Wei - Taiwan Semiconductor Manufacturing Co Ltd - President & Co-CEO
Donald, our 16 FinFET right now, most of the volume come from one single customer. So it will be very difficult for me to give you the exact number
and say whether -- that it's a what percentage. Otherwise I release too much of information.

So it's very possible that TSMC produces a lot of chips for Apple. Then the 16/20nm lower than expected revenues could be due to quickly declining 20nm and/or poor 16nm yields. Or even simplier: the lack of sequential growth is just because Apple has been stockpiling chips since previous quarter.

 

[ThatBuzzkiller]

Are you really trying to say the data collected by the TSMC vs Samsung is completely wrong?

From the earning call transcript:
So it's very possible that TSMC produces a lot of chips for Apple. Then the 16/20nm lower than expected revenues could be due to quickly declining 20nm and/or poor 16nm yields. Or even simplier: the lack of sequential growth is just because Apple has been stockpiling chips since previous quarter.

Having poor yields on 16nm when they are saying that it has a faster ramp than 20nm is practically a contradiction ...

As for declining 20nm revenue, that is to be expected since Apple isn't ordering A8 and A8X processors anymore but if you took a look at the monthly revenue reports they stopped seeing a lot of revenue gains at the end of Q2 yet 20/16nm revenue share has slightly increased so it's unlikely that 20nm saw a sharp drop in revenue ...

Stock piling months before a new device hits ? That's not how it works and even then they didn't report any 16nm revenue for Q1 or Q2 so I ask again. What exactly is TSMC producing on their 16nm node that justifies such small contributions ?

Matter of fact why do they only have an 8% gain in revenue for 20/16nm compared to last quarter ? If TSMC got $50 for every processor they made for Apple it would amount to less than the rumored 30% allocation ...

 

[Rakehellion]

That would still be a statistically useless test. Look at it this way, you're trying to find a correlation between battery life and SoC manufacturer with a sample size of 10 for a phone that sells tens of millions of units. ANY battery life comparisons that are made by reviewers or otherwise would be unreliable unless a vastly greater sample size was used.

Except the tolerances for a high tech piece of electronics are precise enough that such a large sample size would be useless. Every device is tested before it leaves the factory and is guaranteed to be within a certain level of performance.

 

[Cakefish]

How would the A9X fare with the rumoured 2334 x 3112 resolution of the iPad Air 3? That's 2.31x the pixels of the Air 2 (7.26 million pixels vs 3.14 million pixels). Apple claims that the A9X has 2x the GPU power of the A8X.

 

[Mondozei]

Regarding the Samsung vs TSMC issue, from Beyond3D

I've just had my first good look at both chips. There are significant layout, utilisation and library-level differences between the two variants, and no two layout blocks are identical. So there's a lot more at play than pure transistor manufacturing between the two. I know that's kind of obvious from the low-res imaging that Chipworks put out there, where you could see some differences, but it's worth pointing out they're not minor.

From this thread.

By the way, do we have a database yet where people crowdsource battery tests to get a decent databank. I think the kind of tests that Tom's did or that Chinese site did are useless, because the amount of phones we need are in the hundreds if not thousands to get any valuable insight.

 

[stingerman]

Are you really trying to say the data collected by the TSMC vs Samsung is completely wrong?

From the earning call transcript:
So it's very possible that TSMC produces a lot of chips for Apple. Then the 16/20nm lower than expected revenues could be due to quickly declining 20nm and/or poor 16nm yields. Or even simplier: the lack of sequential growth is just because Apple has been stockpiling chips since previous quarter.

Don't forget that accounting rules wouldn't allow them to count the revenue until Apple received and/or accepted the part. We'd have to read the contract terms. So TSMC would have been storing the parts ahead of Apple using them, thus creating the cost without the associated revenue, since Apple's pre-payment would have been considered an accounting liability.

Thus the revenues would start getting counted as the processors were shipped and only the ones that were shipped. However, TSMC would be eating all the fixed costs, administrative fees, etc on paper. My point is that you won't see the beneficial aspects of the sale to Apple for a couple of Quarters. The first few months will just be making up the costs, and depending on their margins, it may take longer.

What you really need to look at is their cash flow, cash and liability sections, ignore the profit and revenue.

 

[antihelten]

Do you care you revise your statements regarding statistics now that the results have completely flipped?

http://demo.hiraku.tw/CPUIdentifier/

You were so sure you had enough of a sample to make a definitive conclusion.

The problems I pointed out are very real, and rather than argue the specifics of statistical theory with you two, I decided to wait. Statistics can be useful, but too often ignorance of the math behind statistics leave people unable to identify its limitations.

Idontcare acknowledged that the share might change over time as Samsung's yields get better, but I think we can all agree that the phones being tested at this time are all still "launch phones", and not enough time has passed for changes in yields to likely make any difference in distribution. We'll know more in a month or two as the 2nd and third waves of shipments arrive.

I agree with Idontcare that none of this really matters much, but I just get annoyed when people overestimate the powers of statistics. Statistics can add to certainty, but the do not create certainty. Especially with a small sample size where no information is known about the distribution.

Why in the world would I revise my statements, when it was perfectly accurate then and still perfectly accurate now.

I said that the sample size was not an issue as far as statistical analysis is concerned (and it wasn't), and furthermore that statistical analysis has to assume a random sample (if your sample isn't random then you generally can't perform statistics on it).

Given the shift in numbers, it would appear that the sample was in fact not random, and thus statistical analysis was in hindsight impossible. More importantly if the numbers were non-random initially, then they are almost certainly also non-random now, and thus the data is still useless, as far as statistical analysis is concerned.

Of course as others has mentioned, the truth is probably that the numbers are in fact random, but also time-dependent (currently the time dependent factor would be batches and yields could most certainly have changed from launch, remember it might still be early days from the consumer perspective, but from the foundry perspective they have been at it for quite a while). This means that the data should really be analysed as a time series, which unfortunately isn't possible given the way the data is formatted on the web site.

So currently the only option would be for someone to either constantly monitor changes in the numbers on the website to ascertain the time factor, or alternatively wait until the differences in batches have evened out and are no longer a significant influence.

This has nothing to do with overestimating statistics, it has to do with actually understanding how statistics and sample sizes work.

 

[sxr7171]

Apple has created a true powerhouse for the iPad Pro. A9X (quad core) will beat Skylake Core M (dual core with HT) without a doubt in CPU performance. GPU performance comparison between the two remains to be seen.

http://www.anandtech.com/show/9615/apple-announces-the-ipad-mini-4

"According to Apple A9X is 1.8x faster than A8X at CPU tasks, and 2x faster at GPU tasks. Given that A8X was already the fastest ARM SoC in a mobile device this is quite an accomplishment, and is likely the result of architectural improvements, higher clock speeds, and possibly the addition of a fourth CPU core. We'll have to wait until we get our hands on the iPad Pro before any of this can be confirmed though. Apple also noted that A9X is built on a new "transistor architecture" which means it's being fabricated on either Samsung's 14nm or TSMC's 16nm FinFet process."

Apple has gone all out with A9X and my guess is the A9X is > 150 sq mm. Skylake Core M die size is just below 100 sq mm.

http://www.anandtech.com/show/9582/intel-skylake-mobile-desktop-launch-architecture-analysis

Apple is challenging Intel to bring true quad cores to Core M. Will we see Intel bring true quad cores to Core M next year with Kaby lake or at 10nm with Cannonlake ? It would be great if Apple can bring simultaneous hyperthreading next year to their custom ARMv8 core . That would be awesome competition. A10X vs Kaby lake (4C,8T).

A multithreaded score of almost 8000 assuming Apple get close to 1.8x improvement over A8X in Geekbench 3 is mind boggling. For comparison Skylake Core M gets around 5800

http://browser.primatelabs.com/geekbench3/2637682

http://browser.primatelabs.com/geekbench3/search?utf8=✓&q=a8x

This is going to be interesting. iPad Pro is going to be faster than the latest Macbook Air (even with Skylake Core M). Is the Macbook Air going to be cannibalized by iPad Pro. I think yes. Apple gets to make better margins as A9X is their custom ARMv8 design fabbed at TSMC (most likely) while Core M is Intel's product.

http://www.theverge.com/2015/9/8/9275157/apple-macbook-air-pro-retina-redesign-phase-out

We knew the new Macbook would be weak (BTW it's not Air anymore) but it performs decently - much better than the Yoga 3 which was disappointingly slow with its Core M.

The iPad pro may be faster but you couldn't pay me to use a tablet for more than 30 minutes. The form factor is perfect for certain short term uses but the regular iPads and even the mini are the antithesis of good ergonomic design.

On the couch holding a tablet up with one hand while using it with another is not fun. In bed holding it up with one hand and using it with another is something I can't do for more than 5 minutes without my hand hurting. On a table unless it has a stand or carry a stand with it it will lay flat on a table which is murder on your neck if you use it that way.

The new Macbook rests on your lap on the couch. If you put your feet up on an ottoman you have a wide range of where you can rest it relative to your eyes. In bed it rests easily on your chest. You will use one hand not to bear it weight but simply to stabilize it while your chest handles the weight. You can also very comfortably have the screen within 10-12" from your eyes - which makes a 12-13" screen occupy a greater field of view than a piddly 21.5" iMac.

On a table obviously the Macbook can rest on it and tilt the screen to your desire. The best part is - you're not holding it the whole time.

The iPad pro may be faster but a laptop design is more usable in almost every single situation. Even content consumption.

This also brings me to the point of stupid designs like the new surface book and all tablets with flimsy detachable keyboards. They are heavy and bulky when you need or carry the keyboard. The Macbook weighs around 2lbs but keyboard and all instead of 3.34lbs for the surface book. Of course that's a little apples and oranges - but keep in mind there are so many great products with full laptop CPUs starting with the Lavie Z at 1.72lbs, the Dell XPS 13, the Thinkpad X1 and so many more that keep under 3lbs including a keyboard. Any serious use requires a keyboard and no onscreen keyboard seems to change that.

iPad sales hit a peak and are on a steady decline after people realized they are pretty much only good to watch movies on flights assuming you have that case/stand - which makes it more heavy. I know people personally who replaced their laptops years ago with an iPad and they claimed to love it - the idea of a $499 thing replacing a laptop was great. I knew they were suffering the whole time. Then when the Macbook came out some of them switched immediately and their iPads, as every single one I've ever bought are serving are dust catchers/drink coasters. Not one who switched thought that using an iPad was more comfortable - not even for the most basic browsing or content consumption.

Apple wasting their time on a declining product category is irritating. I would have rather they used the resources to shave even one month development off the new Apple TV than create this iPad pro. IMHO the development and release of this product was a huge waste of time.

 

[stingerman]

The iPad pro may be faster but you couldn't pay me to use a tablet for more than 30 minutes. The form factor is perfect for certain short term uses but the regular iPads and even the mini are the antithesis of good ergonomic design.

I've used older iPads with an old Apple bluetooth keyboard and it's amazing how useful they are as a notebook replacement. Keyboards and Pens have been part of the iPad 3rd party eco system from nearly the start. The iPad Pro formalizes what many iPad users have been doing all along.

The only issue I have open with using the iPad Pro is: How will it work with an external display. I really enjoy my 27" display when I'm at my office. My gut feeling is Apple will address this with the next revision, but who knows if there is an adapter cable in the works...

 

[sxr7171]

I've used older iPads with an old Apple bluetooth keyboard and it's amazing how useful they are as a notebook replacement. Keyboards and Pens have been part of the iPad 3rd party eco system from nearly the start. The iPad Pro formalizes what many iPad users have been doing all along.

The only issue I have open with using the iPad Pro is: How will it work with an external display. I really enjoy my 27" display when I'm at my office. My gut feeling is Apple will address this with the next revision, but who knows if there is an adapter cable in the works...

Read this: http://www.technologytell.com/apple/143054/ipad-mouse-support-is-desperately-needed/

Apparently I needn't have written all that, I could have just cited "Gorilla Arm".

What I'm saying also is that any iPad plus even the lightest keyboard is pretty heavy (0.84lbs) There is no weight argument to be made in a world where 13" laptops weigh as little as 1.72 lbs. It is not a good notebook replacement when it cannot even be rested on a lap. It needs a stand to even rest on a table in way that a human being can see the screen. You need to carry both a stand and a keyboard for it.

But maybe there are a small number of users who must have iOS over Mac OS or Win 10 and I suppose the product will fill that very tiny IMHO niche. Especially in a world with $200 Chromebooks.

Apple has had some very boneheaded restrictions on iOS that even this "pro" version doesn't address. Where is the mouse support? There is a huge difference between having to input on the screen on the device or having the option to have input from a distance with a mouse. It is disgraceful for them to release a "pro" device without that capability that every other OS offers.

To your point about displays the fact Apple that does not support Airplay to and from Macs and iOS devices is also very stupid. This prevents me from using my 27" Retina iMac as a display for any iOS device. Add to that the next Apple TV won't support higher than 1080p precludes any use of a 4k or 5k external display for iOS devices. It might have been nice to Airplay iOS retina content to an Apple TV. Let's not forget that wired connections to the iMac Retina are not possible because it does not work in target display mode at all. Couple all that with the lack of a 5k cinema display and I don't understand what method there is for an external display beyond 1080p Airplay or wired via displayport adapter to a legacy display or iMac.

On top of all this you have people going ga-ga over Microsoft Continuum like it's the next coming. I laughed to learn that it still relies on a wired connection to the display. This is because for last 3 years or so I've been hearing Miracast, WiDi nonsense and supposedly my laptop supports this technology but not once have I seen it actually work.

Meanwhile Apple has perfected wireless display, it works amazingly well with the Apple TV. I have no doubt that the technology for Airplay up to 5k is present in modern Macs and even iOS devices. Airplay at 1080p has been functioning for 3 years on phones with hardware several magnitudes lower than what we have now.

Through artificial restrictions such as no mouse drivers and lack of Airplay to Macs they have gimped this supposed "pro" iPad and for that matter all iPhones and iPads. If those features were enabled it would be basically Continuum with wireless display. Which of course makes it far superior to Continuum. Even with Continuum developers have to design and implement this mouse driven UI on a per app basis - it isn't the full Win 10 UI at all. There is no reason an iPhone could not just be in your pocket while you sitting in front of your iMac and using it with your existing wireless mouse and keyboard. Very boneheaded by Apple. It could have been a Continuum killer.

In observation the best use for iPads and tablets in general is commercial. They are great in restaurants for wait staff to use in table selection and order taking or in gimmicky places for customers to pick wines or menu items. There are great anywhere a business needs to interface with customers such as check in at doctor's offices or self serve kiosks of any kind. They are an amazing tool for lots of commercial applications and we'll be seeing them at more businesses. They are best used for short periods by any one person.

The Gorilla Arm issue is real and it explains the declines in the tablet year over year over year. Apple just managed to make tablets gimmicky enough again that millions of fools (myself included) bought them. Lets face it for many decades we've heard about handwriting recognition or voice recognition as the holy grail. We thought these technologies will kill the keyboard. Even if perfect handwriting recognition came out today anyone doing input beyond writing a sentence every few minutes will prefer a keyboard. Even if perfect voice recognition came out today it would take a cultural revolution to see people talking to devices as their only input method with it. Of course there will niches but none of these technologies will be the basis of a "game changing" product. The best input method outside of a keyboard would be neural recognition wherever that happens.

Back to the present if the tablet form factor were sufficient for my needs and ergonomic situation I'd still say if Microsoft sold just the tablet half of the Surface book it would be IMHO a far better device. Win 10 is just a more capable than iOS. iOS is nice but you have to admit it's old now and new functionality just gets tacked on. It's often buggy these days and when I take my iPhone to Apple Store for running slow or eating battery they run some test and say processes are crashing. They tell me clean install iOS meaning I cannot restore from any backup. It is no different from the days you had to reinstall Windows 98 every few months. I can assure you Win 10 is actually far more stable than what iOS is becoming.

iOS needs a fundamental rethink from basic ergonomic principles. The back arrow is on top left of a screen is in a place that is impossible for many to reach while holding the device with one hand. The home button double tap is just a bad way to access the app switcher. 3D touch as it stands is a great concept that is so half baked in it's implementation that it seems like a placeholder. The feature is revolutionary but they couldn't be bothered to even outfit the apps that come with the phone, that cannot be deleted, with functions for this feature. I still love my iPhone and my Macs but I can't help but point out this company is not operating at the level it used to.

 

[trivik12]

Tim cook has said Apple has no plans for merging mac and ios. So I am thinking they will keep best of both worlds. x86 based mac os which sells in boat loads both in corporate and consumers. Plus Ipad for consumers and education and tiny corporate segment. Apple will negotiate big with Intel on chip prices plus has huge margin on these computers.

Rumor on Apple getting custom(Zen based) chip from AMD could make things interesting.

 

[asendra]

Nothing really specific, but here's an interview with Johny Srouji, Apple's SVP of Hardware and in charge of Apple SOCs.

http://www.bloomberg.com/features/2016-johny-srouji-apple-chief-chipmaker/

I find it interesting that the plan was to release the iPad Pro with the A8X, probably around March, but because of it being delayed several months and thus coinciding with the release of the 6s with the A9, they had to shorten the A9X dev cycle by 6 full months.

Makes you wonder what things they had planned for it but didn't make the cut due to limited time, if any.

 

[Arachnotronic]

Nothing really specific, but here's an interview with Johny Srouji, Apple's SVP of Hardware and in charge of Apple SOCs.

http://www.bloomberg.com/features/2016-johny-srouji-apple-chief-chipmaker/

I find it interesting that the plan was to release the iPad Pro with the A8X, probably around March, but because of it being delayed several months and thus coinciding with the release of the 6s with the A9, they had to shorten the A9X dev cycle by 6 full months.

Makes you wonder what things they had planned for it but didn't make the cut due to limited time, if any.

Nice how instead of delaying the A9X/iPad Pro further the chip team just delivered.

A certain large PC processor maker could learn a thing or two from these guys.