Apple walks Ars through the iPad Pro’s A12X system on a chip

E.g. this from 2015 (!): https://www.anandtech.com/show/9396/samsung-sm951-nvme-256gb... This from this year: https://www.anandtech.com/show/13438/the-corsair-force-mp510...

In the Jobs' days they'd promote the practical value of a product. For example they might say how the new iPod can have 50k songs, not how big the storage is, let alone the storage type. They'd mention a smaller form factor or improved battery life, not about HDD to SSD.

So when they talk about SoCs, cores, GPUs, intel or anything else hidden, are they signaling to customers? Maybe its signaling to investors that apple is innovating and that ought translate to profits; maybe the inner geek in all of us?

The dirty trick with hidden tech's performance figures is that they don't directly translate to customer value. As you mentioned, you're dissuaded from buying because the new stuff will be so much better. Maybe it will but the old apple would tell you that the new macs can process video in final cut 10x faster, or you don't have to buy a separate gaming rig, i.e. you can do more stuff better.

The good part about focusing on what the products can do is that you can't fake it. You can't fake it to [geeks] like me, [non-geeks like] my mom, my kids, investors, etc.

Sample size of one and all that - but for me it was a selling point. The PowerPC G5 (PPC 970) was intriguing when it first came out. Having been with Intel, Cyrix and AMD systems since the 286 and used them since XT (Intel 8088) days, it was nice to muck around with something new and paired with OSX Tiger it was such a fun world to explore. The move to Intel felt like a bit of a letdown, but by that time I loved OSX and Snow Leopard cemented it as my OS of choice.

Apple desktops/laptops moving to custom silicon would excite the nerd in me. I want competition.

1. It helped them jump from a failing CPU platform to a non-failing CPU platform. PPC was not keeping up with x86 anymore and Apple's two PPC vendors were going in opposite directions because there wasn't enough of a market for CPU's for Macs. (There were even rumors of future Power Macs migrating to a full POWER CPU rather than PPC.)

2. It meant you could run Windows, and hence Windows apps, on your Mac if you wanted to, without CPU emulation. This is still a fairly important use case.

3. It may have also simplified matters even for Mac application development, since you didn't have to switch ISA's in addition to switching operating systems. Making matters worse, PowerPC defaults to big-endian and x86 is little-endian.

How does this apply to a potential ARM switch?

1. You can't really say x86 is "failing" if it's still the industry standard, but Apple might believe (rightly so, given the market size of iOS) that they finally have the ability to sustainably outperform the performance of x86 on their A-series chips.

Most of the PowerPC bet was that a newer and more elegant architecture would outperform x86 and provide Apple a competitive advantage, and while that may have occasionally been true sometimes, it was never a huge deciding factor. Intel and ARM kept up because they were able to make investments in keeping x86 afloat. Ironically, Intel themselves also bet that a newer, more elegant architecture would make x86 obsolete, namely Itanium, only for AMD to invent x86-64. Not even Intel themselves could stop the x86 train.

With the rise of mobile devices, ARM now has the same market power as x86, if not more, simply because there are many more ARM-based devices manufactured and sold than PC's. Apple in particular has been able to invest heavily in their A-series chips and has full control of their CPU roadmap and destiny. Perhaps this time, x86 may finally be rendered obsolete. Don't count on it, though.

2. This is really mostly dependent on Apple's strategic priorities. With more and more application functionality moving to mobile and the web, being able to run Windows is less and less important. At the same time, being able to run Linux is more important; for many developers, running a Linux VM in Vagrant or Docker lets us develop in a similar environment to the servers our code will eventually run on. Sure, you can run Linux itself on ARM, and perhaps there will be more Linux distros that support ARM when and if Apple switches the Mac, but it won't actually be the same as the server unless ARM makes serious inroads in the server market.

Maybe they're betting they can surpass x86 enough that they could emulate x86 at respectable speeds. Since they would be migrating CPUs again, they will probably provide a CPU emulation layer again, like they did when migrating from 68k to PowerPC and then from PowerPC to x86. Keeping this emulation layer around would have more of a benefit because, after awhile, nobody needed to run 68k or PowerPC code anymore. This has never been true for x86 code, and it won't be for a long time, so look for Macs to continue to run x86 even if Apple switches.

3. I think A-series is also little-endian by default, and for x86, see above. Maybe Apple is banking on getting more value by running cross-platform iOS/Mac apps than cross-platform Windows/Mac apps. This will probably impact Mac gaming the most, but that's never been a priority for Apple.

while I have seen many laud the iPad Pro for its power I haven't see any mention of heat or how long it can sustain a workload. Personally I do not want to see the Mac line change processors again for two reasons, like I mentioned prior many of us have a lot invested in software that runs on OSX as well as Windows that these machines can run, second I don't need that wall to go any higher.

I hope I’m wrong though.

Would moving off of traditional desktop CPUs harm that? Is there a way to do compatibility at the OS level without sacrificing half of the performance gains?

I’m thinking you’re going to have a hybrid architecture for MBP. The T2 will expand to support all of Apple’s own software plus all upgraded one sold through a revamped Mac Appstore (plus hopefully your own compiled stuff when security settings are off). This can power down (all but ~2 cores) and instead power up an x86 coprocessor that supports everything else. If Intel doesn’t deliver that, AMD will. For Macbook I think you’re on the right track, just that the Air will be replaced with a 13 inch Macbook model with the same architecture.

This eloquently summarizes much of the disconnect customers have been feeling about the Mac product lineup. Jobs never would have allowed this to happen.

From an Apple perspective you could easily justify the price for all the old Intel Chip. 1. Intel has a 2 years+ node advantage which you cant get anywhere even if you pay. 2. Intel has the best performance / watt CPU on the market, also highest performance Core on the market, you cant get it anywhere even if you pay. 3. x86 compatibility, which is more like a x86 tax. Although you could get it from AMD.

Now the first two is gone. TSMC has now edged Intel in node, Apple themselves has the best Pref / Watt work in A12X. AMD has proved to be very competitive at the high end. And Intel is still charging Apple the same when they have far less value.

Apple is now being hold up by Intel, but I don't think Apple could dump Intel just yet. There are two thing that is holding Apple up.

Thunderbolt - TB is currently still an Intel only technology. There are no other host controller on the market other than Intel's one. And they cost a fortune ( relatively speaking ) Apple has invested a lot into TB, but Intel is making all the same Firewire mistakes. May be Apple is working on a USB 4.0 solution, and they would dump TB once and for all in 2020. Intel promised to make TB as an open standard in 2018 but has yet to do so.

Modem - Apple relied on Intel Modem for iPhone, which is Apple's bread and butter. Before any move on the Mac side Apple will need to think about its consequences on Modem. In worst case Apple could switch away from x86 and Intel decide to hike the price of Intel Modem. I think the revenue of Modem is roughly the same as revenue of x86 from Mac. Intel 10nm isn't performing. And Apple is not happy with Intel on both front.

That sounds like the most un-apple thing I can possibly imagine. Their ethos was always to build a device that does something incredibly well - a phone that can be a phone but also a desktop computer is everything but this.

If Apple invests in enhancing CarPlay, it might be a sign that they are scaling to a wider convergence market. If they don't invest or abandon it, then maybe convergence won't happen. They're famous for saying they're not working on projects that they are actually working on, so we have to read between the lines.

After upgrading from an iPhone 6 to an iPhone XR, I've been thinking how close the latest gen devices are to traditional computers anyway. The configuration/settings and features are so far beyond the first gen devices that I think convergence will happen, nobody knows what it will look like yet.

Because that would mean Apple doesn’t get to sell you two devices.

Sure - one device could do everything, but that doesn't drive sales or profit.

But attach a proper(!) keyboard with touchpad to an iPad Pro, put macOS on it instead of iOS, and there's your next MBP ;)

What console-quality games actually ship on the iPad?

The problem with games is that to actually match an Xbox One S's graphics you don't just need to match its 5 year old hardware in performance. You also need to have the capacity to actually fit the game & its textures. All 40-80GB of it.

Who is going to ship an actual console-quality game at console-quality sizes on a device whose base model can't even hold it?

And Apple is notoriously stingy with RAM. This new one bumps it up to 6GB, which is nice, but still less than the 8GB in an Xbox One S. How much does iOS reserve of that, and how much do the games actually get?

Wheres the example of software that truly shines on these chips? Wheres the software like After Effects, Houdini, Octane Render that you can truly see the power of the machine rev up when on the right hardware.

They're wheeling Photoshop out as proof of this devices power yet as a designer I certainly don't consider Photoshop a heavy piece of software anymore and the only reason it ever chugs is it doesn't use the machines power effectively, mostly single core and disk speed constrained.

This power has been available to iPad developers for a few years now so shouldn't we be seeing truly powerful pro apps that take advantage of it emerging? Are these chips actually powerful for real world pro tasks or are they just talented at providing Geekbench scores.

That's the feeling I got, as well as perhaps being a human abstraction layer between the engineers and the executives. If you can describe in relatively understandable terms something that is technically difficult to thousands of laypeople (well, that's unfair - Anandtech was for nerds but being a nerd doesn't make you an IC/EE engineer), you'd be an asset to both the corporate and engineering teams.

And that translation works both ways - understanding the direction of the company with regards to future products vs what you want to get out of your silicon teams (e.g. when Anand mentions thermal envelopes, that might include an understanding of the limitations stemming from the potential form/design and material of a future product).

With Apple's focus on on-device ML, I would guess this will be the first part of the A-series trifecta (CPU, GPU, Neural) to be included on a Mac and exposed to developers. I can imagine a bunch of possibilities for such a chip, not just FaceID.

You're left with an ARM chip

FaceID is not solely dependent on ML, it's also managed by the secure enclave co-processor which is also used for Touch ID which is available on Macs now. ML helps to reduce false positives.

Apple's T2 chip is an ARM-based processor that's already in almost all of the newest Macs, it is used as a storage controller (which allows Apple to encrypt the drive very fast and transparently), security enclave processing (touch ID on MBA), Siri processing, and more. Every year, more and more of the processing is moved to Apple's T series co-processor.

Apple's custom silicon allows them to integrate software and hardware on a deeper level. Intel develops CPU for the mass market. Apple develops for their own customers only.

To think that deep in the Apple labs that don't already have A-X laptops running - and have for, for a while is not thinking like Apple would.

Despite their ongoing efforts to make the iPad more capable, I think and hope they'll recognize the value in keeping it simple enough for anyone to use, and thus having a separate macOS experience with more tools/flexibility in a laptop/desktop form factor.

Anyways, Apple has never been one for smooth transitions. Their history is dotted with big, bold changes. If they kept x86 they would slow the adoption of their new architecture. Apple will likely take a "take it or leave it" attitude like they did with the CD drive and headphone jack.

Depends on what you mean by "much". They have made good backwards compatibility an important part of every single architecture transition so far, and on the Mac there were good 3-4 year official transitions at least (the 68k emu still ran under Blue Box/Classic Environment so it lasted through 10.4 Tiger, Rosetta lasted through 10.6 Snow Leopard). And that's official, in practice there have continued to be longer last options.

It's certainly not the degree that Microsoft has traditionally cared, but it's not at all been blown off either.

But from a software usability standpoint, it's not.