thenewperson
macrumors 65816
That actually makes a lot if sense lol. Their dev account is about to be terminated anyway...
Apple can sue them though right?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple DTK teardown
- Thread starter tripleintegral
- Start date
- Sort by reaction score
That actually makes a lot if sense lol. Their dev account is about to be terminated anyway...
Apple can sue them though right?
thenewperson
macrumors 65816
Oh I thought the account that was stating was just the one that had Unreal Engine stuff associated with it.
MacPoulet
macrumors 6502a
Apple's already suing them... This would just add more fuel to the fire, no?
Epic is suing Apple. Epic applied for an injunction against Apple, and got a split decision. Fortnite stays off the App Store, but Apple must reinstate Epic’s developer account, as there are people who use the Unreal engine for their games, and would not be able to update their games if Epic’s developer account were to be revoked. Judge said it would harm people not part of the lawsuit, and that Apple’s actions were punitive.
Boil
macrumors 68040
Yes...
Sorta...
The presiding judge gave Epic the choice to go back to the terms of the original contract while awaiting their trial date, and Apple said they would allow Fortnite back on the App Store if Epic did so. Epic decided to continue to whine about anti-competitive B$ & chose NOT to return to the way things were while awaiting their court date...
Apple is still going to kill the Epic Games developer account this Friday, the judge's ruling regarding developer accounts has Apple barred from killing off the Epic International developers account, which is the overall account for Unreal Engine...
Bug-Creator
macrumors 68000
Few thoughts:
- At WWDC they had ArmMinis running the 6k display
- Apple has had to have some Rev0 ArmMacs for at-least a few months internally and at selected external developers (Adobe) to get the SW to the state it was at WWDC
- Apple has 0 interest in sending out DTKs with SoCs they haven't announced yet
Speculations:
a) they used some lower refresh mode to get that DTK to 6k and if you'd opened it you'd find something based on the iPadPro + hotglue and patchwires to make it work
b) they had made plenty Rev1 DTKs back in late 2019 which they used internally and shipped out after WWDC, while they used a A13/A14 Rev2 DTK for the presentation
c) they had an A13/14 DTK all along but decided to downgrade if for the "public" DTK to curb expectations
Both b) and c) would make a good argument for an DTK-based MacMini upgraded to A14 to be released early on in the transition period
- At WWDC they had ArmMinis running the 6k display
- Apple has had to have some Rev0 ArmMacs for at-least a few months internally and at selected external developers (Adobe) to get the SW to the state it was at WWDC
- Apple has 0 interest in sending out DTKs with SoCs they haven't announced yet
Speculations:
a) they used some lower refresh mode to get that DTK to 6k and if you'd opened it you'd find something based on the iPadPro + hotglue and patchwires to make it work
b) they had made plenty Rev1 DTKs back in late 2019 which they used internally and shipped out after WWDC, while they used a A13/A14 Rev2 DTK for the presentation
c) they had an A13/14 DTK all along but decided to downgrade if for the "public" DTK to curb expectations
Both b) and c) would make a good argument for an DTK-based MacMini upgraded to A14 to be released early on in the transition period
Last edited:
Boil
macrumors 68040
This is doubtful, Apple would have any third-party developers sign a pile of NDAs & they would still be working in Apple labs at the Mothership...
Loose lips & all that...!
dmccloud
macrumors 68040
Given that Microsoft already has Windows 10 for ARM, it is a safe assumption that both Microsoft and Adobe already had the bulk of the work done regarding transitioning from x86 to ARM, so the only real changes they would need to make would be MacOS - specific. Microsoft uses the same codebase for Office on Mac and Windows, and I believe Adobe does for at least some of its apps. They likely had Microsoft and Adobe reps working onsite (at an Apple facility) to complete the transition processes, as they couldn't risk leaks of the DTK prior to the WWDC announcement.
On the other hand, 'everyone' knew that Arm was happening months before it was officially announced!
Jorbanead
macrumors 65816
They did exactly that. Craig said in an interview during WWDC that they brought in a very small team from Adobe and it didn’t take them very long at all to create an Apple silicon version of photoshop.
Kaida
macrumors 6502
Everytime i see an update on this post, i was hoping Epic teardown their DTK for us 😘
dmccloud
macrumors 68040
There is a difference between knowing Apple planned to switch to their own silicon and knowing details of either the DTK or initial AS hardware that will hit the market...
Mainyehc
macrumors 6502a
If you mean a full-blown disassembly, the same way we may get teardowns of the AS one, i.e. if any unit is reported stolen or destroyed and isn't, or if Apple doesn't see through to its recall, like what happened with this dev:
Not that I think that that would happen anyway; judging by all of Apple's recent litigation, and by just how quiet things are on that front, the NDA covering the AS DTK must be really solid and aggressive, so they'll likely be much more rigorous about either/both that recall process or/and doing serial number matching if any high-res photos of it pop up on the internet (if you only have a handful of specimens to go by, how will you know for sure which parts of the logic board and other components are safe to leave unblurred, and which might have some form of device-specific ID encoding hidden in plain sight? Sure, anyone with half a brain would blur out any QR-like codes and obvious SNs, but would any of its internal markings, even seemingly innocent part and batch numbers, be guaranteed to be just that? On the closest thing you can get to an actual prototype from Apple? And in a potentially bad and “alpha-like” sense of the word, at that, as it may be embarrassingly kludgy? Ehhhhh).
Yep. It used the legendary latch from that case, which lasted all the way to 2013 in the proper Mac Pro. It was literally a slightly modified PowerMac G5 case, with shared components down to the power supply, as you can see from the article where I found that tweet: https://www.macstories.net/stories/this-is-not-a-product-the-apple-developer-transition-kit/ (You likely saw these before, but I'm liking to that mostly for the benefit of all the young'uns and recent switchers).
Yep. It was basically an eviscerated PowerMac G5, with the rear two thirds of that steel mid-plate assembly, which held the massive DP/Quad heatsink and fan assembly and separated the thermal zones, completely cut out. If you told me that was a Martin Molin/Quinn Nelson-esque angle-grinder hack job I wouldn't doubt you for a second… 😛 But hey, at least they kept that inner Plexiglas cover thingy for airflow enhancement completely unchanged (why they left it there in the first place, considering it now had a hole right behind it, is an entirely different question altogether 🙄).
And yep, I see where you're coming from with that assumption that the AS Mac Mini will look nothing like the Intel versions. Indeed, apart from the aforementioned latch and overall layout of the front panel IO and optical drive bay(s), the Mac Pro was nothing like the G5 on the inside. Besides the slightly raised upper optical drive bay and the additional of the tell-tale, secondary one, gone were the (… seven?) discrete thermal zones and said Plexiglas cover, the IO port layout on the front and on the back was very different, the power supply moved to the top and forced a switch from that pair of weird latching hard-drive bays to the much nicer four sliding tray bays, the PCI expansion slots were push downwards, etc. Even though these two were as similar from the outside as, say, the entire G3 B&W and G4 B&G/Quicksilver/MDD family, the latter were much more consistent on the inside.
However, there's the flip side of the coin: back then, Apple was switching from their proprietary PowerPC architecture, and generic Intel-compatible boards were universal and incredibly easy to source, as were the oversized G5 towers rated for the beastly Quad, which made the entire prospect of developing dedicated prototypes just for that purpose incredibly wasteful and misguided. Adding to that, the contingent of Mac OS X developers, the scheduling – or, better yet, the lack of a fixed one – of its releases and Apple's concerns with the environment weren't even comparable to what they are today, so the apparently counterintuitive combination of Apple's relative lack of logistical prowess at the time (yes, the iPod was indeed at its peak and sold like hotcakes, but the Mac distribution chain was still abysmally unstreamlined on a global scale and the iPhone was still just a well-guarded project doing the rounds of the legally-blind rumour mill) and going for that beast of a case wasn't that big of an issue in real life, anyway.
Of course the Intel DTW would look nothing like any other past or future product and just take the box where it would fit more snugly; however, there's an important factor to consider: other Apple machines with similar TDPs different processor architectures retained most of its internal design and overall shape and/or size when doing the switch… Compare and contrast that weird PowerMac G5–DTK–Mac Pro evolutionary pathway (while the Xeon towers weren't as barren as the DTK, they were indeed more spacious and expandable than the Quad G5s) to, say, that from a late-generation PowerBook G4 to a first-generation MacBook Pro, that from a G4 Mac mini to an Intel Mac Mini, and especially that from a Rev. C iMac G5 to a first-generation Intel iMac (the resemblance between those is so uncanny, both on the outside and on the inside, that it's easier to just boot them up and open the About this Mac window instead of flipping them around to check whether their display output port is Mini-VGA or Mini-DVI, and if you were to tear one down without figuring that out first I wouldn't blame you if you forgot which heatsink angle corresponds to which processor and got them mixed up). Apple took the power savings and upped the performance, as their designs were already above average and over-specced for those newfangled Intel Core chips anyway.
Now, the tables have turned when it comes to the suppliers, but from Apple's standpoint the situation is analogous (if more dramatic). So, seeing how a. Apple has a tendency to keep the external enclosures and even internal layouts of many products seemingly unchanged between ISA transitions (besides those examples I pointed out, they did the exact same thing with some Performas and Quadras, IIRC) and b. one of the main reasons why Apple got fed up with Intel and was so bashed by a ton of reviewers over thermal throttling was the fact that they were counting on more efficient chips from their laggard partner and ended up having to release products which depended on those by design, turned out to be a hot mess instead, but would greatly benefit from their own, highly efficient silicon… does anyone here seriously expect a wave of radical case or even internal redesigns any time soon? In most of the Mac lineup, the transition will mostly come down, at least for the first few iterations, to a massive boost in either raw performance or performance/watt ratio depending on the product line (possibly including the resurrection of the venerable-turned-dud 12'' MacBook Retina), and the belated addition of advanced features (FaceID and HD FaceTime cameras, anyone?) and other dedicated IC module+software optimisation wizardry we all came to expect from Apple by being exposed to them in our daily usage of iPhones and iPads, and by reading about – or indeed using, but those who have that privilege are a tiny minority – products like that ultra-high-end, FPGA-based Afterburner doohickey.
Where does that leave the Mac Mini? Since its inception, it only had two major internal+external design combos, spanning two ISAs, which were overall very coherent and only underwent very incremental, minor evolutions (like the shape of the case, yes) and additions (such as the IR sensor and overall IO speed and versatility), and sometimes even regressions, some of them positive (like the return of slotted memory), some of them somewhat neutral (like the removal of said IR sensor and the optical drive slot, at a time when other more advanced means of wireless communication and data delivery were ubiquitous), or even outright negative ones, like the soldered-on memory they eventually backtracked on and the infamous processor “update” that made slower than older Quad Core models 🙄). It's kind of a weird, step-child-like machine, probably from its inception and likely will stay that way forever, as Jobs and Ive adhered to that Raskin-esque philosophy of treating computers as appliances and Cook isn't far behind them with his professed love for all their “post-computer” devices, so I'm guessing that the BYODKM concept, at the very edge of what is tolerable by Apple's standards of full control of the entire widget, was always seen as inherently “ugly”, visually and in concept, by upper management. It's a great machine for cluster servers, tentative switchers, and antiquated but still selective and devout consumers or just regular prosumers on a budget. It serves a niche of a niche of a niche, which Apple management doesn't particularly enjoy throwing money at, but still deems necessary to address for the sake of completeness and flexibility of their ecosystem. That's all.
Maybe they could have a change of heart and “develop” a NUC-sized machine running macOS… But will they, really? They wouldn't have to develop much of anything at all, as that would basically amount to “hacking” (i.e. flipping some flags in software here and there) macOS into an Apple TV and maybe stick a few more or just beefier ports on the back… And for what benefit? If anything, it would muddy up the waters and mess up their product line matrix even more than they already did – albeit acceptably so and for a good cause, as their use cases can indeed be very different, and there's a lot of money to make in both branches – with the overlap of the iPad Air/Pro+Magic Keyboard combo and the entire MacBook Pro line… and further cannibalise the market for a souped-up but still affordable Apple TV that could behave more as a console if it was marketed and bundled as such. Nah, Apple is likely keeping the Mac Mini as is, design-wise (the entire thing), at least for the next few years, as there's an entire industry standardised on stacking the damned things on racks and suddenly providing it, along with all the other niche users I mentioned, with what is effectively a beast of a modular/headless desktop computer just makes business sense. It's going to have a similar TDP envelope and get a really fast processor, all either at the same price or never lower than the lowest $499 tag it ever reached waaay back in 2014, mark my words.
And yes, I stand by that, even – nay, especially – while assuming the AS DTK is just effectively a souped-up iPad Pro board and a custom multi-port hub stuck into a Mac Mini case. I'd venture a guess and say the only reason they didn't go with a modified Apple TV instead is precisely because much like in that presumptive NUC-like redesign, the ports wouldn't easily fit and it would become either extra kludgy or limited, and its heat dissipation capabilities, currently targeted for an A10X processor, might indeed be a wee bit insufficient and require too much of a redesign for such a limited run. That, and not really having to stick it to Intel, right in their faces, that their processors are such crap that even an Apple set-top box with a tablet chip can eat both a low-power NUC and a beefier i3-based machine for lunch, because what drives Apple's business decisions are usually more mundane and utilitarian calculations, in this case environmental impact, economies of scale and speed in distribution vs. product development expenses and convenience for developers. And erroneously lead Mac Mini customers (yes, the whole lot, whatever their profile) to think that it will become any smaller (or at least Apple TV-sized) in the near future…
That would be a lose-lose proposition, and using the Mac Pro chassis for the DTK instead would be ridiculously overkill, way too soon after that product's unveiling and introduce a similar degree of confusion and unease – as Apple Silicon and, to some extent despite the unveiling of the Fugaku supercomputer, ARM itself have yet to prove their worth in general-purpose, high-end workstation applications – at the opposite and equally valuable end of the spectrum of their market, so… the middling and boring Mac Mini box it was, then, with the added and inverse benefit of really surprising everyone with the ensuing boost in performance.
But hey, I would still LOVE to see the innards of that DTK thing regardless – or, in this case, especially because – of how far removed it is from their actual product development process. I would always very much want that, out of sheer curiosity, but the fact that it might be kludgy and that Apple really doesn't want us to makes it outright morbid and irresistible. xD
Last edited:
dmccloud
macrumors 68040
While I would like to see a DTK teardown strictly out of curiosity, I don't see how it would offer any insight into what the actual Apple Silicon Macs will be running. I also think that the DTK is most likely running a modified version of an iPad Pro logic board, but what I didn't consider was the possibility of those ports on the back actually being some sort of custom USB-C hub rather than part of the actual logic board. I would wager that when the program ends and people start sending their DTKs back to Apple, we will see someone do a teardown. Unfortunately, by the time that happens, we will likely already have at least one (if not more) Apple Silicon based Macs on the market that have been taken arart by iFixit, JRE, etc.
crystalidea
macrumors regular
Mainyehc
macrumors 6502a
That's what makes this thing interesting… Is that known for a fact, or are people just too afraid of even popping up the bottom plastic cover with a spudger? That design itself, even as a proper Mini, evolved from the very friendly twist-to-lock/unlock mechanism to a more permanent and solid snap-on/off one, and I don't see Apple changing the latter that much. Did they use a different, yet more stiff and permanent/inviolable style of clips (i.e. those classic teeth-like ones, only wedged so deeply into cut-outs on the aluminium case or otherwise designed in such a way that you can't pop the cover off without either breaking those altogether or completely messing its looks, or even both), or some weld points instead of just screws under it?
I don't know, but all that seems a tad overkill. Writing an ominous NDA seems much easier than going through all of that trouble. The Mini is already enough of a pain to open as is, with those idiotic TR6 Torx Security screws and whatnot (yeah, I even had to take a one hour drive to my go-to shop's main warehouse, all the way up in Effsville, Nowhere and right after Christmas, because that stupid rarity of a screwdriver was out of stock on their main location in the city, some friends of mine had a plane to catch in a few hours and I had to perform an SSD upgrade on their Mini, STAT; sure, I should've checked the iFixit guide's parts and tools list more attentively, but I already had the equally specific logic board removal bent-wire thingamajig on hand and the last time I had opened one of those it still came with regular T6 screws, so I just assumed they wouldn't be so stupid as to make it even less user repairable/upgradeable, for no good reason, than it already was, especially considering how uninviting the new and outright concealing pop-off cover made it look), and it would not be too difficult for them to swap some of those with tamper-evident screws like the ones they used on the 2016 12'' Retina MacBook ( https://pt.ifixit.com/Teardown/Retina+MacBook+2016+Teardown/62149 ), really.
There's also another benefit, besides cost, to that approach that we should consider: using a case (and bottom lid, and everything) as close to that of a current Mac Mini and just bolting some internal mounting frames/brackets for a hub and a modified iPad logic board onto it also prevents something as discernible and obvious as a Mac-Mini-sized logic board board with a freaking A-series chip (or bare BGA pattern, whatever) on it, or even something as subtle as a just a Mac Mini case with different mounting points, from getting leaked and plastered all over the news. For all we know, the only undeniably bespoke and proper-looking part on that DTK, which might indeed give its existence away in advance, is the plastic back I/O faceplate (and, even then, it could still be mistaken for an ultra-low-end Mini prototype part or something, as they even went to the extra trouble of keeping all I/O ports to the sides of the vent in the exact same position as in the production model, centring the remaining, reduced set of ports above it and even leaving the gimped USB-C ports completely unmarked, thus obfuscating the fact that the DTK has no Thunderbolt connectivity and further cementing the plausibility of it just being a part from an earlier EVT- or DVT-stage device with tentative I/O variants).
It may look like I'm overthinking, but that's quite the contrary. If anything, I'm underthinking, but likely on the money here. I've been following this company for almost 17 years now, and while I'm not an engineer or product designer, I did indeed study graphic/communication design, communication theory, semiotics, marketing, etc. If there's anything Apple is really good at is at manipulating public perception of everything they do. People talked about Steve's legendary “RDF”, and, sure, he was a great salesman, but there's always been much more to that behind the scenes. Something as simple as an unmarked port carries a multitude of meanings, and in hindsight, these assumptions are extremely easy to make. It's only when you start taking all of those together and connecting the dots, extrapolating from there and making new ones for the future that it gets really fun.
My theory is that with Apple's current manufacturing prowess and the relatively humongous run of this thing (no, really; it's nothing compared with an iPhone or iPad pre-launch ramp up, or maybe even with a regular Mini production batch run, but bigger than those of many niche third-party accessories), all internal AS DTK parts are probably beautiful in their own right (maybe the boards aren't painted in prototype-red/purple or some other weird-ass colour, as so many of those popping out of some factory in China might raise a bit of suspicion and attract too much attention, but in Apple's signature premium-black, and any supporting pieces are likely sturdy and made from some nicely stamped and deburred metal, so as to prevent injuries in the assembly line and, well, look good even if no one ever sees them, because this is still a non-product-“product”, designed by the “the-company-itself-is-Apple's-biggest-product”-company's team for “developers-are-our-most-important-customers”-level special customers, we're talking about; come on, even the box it comes in and its pack-in accessories and information/greeting booklet look as much like those from a retail Mini as they could), but the the internal layout of the thing is definitely kluuuudgy af and very rich in empty space. It's the baby, premium version of the Intel DTK, to a tee.
TL;DR: Same formula (pre-existing parts hacked together), and for the same reasons (keeping it simple and out of the rumour mill) —> similar result, in the inside as on the outside (a glorious jury-rig of a thing that appears perfectly normal at first glance but is anything but once booted up or peeked into), with specs, both architecture- and performance-wise, equally mismatched from those of final product(s) (a P4 with a regular BIOS on a generic board was almost as much of an evolutionary dead-end and a Frankenstein that wouldn't be fit for purpose in most of the Mac product lineup as a year-old A13Z with some RAM slapped on top and running a marginally more open and flexible OS is now).
That thing can't even run virtualisation software at all (on that regard, no DTK or any other Mac-shaped boxes were visible in-frame anywhere near the Pro Display XDRs used in the WWDC'20 demos, so I am betting that Apple VP Andreas Wendker was actually running Linux on Parallels from an even kludgier prototype A/M14 MacBook board in a shoebox hidden somewhere under that desk, and it's such an esoteric and specific topic that it took me a while to even find this link and confirm that quirk: https://hacker-news.news/post/23922846 ) and is missing a ton of pretty basic and creature-comfort features (the link to this big exposé on Twitter, incidentally the best compilation of the DTK's limitations I've seen so far, was also buried deep in that thread: ), which further points to how much of an incomplete/rushed hack job it really must be on the inside… All those quirks do fit the “iPad logic board + USB-C multi-port hub” theory very neatly, now, don't they? 😉
Last edited:
They more than likely just epoxied the bottom plastic cover in place. Not complicated, and easy to see should somebody attempt to remove the bottom cover.
The DTK is exactly what the term "DTK" means. It is just a chunk of hardware designed to allow SOFTWARE developers to see how/if thier software runs on Apple Silicon. Whether it was a kudged up iPad Pro motherboard, or was a specofically designed DTK motherboard using an A12Z SoC really must await somebody getting one open. The key part for Apple is that it got an AS powered unit into the hands of software developers to let them do what they need to in order to get ready for the transition. I would tend to think that this is a board that was designed for this unit, in order to reude the Mac Mini power supply and its connectors. I also think this was used internally at Apple for a while (in the past) to allow for internal software development (for MacOS, FCPX, Logic, and the other built in apps that ship with MacOS) needs. and more that likely, some were likely used by the "big" software vendors (think Adobe, Microsoft sized companies) for thier own software development even prior to its being made avaialble to the smaller developers. I also think that MacOS development for AS SoCs has been going on for years, it was just never made public. It takes time to do OS development, and then thoroughly test it; certainly far longer than the few months between the announcement of MacOS Big Sur and the availability of AS Macs to the general public.
Of course, the Apple internal development team has been running on genuine AS Mac SoCs for a while now, as the DTK can only take you so far. The iMacOS developers write to the hardware, so need the actual hardware to write to and then do verification against. It may have been a 7nm or even 10nm version of the AS SoC, but it would have been feature complete, not an iPad SoC, so it would have had 8P cores (if Apple goes that way), working TB4 and USB4 ports, working HDMI and DP ports, and whatever other modules that Apple will be putting into the AS SoCs for the Mac. So they have moved away from the DTK Mini as a development platform for a while.
The DTK is exactly what the term "DTK" means. It is just a chunk of hardware designed to allow SOFTWARE developers to see how/if thier software runs on Apple Silicon. Whether it was a kudged up iPad Pro motherboard, or was a specofically designed DTK motherboard using an A12Z SoC really must await somebody getting one open. The key part for Apple is that it got an AS powered unit into the hands of software developers to let them do what they need to in order to get ready for the transition. I would tend to think that this is a board that was designed for this unit, in order to reude the Mac Mini power supply and its connectors. I also think this was used internally at Apple for a while (in the past) to allow for internal software development (for MacOS, FCPX, Logic, and the other built in apps that ship with MacOS) needs. and more that likely, some were likely used by the "big" software vendors (think Adobe, Microsoft sized companies) for thier own software development even prior to its being made avaialble to the smaller developers. I also think that MacOS development for AS SoCs has been going on for years, it was just never made public. It takes time to do OS development, and then thoroughly test it; certainly far longer than the few months between the announcement of MacOS Big Sur and the availability of AS Macs to the general public.
Of course, the Apple internal development team has been running on genuine AS Mac SoCs for a while now, as the DTK can only take you so far. The iMacOS developers write to the hardware, so need the actual hardware to write to and then do verification against. It may have been a 7nm or even 10nm version of the AS SoC, but it would have been feature complete, not an iPad SoC, so it would have had 8P cores (if Apple goes that way), working TB4 and USB4 ports, working HDMI and DP ports, and whatever other modules that Apple will be putting into the AS SoCs for the Mac. So they have moved away from the DTK Mini as a development platform for a while.
Mainyehc
macrumors 6502a
That much is obvious… What's interesting is to try and guess when and how that might happen. I'm leaning towards that “burnt in a fire” scenario and the pictures of the board being heavily censored, for those reasons I've listed.
Well, that part seems unnecessary, I think. Who's to say that internally its output isn't just being sent through said “hub” (much like you would do with a regular, commercially available hub via a USB-C PD passthrough port) and directly to a fan? Bear in mind that according to that developer whose Twitter thread I linked to, the fan runs at full speed, which suggests the machine isn't doing any thermal management.
That makes perfect sense; the only exception being Parallels and maybe VMware, those two likely had access to something more advanced and closer to the final hardware (or at least with a chip more advanced than the A13Z).
If Apple's earlier transition from PowerPC to Intel is anything to go by (according to Steve Jobs, Mac OS X had been leading a “secret double life” since the Rhapsody days, and there was no reason why it shouldn't, as NeXTSTEP did run natively on x86 before Apple bought NeXT), and you consider Jobs demoed an early build of then iPhoneOS in January 2007 on actual preproduction iPhone hardware, there must have been builds of Mac OS X / OS X / macOS running on ARM processors since as far back as 2006, or maybe even in 2005 (the decision of going with ARM processors for the iPhone was likely made between 2004 and 2005, and Scott Forstall, the guy behind the team that converted Mac OS X into iPhoneOS, knew Jobs wanted Apple to develop a phone at the end of 2004: https://www.theverge.com/2017/6/13/...et-history-iphone-brian-merchant-book-excerpt ).
Much like pre-Rhapsody NeXTSTEP was a great starting point, once they had the core OS ported to ARM and had to keep it updated anyway, there was no good reason for them not to maintain a complete, native build of macOS on ARM on the side, just in case; and as much as that may irk some Apple users, Apple will likely keep x86-64 builds updated internally.
Evidently. The DTK is indeed very limited, in that it doesn't even use the A13Z's Efficiency cores. They're just keeping those M-series-based (or whatever name they come up with for Mac-specific SoCs) machines to themselves for the sake of secrecy, and because most developers indeed don't need those features anyway. Whatever those machines are, they're likely what they used in their WWDC demos. Maybe they look like a 2019 Mac Pro version of the Intel DTK (which would make getting some of those in very high-end developers easier), or maybe they're just those weird, exposed boards you see whenever you get a glimpse at their labs or prototype MacBook boards connected to peripherals (which would entail having them covered or hidden, and bringing in those developers to Apple's HQ instead), but whatever they are, they're just too kludgy and impractical to ship to the majority of developers.
Last edited:
It is possible that they were told to remote in to the actual release level production hardware. It would allow for software development on some heavy duty machines (iMac Pros or Mac Pros), and allow for running on actual production hardware without actually allowing anybody to see the actual hardware.
The DTK is probably a custom PCB made for the purpose. It really doesn't cost that much to do a PCB, and lets face it, Apple can afford it.
The DTK is probably a custom PCB made for the purpose. It really doesn't cost that much to do a PCB, and lets face it, Apple can afford it.
2006 I don't know. We do know an itern was working on the first steps to boot OS X (Darwin) on ARM back in 2010.
Porting Darwin to the MV88F6281 | TU Delft Repository
dmccloud
macrumors 68040
I don't think Apple has been working on ARM-based Macs since 2006, or even 2010. My guess is that 2010 is when they started theorizing on what an ARM-based processor would need in order to run MacOS effectively and efficiently, then waited for the processors to catch up before starting on the actual design and transition. My guess would be that they started working on an Apple Silicon based Mac 4-5 years ago when in-house development began on both the A12 series and Catalina, as dropping 32-bit support made the transition process easier for Apple.
Mainyehc
macrumors 6502a
They haven't been building ARM-based prototype Macs per se since 2006, no. Notice how I said “ARM processors”, not “ARM Macs”. 😉 I can almost bet they kept at least some builds of OS X running on iPhone and iPad boards, not unlike the one inside the DTK, ever since they had those available internally, and even more likely from the development of the A4 onwards.
As for your assessment of their internal timeline for actual AS-based Mac development, it does seem correct for the most part. I would say, though, that they likely toyed around with AS versions of the 12'' Retina MacBook, if not from the very beginning of its design process (i.e. before its launch in 2015), maybe shortly after, when developing the T1-equipped USB-C/Thunderbolt 3 MacBook Pro line. The T1 chip, even before the deprecation of 32-bit support, was the writing on the wall for Intel, IMHO, as was the poor performance of that machine throughout its existence. I seem to recall reading a lot of reviews comparing it to its iPad counterparts and it always coming up short on the performance department, despite both lines being equally constrained when it came to thermals and battery capacity.
It's not like I disagree with you regarding their wait for manufacturing processes advances and performance gains on ARM just because I believe they had those prototypes lying around, though; while an ARM-based 12'' Retina MacBook would've been a great machine even in 2015, getting those chips to high end iMac and Mac Pro levels of performance would be a tall order, and keeping a split architecture (like what Microsoft is attempting) is as un-Apple-like as it gets (I know they kept 68k Macs limping along with their more modern PowerPC counterparts for some time, and sold overlapping and messy combinations of Apple II/III/IIgs/Mac/Lisa product lines at various points before settling on the Mac, but I'm obviously referring to modern, post-Jobs'-return Apple). They picked ecosystem consistency over individual product quality and, sadly for all fans of that one in particular, it was the right choice.
2006 I don't know. We do know an itern was working on the first steps to boot OS X (Darwin) on ARM back in 2010.
Porting Darwin to the MV88F6281 | TU Delft Repository
repository.tudelft.nl
While the article you linked to is certainly very interesting, I'll add this counter-interpretation of it:
There’s no ‘secret project’ to port OS X to ARM, because it already exists
Don’t believe everything you read
www.theverge.com
The chip that intern used was the one found in Apple's second-generation Time Capsule, a product launched in 2009. The iPhone was already three generations in, and the A4 was about to be released… And it's an experimental project by intern, likely for training or assessment purposes as the title implies, we're talking about, or at best a dramatically slimmed-down port for a specific processor in what is effectively a UI-less appliance.
The “running on ARM” part was already done by then. iPhoneOS/iOS/iPadOS/tvOS/watchOS are all, conceptually and technically, slimmed- and locked-down versions of Mac OS X/OS X/macOS, albeit revamped through and through on the UI API level. At their core, all run atop the same Darwin core you mentioned, ergo, Apple effectively had to maintain it for ARM out of sheer necessity, in order to support those touch-based and specialty OSes with constant updates and security patches, and not just to test tentative ARM-based builds of their classic (not to be confused with Classic 😉 ), WIMP-based UNIX OS. Except for classic iPods, all Apple products since the Apple–NeXT merger have been running a flavour or another of NeXTSTEP-derived OS code (yes, including stuff as strange as the original Apple TV, which was closer to a Mac running only FrontRow 2, as opposed to later models, whose pre-tvOS software was derived from iOS).
It's all mostly the same thing deep down, so they already a good chunk of that work cut out for them even if they chose not to do much more than that. The thing is, Apple is highly consistently and predictable… When doing a transition from PowerPC to Intel, they also made use of Universal binaries, Rosetta binary translation technology, a hacked-together DTK, etc., and Jobs explicitly said they kept Intel builds relatively updated throughout the years, so… considering how all those other factors seem to be repeating themselves in the very same pattern, like clockwork, the existence of a skunkworks project that kept all later builds running on ARM for all these years is a very safe assumption.
You know, as they say, “if it ain't broke, don't fix it”, and that formula seems to have worked out just great for Apple the first time. It's true that they claim to say “no” to a thousand things, but there are some internal and seemingly endless or potentially useless projects or tasks the company keeps going, just in case.
I wouldn't say they're running SPARC, PA-RISC or even PowerPC builds of their OSes, but I'm betting there's at least a team over there – or just some guys from the processor and core OS teams – keeping tabs on what's brewing in the market, like RISC-V. As for ARM… from the day they bought P.A. Semi and became a comparatively more advanced ARM chip developer (on account of being actually an ARM ISA licensee) than all other manufacturers of chips based on cookie-cutter ARM-designed cores – the biggest trojan horse move in recent tech history, masked by the more immediately momentous switch to Intel –, thus internally setting the goal of developing their own line of processors, it would be monumentally foolish of them not to maintain ARM builds of all their OSes as a backup plan at all times.
Last edited:
dmccloud
macrumors 68040
The biggest reason Apple switched from PPC to Intel in the first place was performance and production issues with the G4 and G5 processors. Now that Intel is facing the same issues which cost IBM the Apple partnership, Apple is going back to RISC architecture, albeit using ARM as the foundation rather than PPC. I have no doubt that Apple has had ARM-based builds of MacOS since the initial iPhone was released (especially since iOS is based on the Darwin core), but that is different from actively building Macs using Apple Silicon and designing processors for that purpose.
aednichols
macrumors 6502
Linus Tech Tips is going for it!
It's a high-reward proposition with seemingly little risk, they have no relationship with Apple.
It's a high-reward proposition with seemingly little risk, they have no relationship with Apple.