MacOS on Arm talk intensifies: Just weeks from now, Apple to serve up quarantini with Kalamata golive, reportedly Apple is reportedly inching closer towards unveiling its first Arm-based Macs, with official confirmation expected to take place at this month's WWDC conference. The report comes from Bloomberg's Mark Gurman, who quotes loquacious "people familiar with the plans". As previously discussed, the effort to ditch Intel for Arm is …
There is no need for an emulator for x86 Mac software. Unlike PowerPC and Intel, Intel and ARM have very, very similar capabilities. Most importantly same byte ordering. Practically all iOS developers use the simulator on their Macs that translates their code to x86 code, without problems. For anyone using Xcode, it's opening "settings", adding "ARM" to the list of supported architectures, and build.
So? What software authors can do with their source code only matters if they do, in fact, choose to do it. Computer owners are limited to the applications that are actually offered for sale at their local computer store.
Of course, Apple could require software companies, as a condition for placing their Macintosh programs on the App Store, to have both versions of their software. But since the Macintosh, unlike the iPhone, does not compel developers to give Apple its 30% cut, even that won't necessarily be decisive.
For the vast majority of software this is literally just a build option. For things which really need specific chip instruction sets, then that's where you get into interesting territory - a simple rebuild will yield poor performance.. probably.
Absolutely. If software has, either in itself or in any of its dependencies, gone and made heavy use of AVX SIMD extensions, then ARM is going to be something of a disappointment. AFAIK there's no ARM that has a vector unit anywhere near as grunty as Intel's AVX.
This was a problem last time Apple switched architectures. Software like Photoshop had been heavily optimised on PowerPC, making very effective use of that architecture's Altivec SIMD vector processor. At the time, Intel were all over the place (MMX, SSE, lord knows what else), and Photoshop for Intel was substantially slower at some tasks. It took quite some time for Intel to finally catch up with Altivec.
[Intel didn't fully match Altivec until around about 2012, which is when I think that Intel finally added an FMA instruction to their x64 line up. Prior to then they'd had an FMA instruction on Itanium processors, and the suspicion was that they weren't putting one into x64 in the desparation to give Itanium an artificial edge. By 2012 they'd given up with that, for they'd finally read the writing on the wall which customers had carved in letters 10 feet high and 4 inches deep.]
Another aspect is that there's way more to it than simply shoving in a comparable sized SIMD unit alongside an ARM core. To keep such a processing monster fed you need a very extensive cache system. So you end up with a chip that's pretty much as monstrous as Intel's are (in terms of transistor numbers). The savings come from not having all the transistors dedicated to pipelining x86 opcodes which will save silicon space / current / heat. Also, Apple can get ARMs made on TSMC's 5nm process. In contrast there's no sign yet that Intel have mastered their 10nm process (equiv to TSMC's 7nm), let alone worked out how to get smaller still.
Apple could have got x64 processors from AMD, which will get made on TSMC's 5nm process; in fact, they could buy AMD with some loose change. If they're not doing that, and are going all-in on their own ARM; well I guess they've got the money to do that, and it would allow them to forge their own path from here on.
There are some pretty powerful SIMD options available for ARM too...
There's even an ARM based supercomputer:
https://en.wikipedia.org/wiki/Fugaku_(supercomputer)
And with the lower power usage of ARM, they could clock higher or add more cores while keeping in the same power/heat budget.
GPUs are also good at a lot of the things that SIMD instruction sets are used for.
Yes, that's the successor to the K machine. The point is that an ARM based chip with a big SIMD unit on it is still a big chip with billions of transistors. Take the SIMD unit out of an 80W XEON and replace the X64 cores with ARM cores, and (all other things like cache sizes / architectures and silicon processes being equal), it's not going to be a 2W chip all of a sudden. Excising the X64 decoders and pipelines would save a good chunk though.
GPUs are good but suffer from being at the end of a bus. The time taken to load data and unload results can be pretty bad, and you have to have a meaty enough CPU to host it in the first place. For supercomputers, and high power embedded processing on aircraft, the "TeraFLOPs per cubic foot" measure is actually quite important. What is quite often found is that, by the time you've stacked up a suitable CPU to feed a GPU, and then overcome the load/unload time with additional hardware, the TeraFLOPs per cubic foot isn't very competitive compared to having just a CPU with a lot of internal SIMD units. AMD's APUs are quite interesting because they overcome this.
This shows up in the real world performance of supercomputers. What made the K machine special was that the real world accessible performance was very close to the benchmark performance that put it at the top of the 500 list, due entirely to the fact that data movement units and SIMD units were very closely integrated inside the SPARC based CPUs Fujitsu built. They're doing the same again, just using ARM instead of SPARC. There's been plenty of GPU based supercomputers that haven't really delivered performance outside of their benchmark scores, or their intended problem type.
Seriously? When I have to generate software that requires builds for both Windows and Mac, my code is littered with compiler metastatements all over the place. Anything that has to deal with the System Registry has to have Mac specific code to achieve some form of equivalent functionality. (And, no I'm not talking parameter storage which can obviously be handled by a common configuration file architecture.) Just about any Networking that isn't trivial requires API calls, etc. Yes, I do like maintaining one "code base", but even that statement is as much semantics as reality.
Haven't done deep software engineering for a while, but that was always cross compiled... and, more importantly, compiled to multiple architectures.
I've moved away from the coal face in terms of software dev, but I never considered the registry to be the "default" case, it was a special case which was best avoided.
Apple is Doomed.
but wait a moment.
MS failed with windows on ARM probably because they treated it as a sideshow.
Apple has a record of going all in on a tech only when they are good and ready to make it work. Their work on ARM silicon for their iDevices is impressive but it remains to be seen if they can translate this into a laptop/iMac/Mac Mini that improves on what Intel and AMD can do in the X86 world.
let me re-phrase the first line on my comment.
Apple is doomed if they mess this up big time.
We shall all have to wait and see what plays out.
I imagine vmare/virtualbox will continue to provide emulated processors as they can now.
Bootcamp... yep, you're out of luck. Can't recall how long it is since I dual booted (excluding the work laptop which booted into linux and opened the work windows install from the HDD until I removed my second HDD (in the optical bay) at which point IT were none the wiser - but i only had it boot into windows twice, once when I collected it, and once when it needed a windows upgrade that could only be done by IT for some reason. The rest of the time it sat in linux, and outlook (spit) was available via a VM.
Apologies - you're absolutely correct...
It's a while since I needed to emulate different architectures - QEMU and KVM where was was used.
Not *quite* as polished as the simple VBox solution, but still pretty easy.
*this* is where I'd like to be able to edit my original post - even if clearly marked as an edit.
You'll be able to use Bootcamp or VMware to run the ARM version of Windows. Which has built in execution of x86 Windows binaries via JIT translation.
Won't be fast, which isn't a problem for certain stuff (does it matter how fast Turbotax calculates your taxes?) but obviously would be for others. Apple going ARM might be the thing Microsoft needs to get developers interested in porting their Windows apps to ARM. At least that subset of applications that Mac users run on Windows.
Even if that ability isn't lost entirely it is likely to be at best crippled, and this simply reinforces Apple's move to turn an all-purpose computer with the ability to run the OS of your choice into one that runs the OS of their choice. I've already stopped buying Apple laptops; this move would stop me buying Apple computers full stop.
I won't buy a sealed, unserviceable, un-upgradeable brick for a small fortune, let alone the large fortune that Apple demands. About the only subset of developers who will be unaffected - or just have to suck it up - are the Apple and iOS app developers.
Microsoft do a demonstrably better job of supporting older software for extended periods of time, while Linux is perfect if you're working outside of the iOS/macOS and Windows bubbles, as most back end and web developers do. (I'd love to see how Apple's sales breakdown has changed over the last five years, and how they predict it will look three years after a move to ARM)
So, in summary, what you're saying is that you buy computers on the basis of religion or belief, not on gear matching rationally developed and evaluated requirements.
Fine. It takes all sorts.
I will first wait for Apple to do whatever it is presently rumoured to do, then let that settle and get the problems ironed out, and then I'll see if it suits my needs. If not, I go elsewhere, simple. I don't see the point of getting all foamy in the mouth about something that may happen, and of which implementation and deployment are presently about as clear as a tear gassed Lafayette Park.
I stopped buying Apple laptops because of their increasing cost and fragility. Fact, not religion or belief.
I moved from Windows to Mac in 2008 because it was a better platform to do my work on. More reliable OS, longer lasting hardware. Fact, not religion or belief.
I would not buy an Arm based Mac in any format because moving to Arm automatically reduces my operating system choices. Fact, not religion or belief.
I will move to Linux when I get my next (non-Apple) laptop. It is a much cheaper hardware/software package that allows me to do my job just as well as I can on my current Apple laptop. Fact, not religion or belief.
I would not buy an Arm based Mac in any format because moving to Arm automatically reduces my operating system choices. Fact, not religion or belief.
Er, what? You buy an APPLE device, which, by merit of its manufacturer, will run MacOS, irrespective of the underlying hardware. In addition, we have already seen from the switch from 32bit to 64bit that Apple is also quite decent in its platform migration strategy, putting in place interim solutions and giving people plenty of time to make the switch.
will run MacOS, irrespective of the underlying hardware
Catalina has shown that that's not true, not even for 32-bit software on the same CPU. There was no real reason to remove support for 32-bit software either.
If they do the move to ARM in the same way then you can kiss most of your existing software goodbye.
There was no real reason to remove support for 32-bit software either.
Sure, that's why even Microsoft has now started to drop 32 bit code. Given just how long Apple has been warning about the switch to 64bit code and how long they kept the wedgeware going that made it possible to run anyway, I think Apple's in the clear here.
The only complaint I have about Catalina is that it's too much Microsoft Vista in its "the mouse moved, allow yes/no" locking down of the system. It gets in the way.
MacOS and Linux both run on arm (MacOS will when this becomes a question) and Windows 10 has an ARM branch because ARM powered laptops are coming, whether you like it or not.
Apple won't be the first to jump, but once they do it will accelerate the market significantly - and Win10/ARM will see significant benefits from that acceleration.
So which OS can't you use with ARM?
Bootcamp, VMware, Virtualbox, etc
And, of course, Parallels.
I think this is as interesting for what it says about Windows - there's always been that nagging problem that there would be some (especially business) software that wouldn't be available for Mac natively and having a route to use it without needing additional hardware was hitherto been seen as a positive.
It looks like Apple probably reckon that most of that market can now be satisfied either by web-based SaaS solutions or by cloud-based Windows virtual desktops. That's not a prediction that will come as welcome news to either Microsoft or Intel.
I remember the transition from 68k to PowerPC (my SE/30 is sitting next to me, as we speak, awaiting a new hard drive). That went without a hitch, although I grumbled like hell at the time.
I remember the transition from PowerPC to Intel (my G3 and G5 systems are sitting in the loft, although the PSU on the G5 is a little on the noisy side). That went without a hitch, although it felt like the end of the world at the time.
I’m older now. Possibly wiser. Mostly mellower. I know that the switch to Arm will go smoothly, and we’ll all wonder what the fuss was about in five years time. I’m preparing for it already - by not buying a new Mac. I’ll wait until after the transition, thanks. Times have changed. Windows is an anachronism now, for all but gaming. The corporate cool kids all run macOS, the hipsters all run macOS, the server rooms all run Linux, those just want to get stuff done use Chrome. The only hold outs seem to be accountants and gamers (and Windows is very, very, good at gaming as we all know). The point is, who cares if you can’t run Windows on your Mac? You probably weren’t going to game on it anyway (I have a Ryzen powered Windows 10 box for the few occasions when I fancy playing games - and most people have a games console), so Mac’s ability to run Windows is neither here nor there.
"The corporate cool kids all run macOS, the hipsters all run macOS, the server rooms all run Linux, those just want to get stuff done use Chrome."
The corporate bigwigs and graphics people want expensive Macs despite teh software they use running on windows without any issues and the minions make do with cheap Windows boxes.
Most LOB apps are still Windows only and we have a Purchase Ledger system based on Silverlight which is IE only, so we have a direct app server runing IE for eth Mac users. 98% of our servers run Windows.
In corporate environment Windows devices are far simpler to manage, maintain, Control and secure. MacOS is a finikerty beast that causes issues and does strange things. Macs only work in a corporate environment due to the workarounds and fixes that are being done daily by your invisible IT support team that "no one knows why we pay them so much".
As for Chrome I can't think of a more insecure (it sends all your data to a someone else, surely that is the definition of insecure) browser that eats all the resources on your computer.
> In corporate environment Windows devices are far simpler to manage, maintain, Control and secure.
Then you get home and your Mac just works for years and years, while Windows requires extensive hand-holding to keep humming. Before you ask, yes I use both on a daily basis.
> As for Chrome I can't think of a more insecure (it sends all your data to a someone else, surely that is the definition of insecure) browser that eats all the resources on your computer.
Edge slurps more than Chrome does. FWIW I'm not a fan of Chrome either.
Actually the Mac's are no harder to manage when you have the correct tools like JAMF and MacOS gives you less grief than Windows generally. We have hundreds of Macs, thousands of iPads and hundreds of Windows computers (SCCM/Group Policy used for Windows, JAMF for MacOS/iOS), Windows is the one that is the pain and causes the most problems, like MacOS in the home, MacOS in the work place just sits there and works.
IBM now has over 200,000 Macs deployed worldwide and is the biggest corporate user of Mac, Facebook has over 13,000 Macs deployed, and Google has over 40,000 Macs.
Interestingly too Macbooks are not overpriced, if you go to say Lenovo's website and pick say X1 Extreme Thinkpad at £2059 and then customise it to match closer to the Macbook Pro 16" like upgrading the screen to have the higher resolution close to the Macbook Pro's and guess what? Hardly any price difference and the Macbook will hold its value longer and be worth something on resale, will have the best trackpad in a laptop you can buy and looks smarter. When you have the same class hardware inside (ie same speed/quality SSD's) the price gap suddenly disappears.
The price comparison is true. While it's easy to bung a few PC parts together, and come up with something with a lot more raw processing power than a Macbook Pro for a lot less, if you actually want a machine with equivalent specs (and not just better on one to two components), the prices are often a lot closer. For a while, the Cylinder Mac pro (which I have on my desk at work, and is a good machine, which, thanks to Covid, I haven't seen since March) was actually the cheapest way to get that particular model of Xeon in a pre-built computer.
Regarding management, we maintain systems for managing both at work. System Center Configruation Manager for Windows, and Munki/Deploystudio for the macs.. We are in the process of moving to Jamf Pro for the Macs (well, when we can go back to work), but we have *far* more problems with the PCs than the Macs.. Yes, Macs do require a different way of doing things, but that is life, and I am worried about IT technicians that can't cope with that kind of change.
I also recall the m68k to PPC transition...
A lot of software and even key parts of the OS remained 68k code for quite a long time.
Apple stopped making 68k machines at i believe the 33mhz 68040, and transitioned to the 50mhz PPC601.
The early PPC machines were often slower than the high end 68k machines because of the emulation overhead.
For software that was 68k only, running a mac emulator on a 68060 amiga was at one point faster than any real mac.
The 68060 itself was fairly competitive with the 601 even running native code.
I assume you didn't try to use an Xterm "cross-endian". Despite X having been developed in a world where both big and little endian systems were common, somehow Apple lost the recipe.
And I recall a problem with the calculator app. The Intel version muffed calculations made in the RPN variant of "programmer mode. No I don't recall the precise error, but I do recall checking it on Yosemite, quite some time after the switch, and it was still broken. Fortunately, I still have my HP-16.
Don't forget that back in those days the only alternative was a Blackberry. Whilst I didn't have anything against Blackberry devices, their pricing was non-competitive for random proles. Special plans and whatnot.
Poor UI aside, WM was really very capable at the time. I used and liked it all the way to the HD2.
Don't forget that back in those days the only alternative was a Blackberry.
In 2008? There were plenty of Symbian phones then ... and the first iPhones were already on the scene ... Android first appeared in September of that year ... and SWMBO had a Hansdpring/Palm Treo.
Apples nailed the last two processor transitions, keeping emulation working well enough for long enough that all but edge case apps worked until they got updated (or replaced).
If Apple announce this, I'm sure they'll be confident in whatever form of Compatibility they cook up.
Curveball option is they start making ARM Macs running a souped up iPadOS (Which getting closer and closer to a general purpose OS), and keep mac pros running x86 MacOS.
I'll buy a new MacBook, when Apple stop using the keyboard as the integral part of the MacBook's subframe, wrapping all other components around it.
It's pretty clear Apple's policy is to design a laptop from the ground up, around the most vulnerable part likely to fail, while also being the cheapest part (in terms of part cost) to replace for Apple.
A part that is so embedded (and unobtainable/costly to third parties), that the labour repair cost for anyone but Apple, is so high as to make it unrepairable. Yes Apple, we can see what you are doing.
Ditto for soldered SSDs and soldered Memory.
I used to be like you and upgrading my laptops and especially my MacBooks. But... (inevitably) I bought a 15in MBP in 2015 (16Gb/1TB/i7 cpu) and it is still going strong. Sure, a few of the printing on the keycaps is wearing out but it has been my goto machine for 5 years. If I can get another year out of it, I'll be happy to switch to an Arm based device. As I've gotten older the urge to tinker with devices I rely on has diminished dramatically.
My Windows Laptop (bought 2014) was a PITA once W10 defied my defences and installed itself so it now runs Linux and is still a PITA so next week I'll be installing a System D free distro. Isn't being furloughed wonderful.
From the article:
This is something Microsoft largely failed to accomplish with Windows RT**, its first foray into Windows on ARM*.
And fail they did, largely due to the lack of an imagination. Before then MS had shown at some trade show Windows 7, full fat, plus Office recompiled for ARM running on a dev board, working just fine (if a little slow). They also reportedly bought an ARM license. The expectation was that they'd do what Apple are now rumoured to be doing. Had MS decided to go for it back then, either assemble the silicon dev team or partnered with someone who'd already got one (Qualcomm?), they could easily by now have gone for it.
I know, I know, Wintel etc. It would have severely disappointed Intel, and that might have had consequences for MS. However, lets consider Intel. They could have licensed ARM ages ago, again, or still, depending, and could now have a healthy Intel SuperARM line as well as a honking great x64 server line. They might even have a presence in mobile devices had they done that. Instead, they've left the field wide open to Apple, Samsung and Qualcomm. Even AMD are dabbling with ARM.
Apple supported Rosetta for a little bit during the transition from PowerPC. If the developer of that critical piece of software had gone out of business, or didn't/couldn't want to port to x86, then you were stuck. I'm sure that Apple has learnt the value of backwards compatibility and long term support. /sarcasm
Apple will have to continue developing Intel-based laptops, particularly for users performing CPU-intensive tasks like video editing
Video-editing is processor-intensive but nowawadays it's just as likely to be the GPU as the CPU doing the work. But the ARM ecosystem, and especially Apple's part of it has had excellent GPUs for years now. x86 is traditionally better at single-threaded stuff but, again, the margins have been coming down for years because ARM is able to deliver improvements faster. The new (10th gen) Intel chips are pretty good but they also quickly reach TDP limits and start throttling. TDP may well be key for any kind of portable device.
As far as I'm concerned, the only acceptable new chip for Apple Macintosh computers, once Apple makes those chips itself, is one that runs 68020/68882 code, PowerPC code, and x86 code, so that it becomes possible for the Macintosh to seamlessly run all Macintosh software ever at full natiive speed. And, yes, that would mean including OS 9 as a component of future releases of OS X.
He’s well known to be not all there. Do a Google of usenet for ‘vat girls’. He has also advocated having the US invade Brazil, for the good of the environment and for the betterment of poorer Brazilians, who would welcome the liberators from the evil government they’re inflicted with. Note that he’s Canadian. Someone, possibly a fan of wolves, replied to the effect that the lead unit in the invasion should be the Princess Pats, with him tied to the turret of the lead vehicle. For some reason he never replied. ‘Tis a puzzlement why not.
Has Apple tried cleaning up the OS? They have compessed virtual memory, which sounds cool until you realize that you're swapping only because the compressed memory buffer is eating huge amounts of RAM. There must be 4 GB of various unused Apple daemons running that never swap out. The filesystem still doesn't seem to cache properly. Mix poor caching with constant compressed memory swapping and now the laptops are thermally throttled during any serious work. Oh, and periodic minor software updates somehow take 35 minutes to install. The battery in my late model Macbook Pro is more like a UPS than anything freeing me from an AC outlet.
I put Linux on an old personal Macbook Pro. It went from being unbearably slow to being responsive and capable for some tasks.
... with Apple futures had a few too many near me at a time you can guess from the rest of this.
Seems there had been an ARM-based Apple ][gs in the labs, but it had some advantages (price, power, stability) over the existing (68K) Macs, so it had to die.
(never worked for Apple myself, but there are plenty of watering holes near that infinite loop.)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
