Apple is beginning to undo decades of Intel, x86 dominance in PC market It took Apple less than a year to seemingly start undoing decades of x86 and Intel dominance in the traditional PC chip market. The Cupertino-based iMonster provided the boost needed for Arm-compatible chips to take noticeable desktop and laptop processor market share away from x86, said Dean McCarron, principal analyst at …
This post has been deleted by its author
I've got shoes that have lasted longer than Apple has stayed on a CPU architecture... Part of it is Apple is a big fan of dropping legacy support and telling their customers to thank them for it. Other manufacturers wouldn't get away with that. Let me know when Dell, HP, Lenovo, or another big name comes out with a non-x86 based PC, as THAT would be big news. Apple switching to the newest shinny is not.
"Let me know when Dell, HP, Lenovo, or another big name comes out with a non-x86 based PC, as THAT would be big news."
Just one? Because all of those places make ARM Chromebooks and some of them have also made some Windows on ARM machines. They're still primarily X86, but they have tried out the idea too.
But since the consoles are not as tightly tied to legacy hardware, they may choose to jump to another architecture if it matures enough. Don't forget, all the CPUs of the 7th generation (the Xbox 360's XCPU, the PS3's Cell BE, and the Wii's Broadway) were PowerPC-based. If someone can deliver an ARM-based system with reliable 4K performance, I could see the 10th console generation making a jump, too. Might also dispense with optical discs while they're at it.
>But since the consoles are not as tightly tied to legacy hardware
Oh, they are. Nobody's making the mistake of launching the next console without support for their back catalog, and it'll take a brave company to try and convert their studios' toolchains. Emulating the prior hardware and doing that kind of conversion is going to be very expensive. Apple pulled it off because they're a closed ecosystem, but it still took nearly a decade of chipping away at emulation problems to get things working.
So what happened with the eighth generation, which jumped from the aforementioned PowerPC-based XCPU and Cell BE to x64-based hardware, with as I recall NO support for the back catalogues of either? Both made the jump in spite of this because it opened up more-mainstream hardware support. The ninth generation is more PC-like than console-like at this juncture, but it could still jump the rails with the tenth generation. The only eighth-generation console that was backwards-compatible was Nintendo's Wii U, which became an infamous flop.
>So what happened with the eighth generation, which jumped from the aforementioned PowerPC-based XCPU and Cell BE to x64-based hardware, with as I recall NO support for the back catalogues of either?
Sorry, you're right, I should have said nobody's making the mistake *again*.
A huge part of the hardware vendor income stream is now based on recurring subscriptions which are significantly based on on-demand access to back catalogues. Have a peek at the landing pages for the PS5 and Xbox Series X. They both almost immediately trumpet their subscription services and backwards/cross-compatibility. That's not going anywhere - today's consoles exist to shift subscriptions and microtransactions, not game units or hardware.
The fact PC gaming and AAA console gaming are now one big happy x86-based family is not a mistake and won't be undone quickly.
I guess you don't know many gamers with multiple consoles: if people have a back catalogue then they also have a console for playing them on. For the manufacturers having blockbusters that get the most out of the new hardware is key and this is increasingly about the GPU.
Thats a nice theory, but people who do more than surf Crackbook and Instagram still need to use KEYBOARDS to input volumes of data and code and do NOT do that work on "phones and tablets."
Phones and tablets are useful gadgets, but they by no means replace a desktop or workstation.
What suffix of WINDOWS-suffix isn't a pain the arse? WINDOWS-360, a stunning blast from the past as MS thinks web-apps really work now that people's expectations are so low? Windows-tablet? Getting the thumbs up from 75% doesn't mean much when only 4 people actually bought them? Windows-[8,9,10,11,...] -- speak for themselves.
ARM is just a cpu architecture; it can't save WINDOWS from being WINDOWS.
* - OK, one benefit of WINDOWS-whatever is that it has a massive security hole, so that you can sign in from an unregistered device, and attend meetings with just a password. This is awesome if your company has an under provisioned VPN; and if your company doesn't have an under provisioned VPN, you should probably dust of your resume. Also, because Teams can forge the corporate credentials, it means even if you use a secure mechanism like webex or zoom, you can launch it from Teams, inheriting the credentials, bypassing their security mechanisms. MSFT for the win! They have brought bypasses to web apps!
Well, given that I have 2014 Intel machines that not only still work but also still get security updates (ditto for an iPhone 6 from that era, by the way), I'd say that I'm not bothered by Apple switching. Their support tends to last well beyond what I have seen from any PC manufacturer.
As for OS support, the 2014 machine has only now fallen off the update cycle (yes, after a mere seven years, clearly Apple is scandalously leaving people behind /s), yet it still gets security updates.
Also, I can still get parts, so feel free to stick with PCs - been there, done that. I'm glad I left all of that behind - I actually have work to do.
7 years is very short for software support for a desktop/laptop class device.
My last PC was seven years old when I decided I'd like a new one four/five years ago. So I gave it to a friend.
They're still using it and it's still supported - albeit only until 2025, as it won't run Windows 11.
My current laptop is a decade old and still runs a supported OS. It's on its third battery, but that's a trivial clunk-click swap.
Hardware on-site warranties are usually 5 years for non-apple desktops. That's free physical repairs, not software updates.
Weird how Mac users are grateful for only 5-7 years of support, while PC users are disappointed if it's less than a decade.
7 years is with mac os upgrades/new os. Security updates or as you call it "supported" continues.
Not all PCs have decades of support on the configuration they were sold with (esp with drivers on Win10).
Having Windows 10 1803 is different from having Windows 10 21H2 and not all HW is upgradable and thus suported. (This distinction with Windows 10 is lost because they have been doing it through version numbers and still calling it Windows 10 despite dropping support, whereas other vendors typically change the product name when support is dropped).
https://endoflife.date/windows
It is very likely your friend is running "Windows 10" but not a version that is currently supported at all.
Going DIY with linux is also possible with mac hw, but that was never the market Apple was/is targetting so you should not ever buy a mac if that is what you want.
That MacOS support comes with some huge caveats though.
Xcode requires the latest OS to install and run, and iOS development requires the latest Xcode. So while I have a 2011 iMac sitting next to me, for the last three years or so it has been completely and utterly useless to me as a machine I can actually use for anything I originally bought it for.
If you want to develop on machines that are that old, you should use to Linux where I think even a 486 could still be encouraged to give you a decent desktop but will probably compile in a time now measured in weeks. That is still just about the most efficient with hardware, but normally developers have fast machines because of the duration of the code-compile-test-repeat cycle.
I don't think it's feasible to develop anything Apple related on gear older than 4 years, that seems to be their shortest OS version cut off time.
"Google for Xcode MacOS compatibility."
Why not? Let's see what pops up. I'll start with DuckDuckGo, though, using your suggested search term.
First result is a person who can't compile for IOS 14.2 because the XCode that can do that requires Catalina and they're on Mojave. I.E. a required Mac OS update in order to support something which is entirely ordinary for a different platform. True, in this case, they can update, but they could have the same situation and be unable. Your score: 0/1.
Second result doesn't really mention it, so we'll call that a draw.
Third result is a user who doesn't understand how this works, and the answer to their question includes this statement: "Basically, there is no current Xcode that will work on your mac and know what to do with your phone." Sorry, 0/2.
Fourth result is the Wikipedia page for XCode. It has this sentence in it: "the latest stable release is version 13.1, released on October 25, 2021, and is available via the Mac App Store free of charge for macOS Monterey users." This means that non-Monterey users can't install it. 0/3.
Do we really have to continue? Apple cuts support for XCode to their latest OS, and that is required to compile for modern mobile devices. They don't want the support costs; I get it. However, it does allow them to be compared unfavorably with Android, which doesn't do that.
"First result is a person who can't compile for IOS 14.2 because the XCode that can do that requires Catalina and they're on Mojave. I.E. a required Mac OS update in order to support something which is entirely ordinary for a different platform. True, in this case, they can update, but they could have the same situation and be unable. Your score: 0/1."
If you're on Mojave, you can upgrade for free to Catalina (and further). Every system which can run Mojave can run Catalina.
I could challenge all of your points and rebut them one by one, but I'll just leave it at this. It's correct that Apple requires new(ish) versions of MacOS to run the latest versions of Xcode, for very valid technology, feature and security reasons; but there's no legitimate reason for not running the latest OS on any hardware manufactured in the last 7 years unless you specifically need to run a certain version in order to run or support legacy systems or hardware; in which case the version of Xcode that runs on that OS will be the correct one for that legacy hardware/system/app.
No, you couldn't rebut the point because you have ignored the one I used. I said outright that this user had the ability to update to Catalina and hadn't yet used it. However, there are OS releases where some hardware is cut off, and XCode doesn't keep compatibility with the old release then either. Whenever OS releases stop for some hardware, XCode releases stop too. In other words, exactly what the original complaint said: some hardware still works but can no longer run the latest XCode.
You can easily argue that this doesn't matter. You could argue that the support lifetime for feature updates (which is what you need for the latest dev tools) is long enough. You chose to argue that they were wrong. They weren't wrong.
"Whenever OS releases stop for some hardware, XCode releases stop too."
Any Mac hardware from 2015 onwards (2013 for Mac Pro) supports Monterey. There's no reason apart from legacy support to not upgrade to Monterey, and if you need legacy support, then the Xcode version for your legacy device will run on the appropriate (in age terms) system.
You're inventing a problem that quite simply doesn't exist.
That is the point that started the thread. The person has a Mac from 2011, which is fast enough to run modern XCode but for OS reasons cannot, and they think that speaks ill of Apple because they can run modern tools on Windows or Linux on a machine of the same age. They are saying that Apple's support lifetimes are shorter than everyone else's, and you decided to respond by denying the truth of that.
You've admitted it now. In order to run the latest developer tools, your Mac needs to be from 2015 (proviso about the Mac Pro excepted). We agree on the facts. Would you like to provide a reason they shouldn't care, because so far, you haven't.
"I don't think it's feasible to develop anything Apple related on gear older than 4 years"
Oh come on. A lot of these developers are building apps for IOS or Mac OS which do one thing and are smaller. They're not building the kernel. If they were, something five years old would be slower, but they could wait that out. However, they're probably building something which can be fully compiled in a matter of minutes and their changes can be compiled and linked for testing in seconds. The newest machines could perhaps do that in fewer minutes and fewer seconds, but it would still be the same order of magnitude.
There are some systems that stall on Windows updates for some hardware reason, but that's really not that many. Until Microsoft decided to change their policy with Windows 11, you could try to run the latest Windows 10 on basically anything, and most of the time, it would work. In order to prove that, people have been bypassing the system checks on Windows 11 and showing it running on really ancient things. While some computers from 2014 have had drivers dropped, there are a lot of machines from 2008 which indeed can run 21h1 Windows 10 (21h2 is Windows 11) and are being used that way right now. Those will continue to get security updates until 2025.
Apple has entirely earned praise on software support for their mobile devices, as they have supported their devices much longer than any competitor, even as some Android manufacturers have been extending theirs. They do not have the same credentials when it comes to desktops. They support their desktops for a moderate time, less than Windows is supported (unless MS's Windows 11 policy continues to obsolete more things, in which case they could pass them in the race). This is entirely without considering Linux which leaves both well behind. Apple still has a large edge over Chromebooks, but unfortunately, they have not earned the praise you are giving them.
I just sold my 2014 laptop. It had a new battery, but apart from that it's a perfectly functional machine and has been from the day I bought it, although I must say that I have never managed to get it much beyond 10GB of RAM in use (it had 16GB) - it just didn't need it for my work.
I reformatted it before selling it, so it runs the latest version of Big Sur with all the security updates (which is a default) and it has never really failed in its entire lifetime, ever. Open the lid, work, close the lid, done (unless you want full disk encryption to work -built in-, that needs a full shitdown). For years.
I had super expensive machines, even Sony VAIOs, which barely lasted more than two years..
"Well, given that I have 2014 Intel machines that not only still work but also still get security updates (ditto for an iPhone 6 from that era, by the way), I'd say that I'm not bothered by Apple switching. Their support tends to last well beyond what I have seen from any PC manufacturer."
Is this a joke? Pretty much *every* intel Mac has been supported much longer by Apple's main competitor (Microsoft) than by Apple itself. As someone else said, 7 years is nothing for a Windows PC.
Oh, and as to Apple's "support" for previous macOS versions, unless you're on the latest version you may have to wait a very long time for getting any updates:
https://arstechnica.com/gadgets/2021/11/psa-apple-isnt-actually-patching-all-the-security-holes-in-older-versions-of-macos/
"Pretty much *every* intel Mac has been supported much longer by Apple's main competitor (Microsoft) than by Apple itself."
This is not always true, and you forget the driver story with Windows/PC. Also things like BIOS and EFI security updates. The majority of these "supported" Intel processors as you put it do not have BIOS fixes for eg for the spectre bug and variants. Only the extent of protections offered by the OS SW based mitigations exist.
When Apple say supported, it is the whole thing from boot firmware, drivers, os, hw, service support.
Put another way, any SW not written by Microsoft is almost certainly vulnerable on older PCs.
Also as for Microsoft, support is dropped but masked by the use of "it is running Windows 10", which is the latest right?
https://docs.microsoft.com/en-us/windows-hardware/design/minimum/windows-processor-requirements
shows how processor support is dropped across versions of Windows 10, and those older versions of Windows 10 are not supported (for consumers). So someone running Windows 10 1511, who cannot upgrade Windows 10 21H1 as their processor or HW is not supported, is worse off. They don't realise they have a system that is not secure, because they think they are running "Windows 10", which is the "latest".
"This is not always true,"
No? So then tell me which intel Mac has seen longer support by macOS than by Windows?
"and you forget the driver story with Windows/PC. Also things like BIOS and EFI security updates. The majority of these "supported" Intel processors as you put it do not have BIOS fixes for eg for the spectre bug and variants. Only the extent of protections offered by the OS SW based mitigations exist."
That is not true. Pretty much everything starting from Sandy Bridge and later has seen fixes for these issues, and BIOS/UEFI fixes have been quickly made available by the big PC vendors such as HP and Dell. Of course, the latter may not be the case for any junk system out there but that's simply the result of buying, well, junk which its manufacturer has no interest supporting.
Besides, it's not that the same CPU bugs didn't affect Macs. For example, the Nehalem/Westmere based Mac Pros are as vulnerable to the bugs in its processors as any other Nehalem/Westmere system. However, while Apple has stopped support for these computers after Mohave (i.e. three macOS generations ago), they happily run the latest version of Windows 10. Which contains workarounds for at least some of the bugs in its processors.
And then there are the unfixable bugs in Apple's own hardware, like the early T2 security processors in some Macs which have an unfixable critical security hole:
https://appleinsider.com/articles/20/10/05/apples-mac-t2-chip-has-an-unfixable-vulnerability-that-could-allow-root-access
"When Apple say supported, it is the whole thing from boot firmware, drivers, os, hw, service support."
Sure, just that Apple may or may not actually fix a problem. And I'm not just talking about the many instances where Apple denied responsibility for problems caused by idiotic design decision (like butterfly keyboards or too short display cables, or spontaneously cracking displays on its early M1 Macbooks). Because unless you're always on the latest macOS version then you may have to wait a long time to get patches for a security problem that was quickly fixed in the latest version of macOS:
https://arstechnica.com/gadgets/2021/11/psa-apple-isnt-actually-patching-all-the-security-holes-in-older-versions-of-macos/
Then there are the various minor issues which remain unfixed by Apple, like the issues M1 Macs have with many standard monitors.
That's Apple "support" for you.
"Put another way, any SW not written by Microsoft is almost certainly vulnerable on older PCs."
The same can be said about a Mac, which normally runs 3rd party applications.
But as shown above, even if you're only using Apple software you may well be left out in the cold waiting for a fix for a critical update.
"Also as for Microsoft, support is dropped but masked by the use of "it is running Windows 10", which is the latest right?
https://docs.microsoft.com/en-us/windows-hardware/design/minimum/windows-processor-requirements"
What you miss here is that "support" in Apple's circles has a different meaning than "support" does for Microsoft. Because just because Apple "supports" a specific Mac model or macOS version doesn't mean it's actually providing fixes for problems that are discovered on these products. For Microsoft, "supported" means that it will actually go and fix problems that occur on a supported platform.
Also, because Apple actively prevents the installation of new macOS versions on unsupported Macs, having a Mac that's "supported" is a lot more important on the Mac side than it is on the PC side. Even Windows 11 installs fine on older PCs (the oldest one I have it running is from 2012, and that's only because I have no PC that is older to try) without any hacks.
>> Which contains workarounds for at least some of the bugs in its processors.
So having "some" workarounds is supported or not supported?
>> Even Windows 11 installs fine on older PCs (the oldest one I have it running is from 2012, and that's only because I have no PC that is older to try) without any hacks.
So if it installs but does not meet ms minimum requirements, and MS says they do not support it, is it supported or not?
>> Pretty much everything starting from Sandy Bridge and later has seen fixes for these issues
Sandy bridge is 10 years old, so older processors are thus not supported. This is from your statement. This means the oldest supported PC would have to be *at most* 10 years.
At the time of writing, https://endoflife.date/macos has 10.15 still receving security updates.
Assuming it will EOL in 2022, the supported HW is https://en.wikipedia.org/wiki/MacOS#Hardware_compatibility is upto mid 2012. So that means a 10 year support from Apple. Earlier version show a 9-10 year support.
>> What you miss here is that "support" in Apple's circles has a different meaning than "support" does for Microsoft.
So what is the definition of support then given you are happy to accept the compromises with MS and call it supported but not with Apple, which you call unsupported?
Intel themselves declare EoL for their products well before 10 years..
https://www.intel.com/content/www/us/en/support/articles/000022396/processors.html
So how is your PC "supported" after Intel has declared EOL?
Anyone running a pre Sandy Bridge config is unsupported.
To be clear being able to install/run a newer OS is not the same as being supported across the product. If you have an insecure driver MS is not responsible for that. If you have a processor than can be hacked, MS is not responsible for that.
You are confusing being able to "run" a newer OS as "supported", when it is just actually ignorantly running an insecure, buggy system.
Apple is responsible for both HW and SW has to choose whether to support a now abandoned platform by the respective supplier - Intel in this case.
HP, Dell have support cycles shorter than Intel's 5 years, and MS is not responsible for the HW nor the drivers. This means MS can keep selling you SW that could be secure, but is actually insecure, broken or buggy on specific HW. They consider that a PC HW manufacturer issue, while HP, Dell and the like have long stopped supporting the platforms.
What you really are saying is that on PC you can run newer OS even if you end up with a buggy/insecure system that has actively exploited and publicised vulnerabilities. This I agree, it is more difficult with a mac, as Apple clearly state that they do not support the HW, as for eg, Intel has stopped support. You as a mac user are informed of the EoL and can choose to run older HW, with awareness, rather than the PC world through ignorance.
You are just getting a false sense of security on a PC because you are installing "newer" MS OS on it, without realising that the security problems on older HW are yours, as MS, Intel, and the OEM has abadoned parts of the system, when taken as a whole.
You are missing or misconstruing a few of their points and drawing incorrect conclusions from them.
Them: "Which contains workarounds for at least some of the bugs in its processors."
You: "So having "some" workarounds is supported or not supported?"
The thing here is that the OS provider, Microsoft, is not the processor provider, Intel. Microsoft made some fixes to Intel's problems, but they cannot make Intel change the firmware. Intel has done that in hardware back to the sandy bridge models. Microsoft has used software to deal with those Intel chose not to do.
Them: "Even Windows 11 installs fine on older PCs (the oldest one I have it running is from 2012, and that's only because I have no PC that is older to try) without any hacks."
You: "So if it installs but does not meet ms minimum requirements, and MS says they do not support it, is it supported or not?"
Good question, and it's subjective. I would say it is not supported. However, it's not supported under Windows 11. It almost certainly is supported under Windows 10 (yes, the latest version of it), so it counts until 2025.
Them: "Pretty much everything starting from Sandy Bridge and later has seen fixes for these issues"
You: "Sandy bridge is 10 years old, so older processors are thus not supported. This is from your statement. This means the oldest supported PC would have to be *at most* 10 years. [And Apple has ten years also]"
Wrong. Intel has fixed a security vulnerability in their hardware for ten-year-old chips. Microsoft has patched it in software for older machines, thus supporting them. If we say that support must include fixing hardware security vulnerabilities, then Apple has a zero-year support lifetime because they have not made any effort to fix their T2 security problems. And that chip, they made themselves. Blaming Microsoft for a thing they had to work around because Intel chose not to fix it is a very different proposal and much less reasonable, most particularly because Microsoft's fixed theirs and Apple's done nothing. In addition, you have AMD chips which didn't have Intel's problems, and they're also supported.
You: "Intel themselves declare EoL for their products well before 10 years.. [reference link] So how is your PC "supported" after Intel has declared EOL?"
The OS is supported because it can run and it provides security fixes for OS problems. The same reason I don't automatically count an Apple machine as unsupported when they say they won't fix some problem as long as they do continue to provide updates.
You: "What you really are saying is that on PC you can run newer OS even if you end up with a buggy/insecure system that has actively exploited and publicised vulnerabilities. This I agree, it is more difficult with a mac, as Apple clearly state that they do not support the HW, as for eg, Intel has stopped support. You as a mac user are informed of the EoL and can choose to run older HW, with awareness, rather than the PC world through ignorance."
Rubbish. If I run the latest Windows on insecure hardware, I have my hardware's security problems, which I can look up, but I don't have the security problems fixed recently by MS. If I was forced to use an older version of Windows, I would still have my hardware problems but now I'd have the OS bugs too. That is the difference. Apple does continue supporting the OS, but not as long as Microsoft does.
All you've done is separate the system from its constituent components and thus give a hybrid statement of "supported", i.e. if one of the constituent components is supported, then that is the supported period of the *system*
>>Apple does continue supporting the OS, but not as long as Microsoft does.
Windows 10 and XP are aberrations. Microsoft sells OS, not systems.
Someone commented earlier that the minimum support expected for a PC is 10 years, such a support period does not exist as a system.
Intel's own microcode fixes were not rolled into all BIOS manufacturers.
>> I have my hardware's security problems, which I can look up
How many users exactly do this?
The point I'm making is that PC *systems* do not have the extended support being alluded to here.
You might as well install Linux on a Mac then and say it is thus "supported". Hell install Windows 10 with bootcamp on mac. Do macs now become supported?
If so - Apple offers bootcamp on Intel macs.
install a MS OS on an intel Mac and you are supported as any Windows PC, and install Linux on a mac and you are as supported as any Linux PC.
Fuck it, I'll bite. (We need a jaws icon.)
Supported, for me, means the ability to run the latest development environment. In Apple's case, X-code. I'm probably about to move to my fourth Mac Mini. (Still umming and uhhhring about renting.) In the same time, I've had two PCs.
It's true that my current Mac is getting sporadic security updates and the latest version of Safari. But without X-code it's worthless. A far older PC can do valid development work for Android and although I would need a newer PC to test on Win 11, I'm at least a year away from that. But I'm happy to call it 3-4. Wintel is still winnng the support war; and even if they were equal numbers, I'd have spent more on fruity hardware. More cash should get me better support.
Cough *BS*. I have a half dozen macs here, the newest of which is 2011. They are still updated by apple -- wtf would anyone run windows on them? -- but they are at a bit lower level of support. CVE's trigger updates; but features don't.
If I wanted to switch them to windows machines, it would still be moving backwards.
What version of MacOS are you on that is still receiving updates?
I have a 2010 MBP that is stuck on Lion and IIRC hasn't seen any security updates for a while. It's had a new battery and the rest of the hardware is fine. While I do understand Apple wanting to EOL some stuff, some of the reasons are bogus: with the MBP they basically decided not to want to work on the graphics drivers and many of the OS versions I've worked with have contained bugs that were only fixed by switching to a newer version.
OTOH commercial assets are generally depreciated after 5 years at the most and hence provide an incentive for replacement.
So what about all the issues that even though iOS is getting security fixes, there is plenty of software out their that will not run on it because it is not supported.
How long a piece of hardware is "supported" for is utterly irrelevant if it cannot run the software I need it to.
This applies across the board for any OS however the issue is most prevalent on iOS.
I don't know about that. While the Apple IIgs wasn't the brightest spot of Apple's history, the 65c816 powered the Super Famicom/Super NES and edged out the 68k-powered Sega Genesis/Mega Drive for the title in the fourth console generation. Not to mention the IIgs had some very interesting sound hardware thanks to Ensoniq and its ES5503.
It has fantastic sound hardware — for me that’s the only bright spot. Otherwise it’s a framebuffer-only machine with nothing even close to the grunt necessary to do decent animation and a vertical resolution too low for productivity, and the memory layout is so arcane that even the official documentation names one of the many overlapping registers as the quagmire state.
The 65816 ends up being a net detriment because it has the same inefficient memory access patterns as the 6502 (there’ll always be two reads for single-byte instructions, read-modify-writes always have a spurious access in the middle, etc) but in a machine where large chunks of the address space are behind a 1MHz bus.
"Apple transitioned much more quickly than anyone expected," McCarron said . . . .
Apple transitioned on the timeline they've been signaling for over a year. "Anyone" = people who read too many Mac rumor sites before hiding under a rock last year.
"It's not so much [Intel] are lagging; AMD went from a modestly competitive to a more significant competitive," McCarron said.
Intel's product performance is behind or on par with AMD's (depending on the job) and well behind Apple. That's called lagging. Does anyone know what a "significant competitive" [sic] is?
Can we please either not center stories around consultants' drivel or find the unicorn consultant who can use words competently? Thx.
Transitioning to this maturity in 1 year is impressive - none of the others, such as Windows on ARM, have transitioned to the same maturity as Apple, despite the latter having announcements far earlier than Apple.
So yes, for anyone who understands what it takes to transition support across processor classes, Apple has completed it far quicker and far better, than other attempts, current and historical. (Your comment reeks of someone who does not understand what it takes, and has never worked with multiple ISAs)
Indeed Apple is the only one who has completed it to the point a consumer can consider it, others like Microsoft call it "completed" but the performance is rubbish to be a viable choice.
So it is much quicker than expected, primarily because of the unexpected performance of Rosetta 2, and the most certainly unexpected M1 x86 translation acceleration HW (which they did not announce, and quite costly to spend die area on).
""It's not so much [Intel] are lagging; AMD went from a modestly competitive to a more significant competitive," McCarron said."
That depends on how you look at it - in the race did AMD as a laggard catch up to Intel, or did Intel as the lead in the race slow down. AMD does not charge the same as Intel (yet), so to me the laggard AMD caught up in the race, with their Zen microarchitecture and Lisa Su's leadership. Intel's product releases have matched their historical generational performance increments.
Comparing to Apple/ARM is a different point, that quote is discussing Intel and AMD. It is true that their historical performance increments is no longer enough with Apple M1 in the picture, but then it is not yet a threat either because M1 is not offered across the segments Intel operates in.
An M1 is no threat to Intel's server market (well yet), so your comment is analytically incomplete and shallow, to just their consumer segments, and to the subsegment where the consumer is willing to pay for a pricey Apple mac.
If you want to critique analysts do it with analysis, not prejudice.
"Transitioning to this maturity in 1 year is impressive - none of the others, such as Windows on ARM, have transitioned to the same maturity as Apple, despite the latter having announcements far earlier than Apple."
Not really. Because the actual transition didn't happen in a year but more like 8 years. 8 years in which Apple added expertise to design its own silicon and switch is OS base (remember that iOS is a mobile OS derived from macOS) to that new architecture. A transition that started in iPhones and iPads and eventually moved towards Macs.
But yes, Apple is unique in having managed such a transition so smoothly, but that is mostly down to the fact that Apple owns both the hardware and software platforms for its products. Which gives it widely more control than other manufacturers like Microsoft have which are still dependent on 3rd parties making their processors (and Qualcomm is mostly busy with milking its ARM dominance in the mobile space than to really innovate, which means everyone other than Apple is stuck with slow silicon). So there's that.
>>switch is OS base (remember that iOS is a mobile OS derived from macOS) to that new architecture. A transition that started in iPhones and iPads and eventually moved towards Macs.
You are confusing the transitions.
Apple has access to the macOS source code. They needed to recompile their source (in C, ObjectiveC, C++, Swift, etc) to ARM for iOS. This is not JIT translation, they do not convert instructions, they *recompile*. They needed an ObjectiveC, C++ compiler for ARM. All can be done on a development PC to generate the final executable. That executable is generated once, The first bog standard ARM processor (from Samsung I think) was enough. The equivalent is an M1 app, what a third party developer would do and update the app store with.
What they do on M1 for Intel apps is different, they run x86 *third party* CODE (already an executable) on ARM processors. They do not have access to the application source to recompile. They need to translate the instruction set to run on ARM without performance penalities. Rosetta 2 was not used at all for iOS. Rosetta 2 does not run on any iPhone not the first one, not the latest one. They are all native ARM executable.
Apple did not choose a VM as Android has done. Remember iOS apps do not run on x86 macs, but they do on M1/arm macs.
These are very different problems/transitions.
>> Which gives it widely more control than other manufacturers like Microsoft have which are still dependent on 3rd parties making their processors (and Qualcomm is mostly busy with milking its ARM dominance in the mobile space than to really innovate, which means everyone other than Apple is stuck with slow silicon). So there's that.
I sort of agree and disagree. Making an equivalent processor, with the die area that Apple is using for M1, is expensive.
The M1 is 119 mm2 with ~12 bn transistors, whereas if we take the flagship snapdragon, it appears the non Apple market at most will accept the price of a ~10 bn transistor processor, ~84 mm2. https://en.wikipedia.org/wiki/List_of_Qualcomm_Snapdragon_processors#Snapdragon_888/888+_5G_(2021). And that processor includes a 5G modem in that transistor count, which the M1 can again use for additional cache, accelarators etc. So Apple has the additional area and the modem area for performance. That's a lot of real estate.
If the market would accept the price I am sure that a more capable processor could be created. But Apple can charge a premium, something difficult for microsoft/qualcomm.
If someone has to pay as much as a Mac, they'd probably just buy a mac?
I will point out that Apple have gone "all-in" on ARM.
Microsoft has "added" ARM as an afterthought, or as a kick in the pants to Intel to get its arse back in gear. Google is also going 2-pronged about it with their Chromebooks working on Intel and ARM.
The problem being, to make a laptop, let alone a desktop, chipset involves a lot of work and a huge investment in R&D. And the current mobile SoCs don't provide near-laptop performance and the units that the chip manufacturers can hope to sell in the current climate are miniscule. The investment needed to make the performance acceptable would take billions over a few years. That would make the price per unit unacceptably high, so just minor tweaks are being made, which leaves the performance wanting, but good battery life.
Microsoft also don't seem to be in a hurry to make their software ARM compatible, after 2 generations of Qualcomm SoCs in Surface devices, great swathes of their product pallet were still Intel only.
Apple, on the other hand have made that investment in tuning ARM for laptops and says it is doing the same for desktops going forward. It announced the move, once it had the first SoCs available for production and announced at the same time, that Intel was dead, going forward, when it comes to Apple products. And the first products they showed off were much faster than the Intel parts they were replacing, whereas the Qualcomm parts are much slower than Intel parts under Windows.
It does mean the usual no-prisoners attitude from Apple. If you have old legacy hardware and software that won't run with/on ARM based kit, tough, buy some new kit/software or you are stuck with the old Macs until they goes out of support.
Again, that is something Microsoft just can't do. Most corporate customers will be unwilling to spend millions upgrading lab equipment or production lines, just because their old control software won't run on ARM and the manufacturer only supplies new software with new lab equipment or production lines. You just aren't going to pay out several million on a new production line, because a 400€ PC needs new software.
(I currently use a Ryzen 1700 PC and have no M1 based kit, so I am not speaking as an Apple user.)
I remember the big PPC Transition Plan roll out at WWDC in 1993. The first PPC Macs shipped mid 1994 and by mid 1995 it was an all PPC lineup. And for the next 10 years we ran fat PPC/68K application until MacOS X broke everything. Unless you were doing low level 68882 / 68851 stuff no problems running under emulation.
As for desktop / laptop PC processor market share. Apples world unit share has never wandered much from the 4% to 8% range in the last 20 plus years so no big change there. I think Intel/AMD can survive with a 90% market share rather than a 100% market share. The problem for Intel is outside the Wintel market and servers its an ARM world. Not just phones but tablets and chrome books.
I was there and mostly agree with you. Particularly the server/performance per watt ARM situation. As far as I can tell, x86 will continue to bleed market share due to ongoing performance pressure though. Still, it will be a slow decline, since there is enough legacy software that needs x86, and gaming prefers it. The only thing that could change this, would be MS pushing more development efforts into ARM. However, I think this is unlikely, considering the direction they appear to be going with Windows 11.
> So no one around for the PPC transtion....
I don't know WWDC. I do know my school got a PPC Mac, I set it up, my jaw dropped. That thing was FAST!!
No trick. I ran Roy Longbottom's PC benchmarks (adapted from mainframe trials tests) and made all the Intel hotrodders on PCHW jealous. Even a "486" (Am5x86, rated 133 MHz and goosed faster) hardly touched that first Mac PPC-604. The early Pentiums hardly nudged it.
WWDC was the developers conference. '93 was the last time I was able to get someone to pony up the money. Almost went in '96 but after '97 no point. As the Apple software ISV ecosystem that had been such fun from '78 onwards, first the Apple II, the the Mac, was dead. Gave up going to MacWorlds for the same reason. All the old faces were gone and even working stands was no longer fun. Working stands in Moscone from 1985 to 1995 was a blast. Trade shows tend to be a drag unless one is working a stand. Everyone comes to you at some stage so you get to see everyone over the two or three days.
The PPC roll out was almost a fiasco. Saved by the emulator guys. Plus the fact was the new 68060's were faster than the PPC 601 when running code compiled natively. The '60 had a RISC core and was superscaler and the existing compiler optimizers did a good job of producing clean object code. The PPC really needed a great instruction scheduler in the optimizer to get good results Without that the results were anemic. A problem never really solved on the Mac until the PPC was abandoned. As Apple had screwed the one company that could have fixed that problem, Metrowerks. For embedded PPC's Green Hills shipped a fantastic compiler that produced very tight code. But Apple plodded on with their very buggy GGC compiler which produce crap code until they took it out an shot it, replacing it with LLVM. Which does a much better job.
The 68K to PPC transition was voluntary, but misguided. The PPC to x86 transition was involuntary and was always going to be temporary. Apple ships mostly iPhones and iPads and the Macs are just outliers by this stage so it was just a matter of not if but when. Saying that, as a bare iron guy the ARM is a joy to work on just like the 68K. x86 is pure hell, sloppy and stupid, although AMD x64 made it less horrible. Slightly. PPC I have mixed feelings about. The pure IBM versions was not bad. But the CHRP versions had weird limitation which made it as irritating as SPARC sometimes.
A single vendor can never achieve this. Yes it has many fans but is tiny in comparison to the hordes of business laptops, business desktops all from a variety of brands.
The closest company to achieving this is Raspberry Pi Foundation. This has monopolized the IoT market. Up to 40% of them sold have entered the industrial market. That is a huge amount.
They had to make a clean switch fairly quickly. People were already avoiding upgrading the 16” MacBook Pro as they didn’t want to be stuck with the previous processor platform.
Also the performance boost being touted by Apple and by independent reviews of the first MacBooks with ARM chips was extremely positive and the switchover seems to be pretty smooth.
Had they dragged it out, they’d have just lost upgrade sales and had to support multiple lines rather than what’s usually a very slim portfolio of models.
Intel’s biggest risk is it’s own inertia. The level of dominance they had for so long has probably not been great for the industry. Hopefully the ARM chips and AMD drive serious completion and we might see some acceleration of innovation.
Also I’m not sure that this will drive change in the PC market, which is basically generic Windows machines.
Apple has always been in its own niche / bubble. I don’t think that really changes and in general PC makers can’t make that switch, as few, if any of them, have the technical resources to do so on their own and they don’t have control over the OS or software ecosystems.
It is the performance parity offered that is impressive. Achieving that in 1 year later is rather quick.
Others offer just about passably bearable performance (windows on arm)
Apple says "you'll get the performance you're used to with x64 SW, but go even faster with M1" For consumer SW vendors now, the fastest SW they can offer now is on the M1 (across PC/mac, Intel/Arm, x64/arm64 consumer segments) you'll have to go workstation class on other architectures.
I estimate this as a five to eight year piece of engineering - JIT translations, HW microarchitecutre optimisation, die area and cost/margin considerations, x64 patent concerns.
Looking back I think they started on this when they announced ARM64 on their iPhone, with the Apple A7. 64-bit had little purpose to mobile devices at the time (and costly having wider buses on die area), but if they decided then to have a converged ARM based product portfolio, ARM64 would have been mandatory.
So in that context, if we knew Apple was working on this for 7 years, it isn't that impressive.
But hindsight is 20-20, they have delivered a piece of engineering that takes that long, seemingly just a year from announcement.
It is great marketing, and kudos to their product and strategy teams.
Most other companies I have worked with realise a strategy change a year before it is required, unhappily give engineering two years (where 8 is required), engineering take a year to develop what they can, marketing announce, and then a year to mature it, just enough to eek out a competitive edge. This is Windows on ARM, where a handful or so of intel SW can run, as passable tolerable performance.
Apple don't just have an edge here, they are on another orbit right now in terms of translation performance in the market, because they gave their engineering 8 years, and can run just about any Intel SW on their M1, *at speed consumers are used to*. They are not asking intel SW users to tolerate anything on their M1.
It is the performance parity offered that is impressive. Achieving that in 1 year later is rather quick.
I don't think it was that surprising. Apple have been shipping their own ARM based CPU with their own OS for some time. You may have heard of it: It's called the iPhone. So the core technology combo isn't new to them.
With Apple being so secrative, we have no real idea for how long they've been working on the ARM varients of the Mac. The M1 may be the first, second, third, or whatever version of their ARM desktop CPU.
Apple has been working on its ARM processors since about the time of the iPhone 4s. Performance improvements year after year. Around the time of iPhone 7/8 I’ve seen articles saying “you could actually use these chips for a real computer”. A few years ago there was the story that the fastest chip in an iPad was faster than 90% of all laptop processors sold. The first prototype for an ARM Mac had the processor from an iPad. And now it’s the other way round, the top-of-the-range iPad uses a M1.
So this has all been in the works for a very, very long time. Apple did strike when their low-end M1 beat all the quad core Intel processors easily. The current mid-range chips are up with a 14 core Intel processor. Next year they’ll combine two or four of these mid-range chips on one carrier, and that’s game over for intels high end.
I'm typing this on a One Netbook,One Mix 3. I have a OM-2 as well, and the original One Mix. I removed windows 10 and loaded Debian, because I'm geeky like that, but I ran the Win10 for a bit, seemed to be just fine. Tablet mode OK, etc, etc. If I could get an ARM with similar performance running Debian it would be on my list. Iff you want to tote a "Gaming Desktop" disguised as a laptop more power to you. With a fully functional USB-C dock, I snap in and dual screen the local 2560x1600 display with an identical monitor bluetooth KB and rat.
The 8100Y is a little light. They have a new i5 with 16G that would be interesting, but frankly this does everything I want with a form factor I can tote, and a reasonable keyboard. I also had several of the smaller netbooks, the HP and the ASUS to name a few. It's kinda more around general performance, and what you might want to do with the machine.
This one isn't about chromebooks, etc... It is about how quickly Apple transitioned its x86 legacy to its modern future. Its modern future is based on cpu technology far faster than the legacy x86 it discarded. Macos is still Macos; love it or hate it, it isn't chrome os, it is still actual UNIX running in a pretty desktop environment.
The original netbooks (like the Asus 701 etc) running Linux were great little machines. Small/portable yet fully functional.
The main problem was that Microsoft could see their lunch being eaten and then forced the various manufacturers to roll out XP variants. While XP could be installed, the performance wasn't entirely great.
Which then led onto the next problem of people complaining about the performance and screen size etc, which then led to the netbooks becoming bigger/more expensive over time and in the end just became laptops.
The PC was successful only because IBM opened up the specification of the architecture so that other manufacturers could also make their own PCs.
It was also the reason why other architectures like Amiga, Atari ST, Sun failed.
If Apple does not open up the M1 and allow other manufacturers to build their own computers e.g. based on Linux then this walled garden will collapse in few years.
Bookmark my comment.
The Arm architecture is readily available to licensees. The likes of Samsung and Broadcom and others are testimony to that.
What Apple has done is copyable by others. Aren't the likes of Google and Facebook busy trying to copy Apple right now?
The evolution of the M1 is right there for anyone to see. They have taken many steps to get there. It didn't happen overnight but as the article says, at least 8 years.
Apple has been lucky in that those steps have gone hand in hand with improving the performance of the iPhone/iPad. That has funded the development of the M1.
Apple is famed for small steps when it comes to product evolution. Only when they were good and ready, did they release the M1 onto the world.
Any of the other ARM Arcthiectural licensees could replicate what Apple has done in terms of silicon.
BUT
As the article says, there is more to it than that. Rosetta 2 is right there making the transition easy for users. They have had macOS running on ARM for years. That has been well known.
I actually pity MS. They didn't see the writing on the wall. Their efforts with ARM CPU's have IMHO been half-hearted. W11 apparently won't run on the ARM surface device. Footgun at the ready... Aim... Shoot.
Apple has shown what can be done with care and attention AND not needing to open up the M1 to others. They don't need to open it up. All the other ARM licensees are busy trying to replicate the M1 and second-guessing where TimApple is taking Apple in the future.
Intel? WTF are you doing? Yes, you have managed to squeeze M1 levels of performance out of X86 but the power consumption is in a different league to the M1. For Servers and Desktops ok but for laptops? Forget it.
The Arm architecture is readily available to licensees. The likes of Samsung and Broadcom and others are testimony to that.
I think that is not exactly correct. Apple only uses ARM ISA (which ARM has developed for Apple specifically), but they develop the CPU themselves in house. You cannot go to ARM and say "can I have M1 design please" and then "oh and just send it to TSMC".
Samsung and Broadcom share the ISA to some extent but these are different processors.
What Apple has done is copyable by others. Aren't the likes of Google and Facebook busy trying to copy Apple right now?
You can copy an ISA, but if they have not shared the internals, which you cannot get from ARM, then good luck.
You're correct that Apple won't give others the designs for the M1, but why should they? The original point is that the ISA executed by the M1 is known, so other manufacturers have the choice to design and manufacture a chip of similar specification and therefore similar performance. Apple doesn't have to give up all their work for ARM to be open. Some work on a standard bootloader would be appreciated, but I wouldn't expect them to do that either.
"Apple only uses ARM ISA (which ARM has developed for Apple specifically),"
I'm not sure if I'm understanding this phrase correctly, but if it's what I think it is, no they didn't. ARM designed it in many steps, some of which were before Apple was using it, and they designed it for licensees including many who compete against Apple.
No, Apples processors are not “copyable”. Apple and a few other companies have unlimited licenses to use the ARM instruction set. But the chip design is 100% Apples property.
Samsung for example is free to create their own design, and sell it to PC makers if it is good, or build their own PCs. But they cannot copy Apple’s design.
PS. Apple seems to have made some changes so that sharing between processors with stricter Intel rules, so low-level code for locks etc. works correctly when translated from Intel to arm code.
"The PC was successful only because IBM opened up the specification of the architecture so that other manufacturers could also make their own PCs."
That's not how I remember it. I remember Compaq using the longest, cleanest crowbar they could find to pry open the workings of the IBM BIOS to get 100% compatibility with IBM's offerings. And ISTR there was a not-insignificant trial about it, and Compaq prevailed. THAT is what opened up the market to the clones.
This is exactly right; IBM threw lawyers at absolutely everyone making a PC clone for a long time. The breakthrough was finally achieving 100% BIOS compatibility without using any of IBM’s original code, and with the necessary legal evidence to prove that the people who wrote the new code had never laid eyes on IBM’s.
IBM’s solution to that was to double down on lawyers and create the MCA bus and the rest of the PS/2 that would be much easier to protect against clones. But the horse had already bolted.
Apple have always tried hard on architecture change knowing that it links heavily to sales. 68000 to PPC had some degree of emulation support maintained right up to OSX 10.4. PPC to Intel was a much tougher challenge on performance. OS9 to OSX had translation layers… though if you tried very early versions of OSX on the minimum hardware of the time (early IMac etc) they were basically unusable (10-min boot time). And in reverse, unofficial hacks kept os9 running on hardware very much later than it was supposed to (useful in some very specific niche applications for low latency).
M1 has really surprised because even x86 emulation is proving very fast. We got burned buying a G5 quad and how quickly support were discontinued (parts especially - those damned PSUs), so,for this transition to M1 we held off on replacing intel 2015 MacBook Pro’s right up to the latest product launch.
Workflows are such now that I can dump heavy cpu tasks off to an inexpensive ryzen/Linux setup. And where UI & audio performance is king M1 is doing us very well. I will be benchmarking throwing a big fat cpu against the dedicated hardware in the latest M1 when they arrive.
It will be interesting to see where Apple go with the next gen PowerMac.
Are they really going to produce an SoC with 2TB of Memory, or will they go
with an expandable slot design and PCIE slots for dedicated GPUs?
And what about devices that sit between the PowerMac with its stupidly high prices
and the iMac pro/Mac Mini. I'd still love to see an expandable Apple desktop, that can grow
with me/my business.
I have a couple friends that are multiple decade careers with Chipzilla. I asked them over drinks about the impact of M1 & ARM on the future of x86 architecture. Zero concern internally about it.
This industry is rife with stories of companies which ignored change in deference to the status quo. TBD whether this will be another one. Let's come back to this story in 10 years and see how it played out.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
