Apple will announce that it’s shifting from using Intel processors to its own ARM-based chips this month at WWDC 2020, Bloomberg reports. The developer conference is due to take place starting on June 22nd with an online-only format due to the COVID-19 pandemic. Bloomberg notes that the timing of the announcement could change due to the health crisis.
Rumors of Apple switching to using its own ARM-based processors in its Macs have been around for years, but a recent report from Bloomberg claimed that the shift was imminent, and that the first Mac powered by an ARM-based processor would arrive in 2021. The company reportedly has at least three ARM-based Mac processors in development based on the next iPhone’s A14 chip as part of Apple’s “Kalamata” project.
While I like ARM bringing the competition to x86, Apple is not the right company to get excited over. ARM in and of itself is already more fragmented and locked-in than x86, and Apple cranks this up a notch. A lot of people are excited over ARM Macs, but I have a feeling they may be in for a very rude awakening.
No more Steam. No more Bootcamp. No more Linux or other alternative operating systems. Far more Electron apps and shoddy iOS ports (developers barely wanted to make native x86 Mac apps, even fewer will make ARM Mac apps). No Adobe applications for the coming years. iOS-like restrictions for ARM macOS.
Rude awakening indeed.
A not-insignificant userbase of Macs rely on dual-booting (or flat out reformatting with) Windows.
It would be a monumentally stupid (or maybe carefully calculated) move to axe Windows support on Mac hardware. Running DOS/Windows software on Macs has been a thing for decades, although for a lot of that time, it was dependent on additional hardware and/or software in order to run it.
Unless Apple considers x86 compatability to be a paid-for addition, getting rid of it completely seems to be a ass-backwards move.
Judging by the install base of Virtualisation tools, and bootcamp in my company, the usage is about 5% of installations.
We have 500 Mac users and 25 of them run either Bootcamp or parallels for their development needs.
If the active user base of intel Macs is 25 Millions, (5 years at 5 Millions Macs a year) we’re looking at 1.25 million users. With a replacement rate of 20% a year Apple could lose 250k sales a year. If moving to ARM help them reach 5.25 Millions sales a year, they don’t care.
There is cheaper hardware which will run Windows, and the PC Master Race people will also trumpet how the hardware is also better. I don’t see Windows compatibility as a reason to stick with Intel and x86. It’s Apple hardware after all, and people buy it for MacOS more then anything.
I think The kick laid out his argument clearly when he said a “not-insignificant userbase … rely on dual-booting”.
You’re suggesting that just buying a very expensive mac isn’t enough, but then they have to spend money on an additional computer? That computer needs to be hooked up and have space too. I’m sure that will go over well…. Your comment makes me think you didn’t think that through.
* Virtualbox
* Parallels
* VMware Workstation
Hypervisors are a thing.
> Hypervisors are a thing.
Hardware accelerated x86 on ARM isn’t a thing though.. Good luck with performance emulating and translating every ARM op to x86.
kallisti5,
It’s not obvious to me that apple will even want to emulate x86 software on these new ARM based macs, I wouldn’t bet on it.
But since this has been done before by microsoft it’s interesting to look at some of those benchmarks:
https://www.techspot.com/review/1599-windows-on-arm-performance/page2.html
It ranges from bad to unusable, haha. If you’ve got some random legacy business application where the ability to run at all is more important than the ability to run well, then cross architectural CPU emulation is probably adequate. But resource-heavy audio/video/graphics applications are unlikely to be anywhere close to satisfactory without a native version.
If you really need Windows computer what is the point of buying the Mac?
You can buy a NUC on a pocket money left after buying the Mac and duct-tape it to the other side of the monitor. It takes no space.
What is the problem with 2 computers?
Tell that to the hundreds of universities and colleges out there that dual-boot Mac Pros for productivity software, such as xCode and Visual Studio, or the multitudes of image editing and media production software, that can only run on one OS or the other.
We’re not talking about Joe User at home, sat on the sofa browsing Reddit. We’re talking machines used for productivity or education purposes, bought en-masse by a large organisation, purely to run productivity and development software on.
Joe User will either buy a Windows machine, or a Mac and an x86 emulation package when it comes to running Windows software, or maybe even try to get it running under WINE. Most organisations wouldn’t want to take that route, but would rather dual-boot than mess with emulation/virtualisation
Also what about homebrew with all the unix apps? Will they even compile ?
They should. Linux and FreeBSD have been on ARM64 for a while, and most stuff has been ported.
To have a look at arm64 (aarch64 to be more specific) stuff I just bought Pinebook Pro and that is supergood hardware for the price. It comes with preinstalled Manjaro and some of the Linux software that I’m used to on amd64 is not available for arm64. And since there is no arm64 port of Electron, the ones that use Electron will not be available in the near future.
Of course if source code is available one can compile them, but the complexity is probably out of reach for most users (including myself).
My remark about Electron on arm64 Linux was wrong. It was not listed here https://www.electronjs.org/docs/tutorial/support but is available under downloads.
I used to work in an Apple reseller from the ppc to Intel transition. Switching to Intel made a huge difference in sales, and we sold a lot of copies of windows.
This also returns things to the bad old days where you are limited in hardware compatibility.
Apple as late has swapped roles with Microsoft by doing stuff like using closed standards like Metal instead of Vulkan. Maybe Steve did do more than I thought and in the moderate term Apple might alienate users
Does it? ARM can support PCIe, and it’s not like Apple has the greatest expandability or the widest hardware support right now.
Hello,
ARM Macs will almost probably (well, 100% certain almost) will not follow the SBSA* specification, however PCIe and the set of devices that you’d expect will be provided. That won’t be an issue for bringing up Linux which is already adaptable for platforms that are the Wild West in that regard. (including the iPhone)
However, this will make booting AArch64 Windows on the bare metal not a very nice effort to attempt.
* SBSA is a standard firmware and exposed set of devices specification that ensure that generic software images would be installable, even for hardware that did not exist yet when the image was built (UEFI + ACPI, DeviceTree isn’t adaptable enough for such a use case)
It will limit software compatibility on the basis of the fact Windows on ARM is dead (bye bye we never missed ya) and hence any attempt to run Windows software on ARM involves some slow emulation (instead of hardware-assisted virtualization with VT-x).
Then there is the question of the huge backlog of x86 MacOS X software, such as the last non-subscription versions of Adobe software.
I am really hoping they will provide some hardware-assisted binary translation in their custom SoC, but considering how Apple dropped 32-bit apps from Catalina without any warning or migration plan, I wouldn’t bet on it.
Hello person from 1996 and welcome to 2020, just a couple of tips:
1) There is no such thing as “DOS/Windows” anymore. All DOS-based Windows OSes are officially out of support and won’t even run properly on modern computers, instead all Windows OSes are NT-based now. If you want to run DOS software on Windows you ‘ll need a full VM (check out DOSBox, which is also available for Macs)
2) By the way, Macs now use a Unix-based OS as their “Mac OS”
3) There is free-flowing HD pr0n out there from sites that won’t trash your PC or heckle you to download malware to get to it.
4) AMD has the node process advantage over Intel now. Shocking I know.
Can’t reply on your 4:38 comment (there is no Reply button).
the fact Windows on ARM is dead (bye bye we never missed ya) and hence any attempt to run Windows software on ARM involves some slow emulation
Windows on ARM is awesome. Also, “slow emulation” is a bullshit. I use Visual Studio 2019 under emulation and it is fine. I even use emulated browser because it is just blazingly fast and I don’t want to bother installing native one
Well, Steve Jobs once said that the Mac will remain to be open (ok – he also said that Mac OS X Server will be continued…).
If Apple is going to close the Mac (to make more revenue from their “curated” App Store) they will loose lots of market share.
macOS is a good choice because it is open and has a working UNIX layer beneath.
markus,
Can you cite the specific quotes you are referring to?
One would hope so, just as one would hope that the executive branch would never be run by an authoritarian, but it’s dangerous to assume it.
To be clear, I have no knowledge of either company’s inside plans, but if they’re hell bent on normalizing walled gardens on PCs, I think they could succeed by doing it together.
You say they’d loose market share, but lets consider where they would go. Some may jump ship to linux, but I think more people threaten to switch to linux than actually do. Most people using macs and mac software are not prepared to use linux. Many mac users are in this situation…
https://community.adobe.com/t5/get-started/adobe-cc-on-ubuntu-linux/td-p/9834114
Some may jump ship to windows, but I don’t think that microsoft is a reliable guardian for open computing. Microsoft are jealous of apple’s walled garden on IOS and they’ve been bouncing around in will-they-won’t-they territory with regards to their own walled garden OS for PC consumers. They’ve had two trial runs with windows rt and windows 10 s-mode, and a new attempt with windows 10 x. They’re sort of chomping at the bit to get it done.
It would be mutually beneficial for both companies to make a walled garden transition around the same time to maximize compliance and minimize people jumping ship. It wouldn’t be a popular move for users like us, but microsoft is no stranger to following through on it’s unpopular agendas quite forcefully despite tons of negative press, People scream and shout about it for a year or so, but once it’s done and the dust settles, the public at large generally accepts it as the new normal.
It would be wise for both companies to offer unrestricted / “pro” versions for legacy x86 mac/windows applications, but they can make it less accessible to normal consumers (most likely through pricing). People complaining about this will be labeled as too cheap to buy the version that suits their needs, meanwhile consumers will be steered into the latest and greatest walled garden versions. Of course the end goal is capturing a critical mass of consumers in the walled garden. It doesn’t matter to microsoft and apple that the developers have unrestricted versions because the developers will still be forced to pay apple and microsoft fees in order to reach their own customers.
Once again, I offer nothing more than speculating here, but when you follow the money it’s no wonder that walled gardens are so desirable for companies like apple and microsoft. Look at IOS numbers…
There’s the 30% sales tax on developer sales:
https://sensortower.com/blog/app-revenue-and-downloads-2017
That’s $11.5B/yr in 2017.
There’s the developer fees:
https://techcrunch.com/2018/06/04/app-store-hits-20m-registered-developers-at-100b-in-revenues-500m-visitors-per-week/
20M * $100 = $2B/yr
On top of that if developers need more expensive unrestricted PCs to develop on, that could increase profits as well.
We can question the viability of such a plan (what would stop it from working?), however I think there’s no question that both companies would like to see it happen.
Mac OS X is not open in any sense of the word (open-source or open to all hardware), it’s “open” to all ISVs who can compile a dmg but that’s not a high bar to clear, and the “UNIX layer beneath” is useless as far as openness is concerned. TiVo has a “UNIX layer beneath”. The PS3 has a “UNIX layer beneath”. But both of these platforms are heavily locked and only run software the vendor provides (officially). Apple could lock dmg installation from outside the App Store in Mac OS X and keep their “Unix layer beneath”-ness intact. Other than giving graybeards a fuzzy feeling when they see that filesystem root, “UNIX layer beneath” doesn’t say that much when it comes to platforms, freedom from App Stores and compatibility, which is what us regular folk care about.
So, no, Mac OS X is NOT a good choice. You are buying into a system without a significant business customer base and a user base consisted mainly of clueless hipsters and annoying fanatics who will buy anything from that particular OS vendor, even Catalina. Avoid.
Prejudiced not much.
You are not “regular folk”. You are an edge case. Real world need you to be living in some…
You seem to argue ARM fragmentation like it’s a bad thing. I see it as interoperability and competition.
Apple have previously stood alone outside the x86 hedgemony before, so are fully aware of the implications of that decision.
In terms of Bootcamp, few if any use it in production environments, certainly at enterprise level, as its unsupported.
Linux… dare you to install any fully functional linux on a touch bar mac. It’s a mess of half support and broken features.
ARM fragmentation means that there isn’t interoperability, though.
It means you have a kernel build for each board, instead of one kernel that works on, say, all 64-bit ARM platforms. And, if developers lose interest in your board, you stop getting kernels.
macOS apps only run on macOS – there is already no interoperability. If you want your apps to run on macOS, you still have to port them. What’s the difference on ARM? Same steps required.
I have no doubt this will effect platforms like Steam (especially their back catalog), but honestly, who games on a mac? And older games can run in emulation. But Apple has gaming on macOS as obnoxious as they can for years. Still – if Steam does support ARM on macOS, others likely will too. Most games are built on engines – and those tend to support multiple platforms with a checkbox. Unity, Game Maker Studio, Unreal – they all already support all the biggest OS and CPU platforms. They’ll do the work, add the checkbox, and it’ll be fine.
Also, Steve Jobs was right about Adobe. They are lazy. But also, they do thing stupidly. Like, why do they have their own whole bloated and slow cross platform GUI toolkit, for just 2 platforms… They’ll show up eventually though, just like they did for x86 – that’s not a real concern.
Which is one of the great mysteries of the FOSS universe. People have spent countless hours writing a Windows emulation layer, but they haven’t bothered with MacOS even thought there is much more commonality and a better base to start from.
It’s almost as if they like being abused. The pathologies of FOSS developers would probably be an interesting paper for a psychology or sociology post-doc.
It seems there is projects to allow macOS binaries to run on Linux: https://en.wikipedia.org/wiki/Darling_(software)
Also, don’t forget that both Darwin (the macOS/iOS kernel) and NextSTEP/OpenSTEP are both available open-source as well. A large and passionate team, using the 3 projects mentioned, could likely build a pretty decent open-source macOS compatible OS.
@The123King
Which was started fairly recently and well after Wine.
My point exactly. OpenSTEP, Darwin, GNUSTEP, GCC Obj-C, and Swift are all there, but people spend time on emulating the Windows API. Madness.
Flatland_Spider,
If it makes you feel any better, it has less to do with merit, and more to do with popularity… like it or not windows is more popular, which is why more effort goes into emulating it.
“Which is one of the great mysteries of the FOSS universe. People have spent countless hours writing a Windows emulation layer, but they haven’t bothered with MacOS”
I find it kind of odd that you see anything mysterious about this. It’s quite simple and rather obvious, applications!
Large parts of the applications available on MacOS also have Windows versions. For those few MacOS applications without a Windows version, most have Windows alternatives delivering the same functionality. In some cases the MacOS software may be better, but seldom significantly better. And in many cases, the opposite is true. In addition there are the hundreds or thousands of Windows application that does not have an MacOS version or equivalent. Even if it require more work, a windows emulating layer gives a significantly higher ROI.
I run games on mac, but there are limited. ARM would make this far worse. Games like Portal, etc still run very well. I actually use a hackintosh so I can reboot to windows obviously but since i do all my work on macos I’ll play games on that when I can to avoid rebooting. I’m sure there are a ton of apps that rely on x86 too. What about homebrew with all the unix apps? Will they even compile ?
Yes, Apple have stood alone with their processor before.
Yes, they have managed the transition from one architecture to another before.
But that is only part of the story. Towards the end of the PowerPC era, the Apple desktops really weren’t that competitive (at a time when more people were still relying on desktops over laptops). What they did have going for them is that PowerPC was, then, much more efficient for laptop use. And the OS was more solid.
The transition was a pain for users. Yes, old apps worked. Slowly. Realistically, users needed to move over to x86 apps fairly quickly, often absorbing more costs whilst doing so.
Since then, x86 has been transformed into having much more efficient chips. Windows 10 is in a lot better place than Windows ME/XP was back then – especially with the latest subsystem for Linux. And Linux itself is in a better place if you care more about having a *nix underpinning than commercial app availability.
And whilst Bootcamp isn’t hugely used, virtual machines running Windows and/or Linux are far more common.
Does ARM really offer enough tangible benefit over x86 to make ARM based desk/laptops a compelling alternative?
It seems more likely that the only users will end up being people who use Apple specific software (e.g. Logic), and developers – particularly those creating iOS apps.
Unless Apple can demonstrate something tangible in providing more power / battery life and reducing costs, anyone faced with the costs of not just upgrading hardware but software too, would be wise to consider moving to a platform that doesn’t have a history of doing this to users, and at risk of being out on a limb again with software companies abandoning it because the market is too small.
More because of IBM not caring about Power in anything other then servers then any fault of Apple’s, which is the same story over and over again for all of the processors Apple has used.
For Apple, probably.
ARM is designed for vertically integrated companies, like Apple, and Apple already has a design team in house. It makes sense to leverage their in house talent to get chips designed exactly to their specs. They’re already doing that with iPhones and iPads, so they might as well add Macs as well.
Windows is NOT a great operating system, even today (sure, compared to ME, maybe…). Their GUI is buggy, back compat decisions questionable, and they routinely have to run everything as administrator. There just is no sane security subsystem, or they haven’t been aggressive enough getting Windows developers to use a sane security system. It’s even routine for some apps (particularly games, but also some expensive “pro” software) to install literal root kits as DRM on Windows systems. Security is SO bad, that you HAVE to run anti-virus, and it does cost you performance. Windows is not “good”. It’s kind of awful.
One benefit I hope, but don’t expect, Apple to bring from ARM chips, is a reduced cost. Mac hardware has gotten too expensive. 10 years ago their hardware was of a quality and provided a set of specs that was actually competitive. Now the price is just super bloated compared with similar quality PC hardware, with better specs. That has to change if I’m ever going to consider buying another mac (which is increasingly unlikely given how much I currently enjoy Pop! OS)
Apple were not exactly successful back when they used PowerPC. It nearly ended in bankruptcy.
It is however interesting nonetheless, and I welcome a viable alternative to x86.
They were reliant on a 2nd party partner to deliver PowerPC chips. Motorola and IBM. When those companies decided to end their investment in the PowerPC platform, Apple had to switch to x86. but that put them in the same boat – they were reliant on their chip partners in Intel and on either nVidia or AMD for GPUs. And it has shown in the cost structure of a mac.
This time is nothing like the previous times – ARM is not a chip vendor, but an IP vendor. Apple will be designing the chips, and then leveraging competing production companies to get the actual hardware built – exactly the way Tim Cooks likes it.
They very much were. When they moved from 68k to PPC they were absolutely at the forefront.
Remember the G4 when it was released in the Powerbook meant it was legally classified as a weapon it was so powerful (for the time, my watch probably outputs a gigaflop these days).
Apples failure at the time was its licensing model and OS, not the hardware. As was shown by when SJ returned and the company returned to massibe profits all while running PPC.
Adurbe,
Apple’s claims were rather dubious and misleading a the time though, it was never “classified as a weapon”, rather the truth was that it was just subject to export restrictions. The US government didn’t want enemies having supercomputers, and for a time in the late 90s the government definition of supercomputer was low enough that ordinary computers managed to fit the government’s definition. CPUs by amd and intel were also subject to the same export restrictions. The main reason apple was associated with this is because their marketing team went out and turned it into a PR stunt.
https://www.techjunkie.com/apples-1999-power-mac-g4-really-classified-weapon/
The higher costs of PPC chips was justifiable in the 90s when they had a performance advantage, but as time wore on, cheap commodity x86 CPUs would surpass PPC and apple consumers would have to ask why they’re paying a premium for CPUs that perform worse. Ultimately I don’t think apple could have predicted the downfall of PPC, but I think it was a bit of an embarrassment for apple to fall back on the intel CPUs used by windows – at least that’s the way it looked to me. Everyone on team mac really wanted PPC to win over chipzilla, it just wasn’t happening.
I’m a little confused as to why Apple would not do this. Apple’s main revenue is from the iOS platform which is well supported by all the major developers and runs on ARM devices (iPhone and iPads). Locking Macs into the same walled garden and having them run iOS apps is surely the obvious next move, no? MacOS is a dying mess and needs something to bring it back from the brink of death.
>No more Steam. No more Bootcamp.
Most Steam games I’ve bought recently warn they’re not compatible with Catalina and the Epic Store seems to have no Mac games at all other than Fortnite! Bootcamp has never been great IMO, and it’s only getting worse.
Parry,
Agree, this it would be logical of them to do just that.
This is their own fault though, if it’s dying it’s because they let it atrophy when IOS was the new baby and took the limelight. While apple went on to release new mac products you could tell they weren’t deeply vested in them and it just continued to sink. Many people like to say steve jobs would have done better, but the truth is the mac neglect started on his watch as apple circled the wagons around the iphone and this just continued through the progression of years when Jobs was gone.
Business-wise it’s hard to argue that going all in with IOS wasn’t the right choice for apple financially speaking, but it would have been interesting if macos could have been transferred to a company that was more vested in its success…it could have been a cutting edge unix platform leading the industry with innovation instead of stagnating under apple. Oh well, it wasn’t to be.
There’s no reason for this other than apple not supporting it well. I suspect it was always more of a “checkbox item” than something they wanted users to do.
Yeah. At the time, dual booting was still the best way to run a second OS. Hypervisors weren’t as widespread, they relied on emulation, and personal computers didn’t have the resource density to truly make them feasible for running something like Windows.
Then all that changed and dual booting faded out.
Microsoft are trying to kill office now as it is, with the push to O365
O365 and Office are symbiotic. They feed into each other. O365 supports Office the best, and Office supports O365 the best. After that, people might as well buy into Azure.
This pretty much kills whatever use remains for “creative professionals,” but only Microsoft not making office for it can kill this development now. Which is very likely if Apple pulls a Windows 8 and tries to force everyone onto their app store.
As long as users have a web browser they will have O365. I cant see apple redeigining pages though…
Exactly, due to the availability of tools on windows and years with underpowered hardware to premium prices, fewer of those those “creative professionals” remain. The only large group of professionals with need for Apple hardware are the iOS app developers. But with iOS lesser market share and the trend with cross platform apps, this is not a growing market. With developers primarily working on other platforms, the use of aging MACs are no problem and less wear also makes the need for replacements declining.
ARM on Mac… sounds just ghastly! Utterly horrible. From now on I’m going to refer to Apple ARM based computers as “The Sadness” LOL!! On a more serious note, I believe Thom said it best:
No more Steam. No more Bootcamp. No more Linux or other alternative operating systems. Far more Electron apps and shoddy iOS ports (developers barely wanted to make native x86 Mac apps, even fewer will make ARM Mac apps). No Adobe applications for the coming years. iOS-like restrictions for ARM macOS.
Rude awakening indeed.
Light passive-cooled laptops with powerful ARM cores. Mmmm… Can’t wait.
even fewer will make ARM Mac apps
That makes no sense. Apps would be easier to port and create. macOS is just a mess.
Adobe has already ported photoshop to ARM (not that I see any reason for average person to install Photoshop).
iOS-like restrictions for ARM macOS.
Lets wait and see.
The consensus is that Steam has already left, so that doesn’t matter.
Hypervisors like Virtualbox, Parallels, and VMware Workstation have largely replaced dual booting.
Linux is a hot mess on Apple hardware, and there are cheaper alternatives from Dell, Lenovo, etc. which support Linux and alternative OSs much better. Apple hardware support is largely the efforts of small contingents within a community and most people will be pointed towards refurbished x86 equipment.
So far Apple hasn’t lost anything they haven’t already lost.
Electron apps are blights on the technology landscape, and it isn’t limited to Apple. Electron effects all platforms unfortunately. It means Linux has gotten more support, but it means I run like 10 versions of Chromium and need ~128GB of RAM. (Not really. I can get buy with 32GB. Technology today is simultaneously amazing and blighted.)
It’s not hard to understand why Electron apps are popular. You do the work, and any platform Chromium has been ported to is supported. It’s a magic bullet, once the non-trivial work of porting Chromium is done.
I’m interested in seeing how this pans out. I haven’t seen any yet, but I don’t see how they could be bad. I’m one who prefers small windows to running everything fullscreen, and many iOS apps should work fine in a desktop setting.
Note that Windows 10 runs on Arm64; many Linux distributions do too; common browsers are natively compiled for Windows on Arm64, as is electron; Windows on Arm64 executes x86 code through Rosetta-style emulation (not a heavyweight VM), etc.
I’m not saying that the Arm64 software ecosystem is mature or complete, but it does exist, and if a large vendor chooses to adopt it, it will grow that ecosystem.
Seems daft to me though kinda understand Apple wants to build that wall higher. Doubtless it will integrate well with that phone and tablet and suppose those happy with that ecosystem might want an arm laptop though not sure why you’d bother. Anybody else won’t be impressed especially at Apple pricing, at least till arm chips get much more powerful.
Loath to say it’ll be a disaster as Apple seem to get away with any crap, however consumer unfrendly, they come up with . Objectivly its a dumb move and sales should dip dramaticaly but less is more , less for you and more profit for them .
“at least till arm chips get much more powerful”
You clearly haven’t been paying attention to the benchmarks. Apple’s ARM chips already in production are more powerful than the laptop processors in the latest generation of laptops. Thermal limiting takes its toll. Server-side, Amazon’s ARM platform (Graviton-2) is more powerful than their Xeon ones, so there’s that. Up at the big end of town Fujitsu A64FX parts have been seen to actually exist.
I’m looking forward to an Arm-based mac as a performance upgrade, as I expect that’s what they will be. I don’t expect to lose any functionality on the transition, but I neither game nor Windows on my mac systems.
Not that I have much code from the App Store on my system, but remember that Apple has been insisting on not just 64-bit but LLVM-bitcode versions for some time, so you can reasonably assume that anything currently available on the App Store, and anything open source will run natively at launch, or soon after.
areilly,
Can you cite the sources of your data? I realize apple has made great strides with ARM, but it’s a lot more interesting to discuss these things over real data & benchmarks. When it comes to performance, there are often claims that don’t pan out under closer inspection because the data was misinterpreted or it wasn’t a fair comparison, etc. Linking to the data also helps keep us informed!
This Ars-Technica review of the new 13-inch MacBook Pro is fairly recent, and has some results graphs that compare (among others) that computer against the iPhone 11 Pro and the iPad Pro. https://arstechnica.com/gadgets/2020/05/2020-13-inch-macbook-pro-review-the-standard-macos-workhorse/ The iPadPro is running cores a generation (year) behind the ones in the iPhone, but also more of them and with more thermal headroom, one assumes. Those results are wall-clock single-thread results, so the fact that the i5 part clocks twice as fast is in the mix.
Server-side, the register has some figures here: https://www.theregister.com/2020/05/12/aws_graviton2_debuts_in_ec2m6g_instances/
Anandtech have had some benchmarks of the Amazon Graviton-2 cores: https://www.anandtech.com/show/15578/cloud-clash-amazon-graviton2-arm-against-intel-and-amd
And don’t forget that Arm have just announced the Cortex-X1, (https://www.anandtech.com/show/15813/arm-cortex-a78-cortex-x1-cpu-ip-diverging/4) which has the floating point hardware to go head-to-head with AVX2 on a cycle-per-cycle basis, and which the author of that article guesses might not quite beat a _desktop_ Skylake or Zen2, but is close enough to be interesting. That’s Arm trying to catch up with Apple…
areilly,
Thanks for posting those links, it’s good to talk specifics. Once again I’ll agree with you that the iphone 11 pro does quite well, however I think saying “Apple’s ARM chips already in production are more powerful than the laptop processors in the latest generation of laptops” overstates it. Realistically comparing it to the 13″ MBP is kind of cherry picking a slower laptop.
These are from the arstechnica article you linked to…
I’m not really a fan of allowing MBP benchmarks to represent x86 chips generally because of just how notorious MBPs are for bad performance. Anyways, the iphone 11pro does better than both of the MBPs on single threaded benchmarks, but it is significantly worse on multithreaded tests and it’s worse than the dell XPS laptop across the board. The arstechnica article doesn’t show higher end laptops, but they’re available.
The data center applications interest me a great deal. Unfortunately ARM servers are still mostly in the domain of custom integrators with large budgets, but assuming that they eventually come to commodity channels, I’ll probably be buying some myself As far as I know, amazon uses intel chips to power ECs, ARM chips should fair pretty well given that intel is still stuck on older process technology. Although in this case I think AMD is really the one to beat and not intel.
Alfman,
Note that the iPhone 11 Pro’s A13 processor has two fast cores and four low power cores. If the multi threaded result is considered as a two core result, or even a three core result, it doesn’t look uncompetitive at all.
Also, I’ll bet that the processor when placed in an iPhone ends up being thermally throttled. The exact same design, when combined with current laptop cooling, may be able to perform better.
The open question to me is not just performance but also battery life. Apple have been implicitly betting for a long time that users value battery life and weight more than raw performance, which is why they end up underclocking CPUs. I’ve been expecting them to move to ARM for around a decade because it should result in a lighter system that runs for longer, even if it’s a (small) performance regression. Current data suggests that they don’t need to regress performance but may still be able to decrease weight and increase battery life.
malxau,
Yes, this came up last time too, but regardless of why the A13 processor performed poorly in multithreaded test, it still performed poorly. Maybe there will be a new version of the processor that performs well using 6 or more cores, and if so I look forward to seeing those results. However for the time being the benchmarks that areilly linked to really don’t make the case that this is beating high end laptops. This doesn’t mean it can’t or won’t in the future, but my normal stance on performance claims is that I’m not going to take anyone’s word for it until 3rd party reviews come in.
I don’t think there’s a reason to question this, ARM has the lead there already
It’s not just apple, most portable electronics manufacturers have favored ARM for a long time to maximize battery life.
I can certainly see it as a decent midrange laptop. However I would argue that the A13 performance as it exists today is not ideally suited for a professional working with intensive compilation, video and graphics, the performance drops are too significant. Maybe things will be better when their new ARM laptops come to market, I’m not expecting any surprises though. We will have to wait and see what the rest of 2020 brings. I am more curious about pricing, I kind of expect these to be priced lower to fill in the gaps in their x86 lineup.
Alfman,
Fair enough, that’s your choice. I’d be very surprised if Apple announce ARM laptops with two high performance cores. From an engineering point of view, adding cores is relatively straightforward. So I’m extrapolating that a four core version of what they have today is quite competitive, and we can certainly reserve judgement on that until it arrives, if ever.
I guess I’m the only one who doesn’t think the sky is falling until I have proof. There’s no reason to assume anything will change regarding supported apps, my guess is that it will all simply be emulated until the recompile is ready. In fact this would be a good explanation as to why 32-bit support was dropped in Catalina – probably the emulation engine is limited to 64-bit.
The only downside that I agree is likely to happen is dropping Bootcamp. But I never used Bootcamp anyway since Parallels and Wine have always been more convenient for my use cases. I’m guessing it’s just a matter of time before Parallels finds a way to run the ARM version of Windows in a VM, probably with no support for x86 emulation within the VM, but by that time most Windows software will hopefully be available for ARM64 as well and then there’ll really be almost nothing to complain about.
We’ll all find out soon enough….
Hello,
The x86 (and the coming x86_64 binary translation layer) on AArch64 Windows do not rely and do not use any special hardware support outside of the Arm specification. XTAJIT is all implemented in user-mode, don’t worry about that.
You might then ask why Windows on Arm was not demoed to run on current iDevices:
On Apple A9 onwards, Apple doesn’t support 4KB pages in the MMU, using 16KB as the smallest page size. However, compatibility with Android (notably does _not_ include GNU/Linux and Musl based distributions or BSDs, which run just fine) and Windows require 4KB page support. Such support is not required but is very desirable for x86_64 on AArch64 binary translation too, so maybe it’ll be there on the A14 generation.
The Apple SoCs also use a non-standard interrupt controller, but that can be papered over quite easily through Hypervisor.framework, which is provided on AArch64 macOS.
Hopefully that gives you some more answers,
Thanks, that was very informative. I guess it’ll be up to Apple whether they want to support 4kb pages for the sake of x86_64 and alt-OS compat.. if they were to leave it out I guess Android development on Mac-on-ARM would be a no-go for the time being?
Hello,
Android was ported to Apple devices with 16KB pages (done as a part of Project Sandcastle). Though, on that, you have to re-link your applications with restricting the linker to make 16KB aligned sections.
As such, for your apps it isn’t much of an issue, but if you want to run regular ones that aren’t Java only and use the Android toolchain… it won’t work.
It’s interesting how quickly and willingly consumers surrender their independence just at the hint of a promise of a shiny new something or other. The price is not even known yet, or the cost!
I’m curious what independence you think that will be lost? Do you expect that the existing ability to develop software or load from other sources will change? That the internet will become unavailable?
I’m expecting the new shiny to behave exactly the same as the current systems, but faster and cheaper. PC processors have always had enormous profit margins compared to Arm mobile SoCs, and Apple don’t even have to pay a chip company, just their own engineers and the fabricators.
I guess we’ll see.
Isn’t it the opposite of transparency?
Apple’s a good citizen, if I film myself murdering somebody they won’t unlock my phone for the bad policeman. They always do the right thing, they are not motivated by profit.
Cynical as I may be, I’m curious to understand how you think a proprietary OS built on and for proprietary hardware preserves your freedom?
When they own and control all the hardware, what happens to those developer fees, a big fat bonus? I must check in on my old IBM, Silicon Graphics and Sun/Oracle friends.
It feels like just one corporate psychopath in the right position on the corporate ladder is all it will take for this to go horribly wrong. Maybe I’ve got too many feels!
Why? Windows runs just fine on ARM64 and also includes an x86 emulator for the legacy apps. Most x86 games have been ported to iOS so a port to macOS/arm64 shouldn’t be difficult.
Why again? Linux will run just fine on macs. The question is if Apple will create a KMS driver for Linux or provide documentation for their own GPU. Otherwise I don’t see any reason for this assumption.
Why again?
Adobe, due to its huge mac userbase will bring ARM64 apps from day one of ARM64 macOS. They did that with the transition from PPC to x86 and at the m68k to ppc transition.
Current macs are as restricted as iOS if you look carefully. They allow the restrictions to be bypassed by a firmware flag or by a plist setting. Othewise, you can’t install Linux on the mac.
Stop spreading FUD. Let’s wait for the unveiling.
I guess I’m the only one who doesn’t think the sky is falling until I have proof.
We are also from development company.
https://sysbunny.com/android-application-development.php
This smells of carefully crafted spam.
Whatever the shortcomings of this move will be overcome by a shinier gold coat and a bigger, pulsating, Apple logo on the lid to let people around know you’re a Mac User. And a higher price tag to empower you to prove you can.
They will probably be very well made, extremely thin, have long battery life, wake up instantly weeks after you used them last, and offer a touch-strip option at twice the price of a full 4K touchscreen on a Windows PC. They will have one port and charge $200 for an external hub and $50 for each cable. They will not be expandable. And they will have a world-facing camera to let you use them as a camera if you feel the 12″ iPad is not large enough. And they will not honor the compulsory 2-year guarantee in Europe.
Outside Apple’s own software, as long as they run Adobe stuff and some popular music and video softwre, they are good to go.
“They will probably be very well made”
Why would they start making them reliable again when the last 5 years of their laptops have been a garbage fire in that category? Rossmann on youtube has almost daily videos of repairing their engineering failures, but it feels like Apple decided Apple Care and getting people to replace their whole computer by only doing motherboard swaps was more profitable. Most Apple users only care if their own laptop bites the dust, then blame themselves for spilling a bit of water on it. Of course it’s 2020 and I can watch old Lenovo videos of Thinkpads surviving pitchers of water being poured on them on YouTube.
Hmm – does this mean that we’ll be able to – a la Einstein – directly compile the NewtonOS for the Mac? How much different is the current ARM assembly (probably especially interrupt and memory-fault handling) from the “StrongARM era”?