Mac With Apple-Designed Arm Processor Coming in First Half of 2021

[hans1972]

Not to mention any number of tools that rely on hardware features, and then those that rely on hardware acceleration of features.

I think very few applications interacts directly with the hardware. They do it through OS APIs and drivers. Apple just needs another way to simulate the hardware feature either with another hardware feature or software simulation.

It doesn't matter if the implementation is slower, because I believe this will be for low end users in the first iteration.

They just need to make sure that the performance is acceptable for regular users not using complex software or special hardware.

 

[bsolar]

Let’s get it right. That was actually settled out of court. It looked like Intel would lose, but the reason wasn’t that a license was not needed. The reason was that Cyrix was using a fab that DID have a license. So the question was whether you can infringe a patent by selling a product that was made in a fab that did have a license. The answer to that was going to be “no.”

That isn’t the situation here.

I stand corrected about the Cyrix case. I'm still unsure whether Intel protections apply to emulation layers though. The article mentioned above with Nvidia seems to hint it should not apply. The articles about Qualcomm are a couple years old and AFAIK nothing came out of these "threats", some of the article questioning Intel's legal standing to prohibit x86 emulation.

It seems Microsoft is preparing an amd64 compatibility layer for Windows on ARM beside the already existing 32 one: it would be interesting to see whether it actually happens and whether it happens with Intel (or AMD?) cross-licensing or not and how Intel would react in the latter case.

 

[cmaier]

I stand corrected about the Cyrix case. I'm still unsure whether Intel protections apply to emulation layers though. The article mentioned above with Nvidia seems to hint it should not apply. The articles about Qualcomm are a couple years old and AFAIK nothing came out of these "threats", some of the article questioning Intel's legal standing to prohibit x86 emulation.

It seems Microsoft is preparing an amd64 compatibility layer for Windows on ARM: it would be interesting to see whether it actually happens and whether it happens with Intel cross-licensing or not and how Intel would react in the latter case.

Microsoft can likely do whatever it wants, both because it may have a license and because it is too powerful. Intel depends on Microsoft. AMD wouldn’t be alive today if Microsoft didn’t support and assist with AMD64 (my biggest fear at the time was that we’d design this great chip and have no software to run on it).

As for whether you need a license to create an emulator or not, it depends on the specific patents that AMD and Intel own, and what the claims cover. I suspect the answer is, yes, of the thousand of patents owned by these companies, at least some claims cover a machine running an emulator, but I haven’t done even a cursory analysis of their entire patent portfolios.

 

[konqerror]

It seems Microsoft is preparing an amd64 compatibility layer for Windows on ARM: it would be interesting to see whether it actually happens and whether it happens with Intel cross-licensing or not and how Intel would react in the latter case.

It's been out for nearly two years now. Go buy a, say, Lenovo Yoga C630 or a Surface Pro X, you can run x86 apps on it.

ITT: people coming up with hypothetical situations over something Microsoft has been shipping for a year.

 

[bsolar]

It's been out for nearly two years now. Go buy a, say, Lenovo Yoga C630 or a Surface Pro X, you can run x86 apps on it.

ITT: people coming up with hypothetical situations over something Microsoft has been shipping for a year.

AFAIK the already existing emulator/compatibility layer is only x86-32, not amd64. I think the patents for the former are likely expired or nearly so, for the latter there are some likely still valid.

 

[oneMadRssn]

Actually, most laptops and desktops likely already have more ARM processors in them than x86.

What? Other than Apple's T-series, what are referring to specifically? Which ICs in laptops and desktops are ARM-based processors?

I'm not being pedantic, I actually want to know because that would be very interesting to me.

 

[MikeZTM]

Again, you're missing the point - I have to assume deliberately.

At the time of Apple's transition to Intel, there was essentially no commonly used software that ran on PPC other than that which targeted Mac OS.

In the ~3 versions/5 years of macOS supporting Rosetta, it was reasonable to expect Mac app developers to rebuild their apps for Intel - there was certainly no expectation of new, ongoing development of PPC-only apps that would not be available for Intel at all.

In this scenario, that is not the case at all. Unless you're expecting the majority of the computing industry to drop x86 by the time Apple decides to drop a hypothetical "Rosetta 2.0"?

PowerPC "PC" only software? They still exist just not as popular as it was.
PowerPC software? Just remind you Xbox360 and NGC/Wii/WiiU was powered by PowerPC.

Right now, mobile market is several times larger than PC market. A brand dominating PC market isn't as important as it was.

 

[rp2011]

Apple needs to produce <u>HUNDREDS of millions</u> of CPU’s every year for it’s iOS devices.

Imagine if it had to pay and depend on Intel to produce them?

Chips designed for the all the various Intel interests and bogged down to support so many of their legacy products?
Yikes!

Whereas Apple only needs a tiny fraction of CPU’s for the failing x86 consumer market. If/when Apple’s CPU’s are close in parity to cut Intel out of the equation, it will. Absolutely.

 

[cmaier]

What? Other than Apple's T-series, what are referring to specifically? Which ICs in laptops and desktops are ARM-based processors?

I'm not being pedantic, I actually want to know because that would be very interesting to me.

Many of the random chips in any computer have arm cores. Power management chips, system control chips, etc.

 

[MikeZTM]

I use one macOS VM, to run one macOS only App, specifically because the newer versions of this app don't support the self-hosted mode they supported formerly, because said app developer said "**** you" to those of us who bought a license to run it self-hosted.
My point about that wasn't that it's a huge use-case, it was that you have literally no ****ing idea what you're talking about claiming that an x86 VM can't help alleviate the 32bit issue for "legacy" Mac apps.



Somehow, you've still missed the point completely. 99% of my use for x86 VMs is not "legacy" software. I use a metric **** ton of VMs to run simulations of production x86 Linux environments, locally. This is not a rare thing for developers using Macs, and while some would get by with an ARM Linux VM, plenty wouldn't be able to.

"just keep using your old computer" isn't really a viable solution if Apple drops all x86 support.

What if your production server becomes ARM servers?

And just keep in mind how those developers doing their work for those "mini computers" on their Intel workstation.

You have ssh.

If you have to rely on not supported features for whatever reason -- be it a money grab action from developer or not -- you are in a died end. Eventually you have to move on.

And if you are not confident enough here's an example:

How about supporting ARM64. · Issue #2053 · dotnet/winforms

Windows 10 supports ARM64 processors. And, Windows 10 on ARM64 can run x86 applications. But, performance is significantly low because of emulation. And, it cannot run native AMD64 applications. So...

github.com

This is Microsoft .net core winforms open source repo(yes that Winforms from 90s).
.Net Core doesn't support Windows 10 ARM64 yet and only support Windows 10 x86/x64 and Linux ARM64.
It just took them a week to make demo app run on Windows ARM64.

It should be much easier for other app to recompile than this huge complex .net runtime.

 

[konqerror]

AFAIK the already existing emulator/compatibility layer is only x86-32, not amd64. I think the patents for the former are likely expired or nearly so, for the latter there are some likely still valid.

The AMD64 spec was released in 2000, so any remaining patents should be close to expiring.

 

[bsolar]

The AMD64 spec was released in 2000, so any remaining patents should be close to expiring.

Maybe that's why Microsoft didn't release the amd64 compatibility layer together with the x86-32 one back then.

 

[MikeZTM]

Why on Earth would you WANT them to run the same apps? The iPad and iPhones are jokes compared to a Mac in capability. Macs are vastly more powerful through the operating system and not limited to Apple only stores (although Apple sure seems to keep heading to them having a monopoly level CONTROL on Mac apps the way they do with iOS apps).

Calm down.
Mac can gain the ability to run iOS apps. And they can also maintain their ability to run macOS apps.
We are not talking about a new iPad with non-detachable keyboard.

If Apple decide to kill apps from outside the AppStore they can do it on Intel Mac also. It's not related to ARM.

Porting current app to a new ARM64 AppKit macOS platform is really easy just like how you can port iOS app to macOS now with Mac Catalyst or how iOS simulator runs iOS app natively x86_64 on Mac.

Catalina already kills all abandoned-wares. A new ARM Mac will run almost everything that runs on Catalina.

 

[oneMadRssn]

Many of the random chips in any computer have arm cores. Power management chips, system control chips, etc.

Do you actually have any proof of this? Or can you name a specific chip in a laptop or desktop? I'm genuinely curious.

I've researched this fairly extensively for my job. I know ARM has cores for such uses, but I haven't actually been able to find any laptops or desktops that actually implement them. Maybe they exist, but I can confidently say no consumer-level laptops or desktops from companies like Dell, HP, Lenovo, Asus, and Acer have ARM-licensed PM or control chips on their motherboards. Apple's T-series chips are quite unique in the industry for that reason. I've learned that desktop and laptop OEMs use less expensive suppliers for that purpose. Why pay ARM a license fee for a simple PMC when any number of other suppliers have perfectly good ones at a much lower costs?

I know of ARM-based PM and control chips in automotive, industrial, robotics, infrastructure, and other sectors - where individual chip cost is less important than performance and capability. But I have yet to see one in a laptop or desktop (other than the T-series in Macs). If you know of one, I'd love to know.

 

[dspdoc]

Why does Apple feel the need to continually fragment things?

EDIT: NVM, stupid Q... $$$

 

[cmaier]

Do you actually have any proof of this? Or can you name a specific chip in a laptop or desktop? I'm genuinely curious.

I've researched this fairly extensively for my job. I know ARM has cores for such uses, but I haven't actually been able to find any laptops or desktops that actually implement them. Maybe they exist, but I can confidently say no consumer-level laptops or desktops from companies like Dell, HP, Lenovo, Asus, and Acer have ARM-licensed PM or control chips on their motherboards. Apple's T-series chips are quite unique in the industry for that reason. I've learned that desktop and laptop OEMs use less expensive suppliers for that purpose. Why pay ARM a license fee for a simple PMC when any number of other suppliers have perfectly good ones at a much lower costs?

I know of ARM-based PM and control chips in automotive, industrial, robotics, infrastructure, and other sectors - where individual chip cost is less important than performance and capability. But I have yet to see one in a laptop or desktop (other than the T-series in Macs). If you know of one, I'd love to know.

Just google. For example, Dialog semiconductor uses ARM cores in its power management chips. So does TI, etc. I don't have a bill of materials for any particular mac, but if you have one you can check the data sheets for the various chips.
[automerge]1582661502[/automerge]

Why does Apple feel the need to continually fragment things?

Not sure what you are referring to is really "fragmenting."

 

[MikeZTM]

Do you actually have any proof of this? Or can you name a specific chip in a laptop or desktop? I'm genuinely curious.

I've researched this fairly extensively for my job. I know ARM has cores for such uses, but I haven't actually been able to find any laptops or desktops that actually implement them. Maybe they exist, but I can confidently say no consumer-level laptops or desktops from companies like Dell, HP, Lenovo, Asus, and Acer have ARM-licensed PM or control chips on their motherboards. Apple's T-series chips are quite unique in the industry for that reason. I've learned that desktop and laptop OEMs use less expensive suppliers for that purpose. Why pay ARM a license fee for a simple PMC when any number of other suppliers have perfectly good ones at a much lower costs?

I know of ARM-based PM and control chips in automotive, industrial, robotics, infrastructure, and other sectors - where individual chip cost is less important than performance and capability. But I have yet to see one in a laptop or desktop (other than the T-series in Macs). If you know of one, I'd love to know.

All silicon motion SSD controllers have Cortex M cores.
I didn't check but I guess same applies to HDD controllers also.

NVIDIA claims all their GPU after Maxwell have build-in ARM core for security usage.

And in reverse: All iPhone with Intel baseband actually have a x86 CPU in it.

 

[firewood]

Do you actually have any proof of this? Or can you name a specific chip in a laptop or desktop?

Nope. And it just might be possible those all those embedded software and hardware engineers at all the companies making disk controller, SSB controllers, WiFi radios, BLE radios, audio IO chips, touch panel controllers, keyboard controllers, PMCs, thermal management chips, camera modules, etc. as still using 8051's. But not likely.

BTW, there was an Intel CPU in the original Mac Classic.

 

[konqerror]

And in reverse: All iPhone with Intel baseband actually have a x86 CPU in it.

Actually only from the Xs generation onwards. They were still ARM before that.

Do you actually have any proof of this? Or can you name a specific chip in a laptop or desktop? I'm genuinely curious.

Adding to the list, all recent Broadcom Wifi client chipsets, Samsung SSD controllers are ARM. Broadcom ethernet chips were on MIPS. AMD's PSP, their version of the Intel ME, is ARM, so their CPU has an ARM core built right in.

Maybe they exist, but I can confidently say no consumer-level laptops or desktops from companies like Dell, HP, Lenovo, Asus, and Acer have ARM-licensed PM or control chips on their motherboards. Apple's T-series chips are quite unique in the industry for that reason.

HP Endpoint Security Controller in their Elite line computers serve much the same purpose at the T-series chip. I don't know what they use, but I bet its ARM. Dell also shipped the Broadcom USH security chip, a.k.a. Controlvault, which knowing Broadcom, is either ARM or MIPS.

 

[MikeZTM]

Actually only from the Xs generation onwards. They were still ARM before that.

I got a iPhone 7 with Intel baseband.
And a iPhone X with Intel also.

In general Intel's wireless baseband have x86 CPU on them. And since Apple acquired that department I guess they got some way to produce x86 chip for baseband legally though not really matters as they are going ARM and could just run the baseband software on ARM.

 

[konqerror]

I got a iPhone 7 with Intel baseband.
And a iPhone X with Intel also.

In general Intel's wireless baseband have x86 CPU on them. And since Apple acquired that department I guess they got some way to produce x86 chip for baseband legally though not really matters as they are going ARM and could just run the baseband software on ARM.

They were Intel but still ARM core until the Xs, left over from Infineon days. Somebody disassembled the iPhone X baseband firmware and proved this.

x86 finds its way into your iPhone

Personal page where to dump sparse knowledge

lcq2.github.io

 

[CarlJ]

iOS is MacOS.
iOS is stripped with a new framework. Under the hood they are both BSD.

Technically a Mach kernel with a BSD userland.

 

[Trusteft]

First time in a long while I am glad I can't buy a new Apple computer.
Unless the new CPUs are going to be X86 compatible, I see no reason to ever consider buying another Mac again.
Such a shame.

 

[Unregistered 4U]

In short, it will be the death of Macs.

Just like it was the death of the Apple ][, the death of OS 9, the death of PowerPC, things come to an end.

Steve Jobs would be rolling over in his grave now

Would that be the “milk the Macintosh for all it’s worth” Steve Jobs? Because if it were him, he’d be wondering what took them so long

 

[smulji]

Just like it was the death of the Apple ][, the death of OS 9, the death of PowerPC, things come to an end.

Would that be the “milk the Macintosh for all it’s worth” Steve Jobs? Because if it were him, he’d be wondering what took them so long

If it were SJ, he probably would have doubled down on the pace of innovation on the iPad to the point where there wouldn't be anymore Mac laptops.