Solved! ARM Apple High-End CPU - Intel replacement

[Richie Rich]

There is a first rumor about Intel replacement in Apple products:

• ARM based high-end CPU
• 8 cores, no SMT
• IPC +30% over Cortex A77
• desktop performance (Core i7/Ryzen R7) with much lower power consumption
• introduction with new gen MacBook Air in mid 2020 (considering also MacBook PRO and iMac)
• massive AI accelerator

Source Coreteks:

 

[soresu]

the fact we have no competition in this space and have a duopoly is crazy

The end of this sentence contradicts the beginning!

A duopoly is not much competition on the scale of things, but it is competition.

It has basically been this way for both desktop CPU's and discrete GPU's for almost 2 decades now, since the death/acquisition of 3dfx.

You can also say that the mobile OS market, at least in the west is a duopoly between Android and iOS, as the main HDD competitors have essentially become a duopoly between Seagate and Western Digital.

It's part of the long stretch of mergers and acquisitions over time that has contracted the IT market in many areas.

Mobile GPU's have stayed more varied due to Apple (sort of part time) investing in Imagination's PowerVR, Qualcomm buying Imageon/Adreno and ARM having it's own internal GPU, which as we know isn't doing so well for the last few years - former Vivante is a bit like VIA, just not well used at all outside of its famous use in Chromecast which may well end with the next CC hardware generation.

Tegra once held some hope for more variance, but the lack of any true mobile oriented Tegra chip since X1 scuppered that, so ironically it may be down to AMD to bring some new life into mobile gfx from their partnership with Samsung.

 

[Richie Rich]

What is your tips for Apple's A14 Firestorm core improvements over A13 Lightning?

1. Adding one more FPU to have 4xFPUs?
2. Adding one more Load AGU like Cortex A78/X1?
3. or adding one more complete Load/store AGU from 2xLSU -> 3xLSU?
4. ARMv9 +SVE2 instead ARMv8.3+NEON?
5. separating ALUs from Branch units 6xALUs (2xALU/Branch) -> 6xALU + 2xBranch
6. adding more ALUs from 6xALU to 8xALU?
7. increasing L1 cache from 128kB -> 256kB?
8. increasing L2 cache from 8 MB -> 16 MB? (A12 has 4MB)
9. increasing L3/SLC cache from 16 MB -> 32 MB? (A12 has 8MB)

Apple is planing bigger changes for completely new design coming with ARMv9+SVE2, similarly to ARM's Matterhorn. Last time Apple did major architecture update when A11 Monsoon was released. So it was A7 up to A10 (four iterations). This year we expect A14 as a last/four iteration of ARMv8+NEON core. If Apple would bring ARMv9+SVE2 in A14 core they won't spread development kits based on A12Z with ARMv8+NEON.

 

[soresu]

Apple is planing bigger changes for completely new design coming with ARMv9+SVE2, similarly to ARM's Matterhorn. Last time Apple did major architecture update when A11 Monsoon was released. So it was A7 up to A10 (four iterations). This year we expect A14 as a last/four iteration of ARMv8+NEON core. If Apple would bring ARMv9+SVE2 in A14 core they won't spread development kits based on A12Z with ARMv8+NEON.

Whatever it is it will have NEON.

The SVE2/TME announcement was pretty explicit about SVE2 capable cores having NEON for backwards compatibility, which makes sense considering the time that NEON has been around and therefore the amount of hand written assembly out there to rewrite for SVE2.

 

[name99]

Ugh.....

This one sentence alone displays a degree of technical ignorance in you, which a mere Google search could reveal if you were inclined to do so and educate yourself.

H.264/AVC is a video codec - AVI is a video container format, which has actually been used as a container for AVC encoded video back when MP4 was not as prevalent, and MKV had not yet been fully developed let alone in mass use.

Loose as it was, AVI was for all intents and purposes a standard in the days of DivX and XviD encoded video, until MP4 and MKV began to be used and replaced it.

You are comparing my desire for file system interoperability to people complaining about AVI being dead, really?

The only reason I can think of anyone to complain about the loss of AVI is because they actually mean XviD encoded video, which is still fairly commonplace across the net even today, in the age of encoding far superior like HEVC, VP9 and AV1, which have themselves surpassed H264/AVC considerably, much to the joy of Google's Youtube bandwidth bean counters.

I imagine XviD is still popular because so much consumer hardware like digital TV's and STB DVD players support it, likely far more than even H264 at this point - that's just a question of age and interoperability.

Right, technical ignorance.
You do realize that I'm referring to the state of the world c 1999, right, not today?
You do realize that's kinda my point! That people whine about anything that's not familiar, complain that the new is not "the standard", that these complaints have nothing to do with technical reality, and that it turns out that, once they get used to the new normal, they forget how much they whined at the beginning.

You do realize that I worked for Apple at that time in the QuickTime group, including optimizing a fair chunk of their h.264 code (along with a bunch of other stuff)?
You do realize that I lived through precisely the complaint I'm describing, when you were, what, not born? In kindergarden? The world did not begin in 2000, and the psychology of complaining about the new and unfamiliar has been with us a long long time.

 

[name99]

As I said before, I was able to access that same HDD from an Android phone - a phone!

NTFS and exFAT were created by MS but are both supported on Linux systems and by some Android vendors.

If you are going to charge through the nose for less performant systems - and boy do they ever, they should at least be able to read/write to storage formatted on cheaper and far more commonplace systems.

Is it that hard to believe that when most of my work and data has been recorded on Windows, which has by far the biggest marketshare of any consumer desktop OS, that maybe, just maybe an incredibly expensive alternate computer system would at least support the file systems necessary to open or write to those files on external, portable media like USB flash and HDD's?

If your answer is no, then you only reveal just how little you expect from Apple for the pleasure of having your wallet reemed, and that is truly quite sad - either that or you are loaded and just don't care either way.

Perhaps we would make more progress if you clarified what your complaints are.
Today you can take an exFAT (or other FAT variant) USB-C drive plug it into an iPad Pro, and it works. So, yes, you can read and write exFAT.

Today you can mount an SMB server on your iPhone or your iPad and access all that work and data you have stored on Windows.

So what exactly are you complaining about? That you can't plug a USB connector into your iPhone? That you want your iPad drive formatted as NTFS not exFAT? NTFS is owned by MS. MS controls who gets to license it from them, under what conditions and at what cost.
Are you just as angry that every IoT device out there, from cameras to cars, uses exFAT but not NTFS?

The issue seems not "I can't perform the data interchange I want" (you can, via exFAT or SMB); it seems to be "I demand Apple do things in the way maximally convenient to me, regardless of the relevance of that to Apple as a whole".

 

[Doug S]

What is your tips for Apple's A14 Firestorm core improvements over A13 Lightning?

1. Adding one more FPU to have 4xFPUs?
2. Adding one more Load AGU like Cortex A78/X1?
3. or adding one more complete Load/store AGU from 2xLSU -> 3xLSU?
4. ARMv9 +SVE2 instead ARMv8.3+NEON?
5. separating ALUs from Branch units 6xALUs (2xALU/Branch) -> 6xALU + 2xBranch
6. adding more ALUs from 6xALU to 8xALU?
7. increasing L1 cache from 128kB -> 256kB?
8. increasing L2 cache from 8 MB -> 16 MB? (A12 has 4MB)
9. increasing L3/SLC cache from 16 MB -> 32 MB? (A12 has 8MB)

Apple is planing bigger changes for completely new design coming with ARMv9+SVE2, similarly to ARM's Matterhorn. Last time Apple did major architecture update when A11 Monsoon was released. So it was A7 up to A10 (four iterations). This year we expect A14 as a last/four iteration of ARMv8+NEON core. If Apple would bring ARMv9+SVE2 in A14 core they won't spread development kits based on A12Z with ARMv8+NEON.

I'm not sure why you believe so strongly that A14 will be ARMv9. AFAIK ARM hasn't officially announced v9 yet, so it is unlikely to be finalized yet. Apple can't ship chips with a spec that ARM could still decide to change.

It takes around three months from start to finish to fab a wafer, then you do testing/packaging of the dies, deliver them to where products are being assembled, and assemble/test them to get them ready to ship. That means the first A14 iPhone wafers have already left TSMC. The first A14 Mac wafers for Macs that will be sold in December pretty much have to start fabrication any day now and more likely are already partway through that process.

While Apple obviously has people heavily involved in defining v9 they don't get the final say on it nor do they have the power to veto last minute changes to the spec. They couldn't have put it in the A14 design that's been finalized for a while now without taking on the considerable risk that they have to disable the whole thing if ANYTHING changes. It can really only go in an 'in progress' chip design where they still have time to make any necessary changes before it is fabbed. If the spec is pretty close to release they could be putting it in A15, and so long as that spec is made final (internally at least if not publicly announced yet) before A15 tapes out (which must happen no later than the end of the year, preferably several months sooner) they'd be OK.

Just look at the timeline for ARMv8, it was announced by ARM in October 2011, Apple shipped the first ARMv8 chip almost two years later in September 2013 - and everyone was shocked at how quickly they'd done that. Now you expect Apple to ship a chip before its architecture is announced??

Not sure why ARMv9 even matters, what do you believe it will bring that you are so hot and bothered about? SVE is already in ARMv8. SVE2 adds some new (mostly non FP related, and probably not all that important to Apple's customers versus what they'd get with SVE alone) capabilities on top of SVE and one would assume it will be marked optional in a future point release of ARMv8.

 

[name99]

I'm not sure why you believe so strongly that A14 will be ARMv9. AFAIK ARM hasn't officially announced v9 yet, so it is unlikely to be finalized yet. Apple can't ship chips with a spec that ARM could still decide to change.

It takes around three months from start to finish to fab a wafer, then you do testing/packaging of the dies, deliver them to where products are being assembled, and assemble/test them to get them ready to ship. That means the first A14 iPhone wafers have already left TSMC. The first A14 Mac wafers for Macs that will be sold in December pretty much have to start fabrication any day now and more likely are already partway through that process.

While Apple obviously has people heavily involved in defining v9 they don't get the final say on it nor do they have the power to veto last minute changes to the spec. They couldn't have put it in the A14 design that's been finalized for a while now without taking on the considerable risk that they have to disable the whole thing if ANYTHING changes. It can really only go in an 'in progress' chip design where they still have time to make any necessary changes before it is fabbed. If the spec is pretty close to release they could be putting it in A15, and so long as that spec is made final (internally at least if not publicly announced yet) before A15 tapes out (which must happen no later than the end of the year, preferably several months sooner) they'd be OK.

Just look at the timeline for ARMv8, it was announced by ARM in October 2011, Apple shipped the first ARMv8 chip almost two years later in September 2013 - and everyone was shocked at how quickly they'd done that. Now you expect Apple to ship a chip before its architecture is announced??

Not sure why ARMv9 even matters, what do you believe it will bring that you are so hot and bothered about? SVE is already in ARMv8. SVE2 adds some new (mostly non FP related, and probably not all that important to Apple's customers versus what they'd get with SVE alone) capabilities on top of SVE and one would assume it will be marked optional in a future point release of ARMv8.

Andrei F has said on a few occasions that (not exactly officially, but strongly hinted behind the scenes) ARM will ship v9 silicon in 2022.
Now starting with that what can you say about Apple?

On the one hand, Apple has (so far) been a lot more aggressive than ARM about rolling out any new feature of the ARM ISA.
On the second hand 2021 is still before 2022; maybe it's the A15 that will ARMv9?
On the third hand, it seems sub-optimal to release a year, and only a year, of transition ARM Mac's.
On the fourth hand, Apple has done this before. The first year of Intel Macs were on 32-bit only silicon, the next year brought x86-64 silicon, and, yes, of course in time that first year of Intel Macs were left behind.

Much of how much this matters hinges on how different v9 is from v8. Is it
- essentially a legal change, an opportunity to establish a new baseline of minimum capabiliies OR
- essentially v8++, so exactly the same as v8, plus the new stuff (SVE2, 8.6) PLUS perhaps a few more instructions (so kinda like ARMv8.7) OR
- is it an opportunity to reset a bunch of stuff that, after ~12 years of experience, clearly could be done better? Maybe some re-encoding of the instruction set? Some changes to details of the memory model? A way to pack the various high bits that are now used as tags by everyone (Swift/Objective C; PAC; MTE; santitizers; ...) differently given that servers will want to use large physical memories and OS's will want higher entropies for their security tagging.

My guesses are
- A14 will be ARMv8, A15 will be ARMv9
- A15 will run both v8 and v9 code, just like we had a few years of simultaneous ARMv7 and v8 code on iOS

My hope is that
- v9 will do something to provide for both larger tags and the same large (or larger) address space -- but I have no idea quite how this could be done well. One can imagine suboptimal (though likely practicable solutions, like PAE [short term, not my choice] or IBM style segments [slightly better, slightly more extensible, still not great]) but I'd hope for something better than those. The obvious choice of going to 128-bit pointers seems too heavyweight? (Thought not ABSOLUTELY crazy! The more your language has abstracted pointers, and can mostly use offset indices [at a variety of lengths from 8 to 64+ bits] rather than pure pointers, the less expensive it is. Safari did this with JS, for a big memory savings. LLVM has done it for a fair bit of their internal data structures. Swift (and Java?) could probably do more of it if they had a good reason to.
All the ALU side of things [adding offsets to pointers] could be done on the lower 64 bits, so you don't really even need to widen ALU; you just need a few data paths where what's propagated as an "address register" is in fact a register pair, even value is the address, odd value is the tags?...)

I agree with you on timelines. However we don't know the extent to which Apple gets to work with ARM REALLY early in the Apple SoC design process. The big possible example here is AMX. Is it independent Apple? Or is it ARMv8.6 matrix extensions (which were announced the same month as A13 and AMX were announced)?

 

[Doug S]

Andrei F has said on a few occasions that (not exactly officially, but strongly hinted behind the scenes) ARM will ship v9 silicon in 2022.
Now starting with that what can you say about Apple?

On the one hand, Apple has (so far) been a lot more aggressive than ARM about rolling out any new feature of the ARM ISA.
On the second hand 2021 is still before 2022; maybe it's the A15 that will ARMv9?
On the third hand, it seems sub-optimal to release a year, and only a year, of transition ARM Mac's.
On the fourth hand, Apple has done this before. The first year of Intel Macs were on 32-bit only silicon, the next year brought x86-64 silicon, and, yes, of course in time that first year of Intel Macs were left behind.

ARM doesn't ship silicon, so did Andrei say that ARM will release v9 core designs in 2022 (which means 2023 for silicon from Qualcomm et al) or that they will release them in 2021 which would allow OEMs using their core designs to ship in 2022?

If ARM plans to speed up their timeline from announcement of v9 to release of v9 cores vs v8 that doesn't say anything about Apple. ARM was slow last time, everyone knows it and they obviously are going to be less slow this time. ARM not dicking around like they did with v8 doesn't mean Apple automatically can or should move more quickly themselves.

If Apple sees a compelling reason to go v9 they might try to turn around more quickly this time, but if they want to hit A15 then v9 needs to be announced real soon now. They had only a year or so from the official release of v8 to when A7 taped out, so in earlier stages they were designing to a spec still in flux. There's only so much of that you can do without risking your overall timeline, and Apple absolutely cannot afford to have a design come late because that would make the iPhone 13 late (though if they call it "13" maybe it is doomed already )

 

[ThatBuzzkiller]

You’d be wrong about your first statement, but I guess that isn’t surprising given your posts.

Well, if you want to be technical about it, you can easily build and distribute applications on a mac without using the App Store. So you’re not locked in on that front, you should know that as a developer. If macOS is so terrible, would you really want to support it with your app? It’s already such a small percentage of the PC market, why not focus on Windows where everything is “open” instead? Oh wait, Windows is completely closed.

If you're talking about manually disabling gatekeeper through a command line that doesn't count since it's not officially supported by Apple ...

You didn’t even answer with a relevant response. Windows on ARM doesn’t support x64 apps. Even if it does have Office, that still isn’t enough. It is clearly half hearted. Where’s the Rosetta equivalent where you can basically run almost any app? What benefits do you get from buying a Windows on ARM machine? Basically nothing, besides the battery life, almost everything else is a con. Performance and compatibility are both missing, so why would you buy it if you had better alternatives for the same price? Anyway, we’ll see what happens, nobody can predict the future, but Apple has done this successfully before, twice, and Microsoft has tried this multiple times with lackluster results each time.

Microsoft Office isn't the only high quality software that they're offering. There's Visual Studio, .NET, Direct3D, WSL2, and they even offer their Azure cloud computing platform as well ...

"Where’s the Rosetta equivalent where you can basically run almost any app?"

This statement can't even be quantified but just to list out some caveats behind Rosetta 2 for you it likely doesn't work with x86-32 binaries just as it is the case on Catalina and it can't translate executables with kernel extensions, x86-64 virtualization or AVX/AVX2 instructions.

The reality is more fragile than you imagine if you believe ARM-based macs will offer far more than Windows on ARM. Compatibility through a translation layer like either Rosetta 2 or Microsoft's solution is ultimately irrelevant since the future will be determined with native apps. AMD and Intel has converged on AVX/AVX2 for some time and Rosetta 2 won't make a difference when the industry decides to converge on using AVX/AVX2 as well so Apple is going to be left hanging regardless even if they have a superior translation layer to Microsoft or not ...

Performance also isn't a good argument to make in favour of ARM-based macs either since Apple discourages low level programming practices and they encourage high abstraction costs like Rosetta 2 which ironically kills one of their main selling point about 'performance' ...

Apple doesn't have to "convince" people to adopt the ARM platform. They aren't buying a Mac to get an x86 CPU, anymore than they used to buy one to get a PPC or an 68K. They buy a Mac, today they get one with an x86 CPU, tomorrow they get one with an ARM CPU. The user base will follow them, that much has been established through two previous migrations.

Well I guess Apple can stop advertising boot camp altogether if they feel this way but they'd best be prepared for another mass exodus mac users since tons of them were using that feature ...

Microsoft has a very different problem trying to get people to want Windows/ARM, because they added it as a second platform. x86 is still the main Windows platform and will remain so for the foreseeable future. So they do have to "convince" people why they should choose ARM instead of x86, and I agree that there is absolutely no reason why they should.

Microsoft has added and abandoned non-x86 platforms too many times in the past so getting developers to port means not only making them believe there will be a market for their application running on ARM, but that their entire effort won't be wasted if Microsoft drops it. It is like trusting Google and relying on some new service they offer, when so many of them disappear into the 'google graveyard'.

Even if it were true that Windows on ARM is a second platform to Microsoft do you believe that Apple making ARM-based macs a priority will somehow erase the very same issues facing them ?

And sure Microsoft hasn't been the most forthcoming when it comes to exotic hardware designs but that's because they alone can't dictate the direction of the industry. Even Intel tried to do this with Itanium and it ended in failure. A single corporation no matter how big like Apple can't solely control industry consensus which will determine extremely important trends like software support and going against this consensus could have far reaching negative consequences ...

Are macs subsisting on scraps from iOS somehow supposed to be superior to subsisting on scraps from Windows and Intel ? Even if Apple does solely transition to ARM-based macs it still won't change the prevailing industry consensus of developing on x86 platforms which persist in the foreseeable future ...

 

[soresu]

You do realize that I'm referring to the state of the world c 1999, right, not today?

Ignorance redoubled once more, are you even capable of using Google at all?

H264 was not even ratified until 2003, and x264 not even remotely capable of producing quality video for years after that.

Even Apple's initial 720p trailers were basically unplayable on brand new machines due to the decoder in Quicktime being a barebones implementation with little optimisation.

 

[soresu]

or that they will release them in 2021 which would allow OEMs using their core designs to ship in 2022?

Pretty sure Andrei meant this latter, Matterhorn is definitely the post Hercules/Hera core as shown from the TechCon presentation last October.

So announcement next May/June is likely, with products to follow early 2022 as you say.

It seems unlikely that v9-A wouild be announced with the cores though, so I'd say it will be announced at the TechCon in October - or through some virtual event if COVID measures are still impeding travel.

 

[Richie Rich]

Andrei F has said on a few occasions that (not exactly officially, but strongly hinted behind the scenes) ARM will ship v9 silicon in 2022.
Now starting with that what can you say about Apple?

@Andrei. said this:"Matterhorn is a Q4 2021/ Q1 2022 silicon product. But let's not get that minor detail get into the way of spreading made up rumours."

News - ARM Matterhorn

Found this on the ARM TechCon blog: Someone mentioned this to me recently as something related to ARMv9.0-A, possibly the post Hercules big core. Link for the ARM TechCon blog here. A quote under another image: "MatMuil will double CPU GEMM performance" "Matternhorn will introduce what Arm...

forums.anandtech.com

So big announcement of ARMv9 and new core Matterhorn core line up (A79, X2 and A59(?) ) is next year (2021).
Qualcomm new silicon is released Q1 usually, so it's Q1/2022. HiSilicon can be Q4/2021. This means HiSilicon and Qualcomm SoC based on Matterhorn are going to be taped out right now (mid 2020). Apple A15 was taped out one quarter earlier so might have running engineering samples in a lab.

On the one hand, Apple has (so far) been a lot more aggressive than ARM about rolling out any new feature of the ARM ISA.
On the second hand 2021 is still before 2022; maybe it's the A15 that will ARMv9?
On the third hand, it seems sub-optimal to release a year, and only a year, of transition ARM Mac's.
On the fourth hand, Apple has done this before. The first year of Intel Macs were on 32-bit only silicon, the next year brought x86-64 silicon, and, yes, of course in time that first year of Intel Macs were left behind.

You are right. Apple's first 64-bit CPU at ARMv8 A7 Cyclone was released one two years after 64-bit ARMv8 announcement. So Apple engineers were working on it three years before. Cortex A57 was announced 2012. Apple A7 Cyclone as a first 4xALU core and being much complex design than simple A57 took more time to finish. Amazing thing is that Apple was basically matching Intel tech level (first 4xALU x86 CPU Haswell was released just two months before A7). Since then Apple was constantly increasing IPC 20% every year while Intel got stuck at Skylake for 5 years.

So probably you are right that A14 will be still ARMv8+NEON based. However I think Apple added fourth FPU to make it more Laptop friendly.

Much of how much this matters hinges on how different v9 is from v8. Is it
- essentially a legal change, an opportunity to establish a new baseline of minimum capabiliies OR
- essentially v8++, so exactly the same as v8, plus the new stuff (SVE2, 8.6) PLUS perhaps a few more instructions (so kinda like ARMv8.7) OR
- is it an opportunity to reset a bunch of stuff that, after ~12 years of experience, clearly could be done better? Maybe some re-encoding of the instruction set? Some changes to details of the memory model? A way to pack the various high bits that are now used as tags by everyone (Swift/Objective C; PAC; MTE; santitizers; ...) differently given that servers will want to use large physical memories and OS's will want higher entropies for their security tagging.

I agree. As far as I know ARMv9 is actually ARMv8.6+SVE2+TME plus maybe some minor changes towards server usage. Apple doesn't support 32-bit on their latest chips so I guess ARMv9 is a new opportunity to throw away backward compatibility in favor of performance leap. Totally opposite aproach to x86 where latest chips has to deal with backward compatibility for instructions from 1986 which nobody uses anymore.

ARM doesn't ship silicon, so did Andrei say that ARM will release v9 core designs in 2022 (which means 2023 for silicon from Qualcomm et al) or that they will release them in 2021 which would allow OEMs using their core designs to ship in 2022?

You are wrong. Andrei said Q4 2021 ARMv9/Matterhorn silicon. So next year.

 

[Doug S]

Well I guess Apple can stop advertising boot camp altogether if they feel this way but they'd best be prepared for another mass exodus mac users since tons of them were using that feature ...

It is already known that Boot Camp will not be an option for ARM Macs to run Windows. Likely because Apple doesn't want to write/support Windows drivers for their GPU. You'll be able to run it inside a VM, which is fine for people who "just need to run something on Windows". It isn't any good for people who want to primarily/exclusively run Windows - i.e. people who bought a Macbook because they think it is better than any Windows laptops. But how many people could that really be? You say "tons of them were using that feature". Based on what, anecdotal evidence? I'm not aware of any hard numbers out there.

I'm sure Apple is plunging ahead into this scenario which you feel will be their doom without any data, that's the sort of thing the most successful companies in the world do all the time, after all.

 

[Doug S]

You are wrong. Andrei said Q4 2021 ARMv9/Matterhorn silicon. So next year.

How can I be "wrong" when I asked a question about what he meant?

For chips to ship in Q4 2021 ARM will need to release the core designs this quarter. Maybe they are planning on releasing the architecture and the core designs at the same time this time around?

If that's the case then ARM would have frozen the spec and sat on it while they completed the design. Not sure what the motivation for doing so would be, seems like that would unnecessarily delay/handicap compiler support.

 

[Eug]

It is already known that Boot Camp will not be an option for ARM Macs to run Windows. Likely because Apple doesn't want to write/support Windows drivers for their GPU. You'll be able to run it inside a VM, which is fine for people who "just need to run something on Windows". It isn't any good for people who want to primarily/exclusively run Windows - i.e. people who bought a Macbook because they think it is better than any Windows laptops. But how many people could that really be? You say "tons of them were using that feature". Based on what, anecdotal evidence? I'm not aware of any hard numbers out there.

I'd be shocked if even 1/25 of Mac users use Boot Camp on a regular basis, and I suspect it's significantly lower than that.

Boot Camp was a big deal back in 2006. In 2020, not so much.

Use of Parallels / VMware is probably higher than the number of Boot Camp users though. The question is how quickly they will release that software on macOS ARM.

IOW, I see the loss of Boot Camp support as a non-issue overall, but it may be more of an issue that people won't be able to run Windows under a VM, since Rosetta 2 doesn't support this.

 

[soresu]

You are wrong. Andrei said Q4 2021 ARMv9/Matterhorn silicon. So next year.

Qualcomm new silicon is released Q1 usually, so it's Q1/2022. HiSilicon can be Q4/2021. This means HiSilicon and Qualcomm SoC based on Matterhorn are going to be taped out right now (mid 2020).

Don't make any bets about HiSilicon plans where next gen cores and fab nodes are concerned.

The current US govmt administration stance against Chinese businesses, especially Huawei has led to serious problems for TSMC (Huawei's main supplier) in supplying them, and possibly even ARM supplying core designs to them also.

It will certainly be a shame if so because Huawei/HiSilicon were getting the first crack at the bat for A73 and A76, months before Qualcomm.

Also Qualcomm's chip next year (875) will be A78/X1 based, so they won't be taping out anything Matterhorn based this early - more likely it will be taped out just as the A78 chip is in first products.

 

[Richie Rich]

For chips to ship in Q4 2021 ARM will need to release the core designs this quarter. Maybe they are planning on releasing the architecture and the core designs at the same time this time around?

Yes. As usual. A78 and X1 was announced this Q2 and first silicon is expected in Q4'20 from HiSilicon Kirin 1000 and from Qualcomm SD875 in Q1'21. This means that A78/X1 SoC was taped out in mid 2019..... so this means ARM LLC fihished A78/X1 core designs quarter ahead? so Q1'19. Same with previous A77 and A76. So Matterhorn cores will keep same schedule. This means Matterhorn design was already finished Q1'20 and vendors are working on SoC layout tape out. Pretty exciting but no leaks yet. I hope there will be more leaks in future due to higher number of vendors

Also Qualcomm's chip next year (875) will be A78/X1 based, so they won't be taping out anything Matterhorn based this early - more likely it will be taped out just as the A78 chip is in first products.

Isn't Qualcomm always releasing new hi-end chips in January? So Matterhorn Snapdragon 885 is about to tape out soon (in Q3'2020).

 

[Doug S]

This means Matterhorn design was already finished Q1'20 and vendors are working on SoC layout tape out.

If that's true, and Matterhorn is v9, why hasn't ARM announced ARMv9 or at least released the specs? What do they gain by keeping them secret?

 

[ThatBuzzkiller]

It is already known that Boot Camp will not be an option for ARM Macs to run Windows. Likely because Apple doesn't want to write/support Windows drivers for their GPU. You'll be able to run it inside a VM, which is fine for people who "just need to run something on Windows". It isn't any good for people who want to primarily/exclusively run Windows - i.e. people who bought a Macbook because they think it is better than any Windows laptops. But how many people could that really be? You say "tons of them were using that feature". Based on what, anecdotal evidence? I'm not aware of any hard numbers out there.

No they don't have to write drivers for their trash tile-based GPUs and they can opt to use the WARP software renderer if any graphical output was absolutely necessary but that's another annoying gripe about Apple going against programming trends. Just about every high-end/professional graphics applications are designed with AMD/Intel/Nvidia GPUs in mind AKA immediate mode/sort-last GPUs so performance goes out the window with their sole introduction of tile-based GPUs. Even Samsung wants to use AMD GPUs because they understand the value in matching programming trends ...

BTW Rosetta 2 doesn't support x86-64 VMs so good luck to any mac users out there doing virtualization!

Seeing as how Apple keeps dropping features like x86-32 either Boot Camp or a dedicated Windows device is practically necessary at this point for running productive desktop software in the modern world.

I'm sure Apple is plunging ahead into this scenario which you feel will be their doom without any data, that's the sort of thing the most successful companies in the world do all the time, after all.

It's Apple that doesn't have any data to backup their move and their success is primarily based on hardware rather than software so why is it hard for you to accept this fundamental basis ?

You can't build a sustainable platform by building software around hardware. It's industry consensus that hardware should serve software. If hardware was all that truly mattered then why should Apple ever stop with ARM ? Would it still be in Apple's best interest to drop ARM for a superior hardware architecture then ?

 

[RasCas99]

No they don't have to write drivers for their trash tile-based GPUs and they can opt to use the WARP software renderer if any graphical output was absolutely necessary but that's another annoying gripe about Apple going against programming trends. Just about every high-end/professional graphics applications are designed with AMD/Intel/Nvidia GPUs in mind AKA immediate mode/sort-last GPUs so performance goes out the window with their sole introduction of tile-based GPUs. Even Samsung wants to use AMD GPUs because they understand the value in matching programming trends ...

BTW Rosetta 2 doesn't support x86-64 VMs so good luck to any mac users out there doing virtualization!

Seeing as how Apple keeps dropping features like x86-32 either Boot Camp or a dedicated Windows device is practically necessary at this point for running productive desktop software in the modern world.



It's Apple that doesn't have any data to backup their move and their success is primarily based on hardware rather than software so why is it hard for you to accept this fundamental basis ?

You can't build a sustainable platform by building software around hardware. It's industry consensus that hardware should serve software. If hardware was all that truly mattered then why should Apple ever stop with ARM ? Would it still be in Apple's best interest to drop ARM for a superior hardware architecture then ?

You are so hostile , also you drone on and on about industry consensus that hardware should serve software , it is of course false and depends on who you are in the industry , the OEM`s love HW progression , they make money out of it , they for all the intended purposes would LOVE this change , if it provides better HW , the "X86" SW houses hate changes like this because it costs them MONEY to change over instead of making it.

But you can just look at the iPhone , went into a billion devices a year market , with established market leaders (Nokia/Blackberry) and by building a better HW , they made it happen in the long run.
The iPhone had no SW at the start , it took time and iterations (first year iPhone was not crossing the %1 market share) , but the better HW won at the end and enabled a SW echo system around it.

You are SO short sighted with your comments , do you truly believe we should encourage or accept that Intel/AMD and Nvidia/AMD should be the only game in town ? I applaud Intel for trying to disrupt the GPU market and Apple for the CPU market , and I am sure that if they make better HW , SW will follow , and for one reason and the best reason for business - MONEY.
You can see MS trying their best to be a player in the ARM space , and they will iterate on their version , they must , if not they WILL lose market share once a capable QC chips starts rolling out and Google Chromebook will be something more then a cheap browser , it happened in the phone market it can happen here again.

You actually trying to say that the way things are today should always remain this way into the future no matter what because "the industry" ? nope , thats not how tech works and it is not how its going to work into the future , the competition caught up with the subpar X86 CPU`s and its game on now , lets let them race , its time for Intel and AMD to put some die size into their CPU`s and give us consumers more bang for our buck , done are the days we are waiting ages for CPU to make progress on the back of node shrinks and new Uarch , its time to pay in die size and margins to be competitive!

"It's Apple that doesn't have any data to backup their move and their success is primarily based on hardware rather than software so why is it hard for you to accept this fundamental basis ?"
Its the above statement that actually seals the deal in regards to your crazy line of thought and your perception of what you know , you actually try and claim that the poster should accept your made up "fundamental basis" that Apple ,who sell computers for decades and write the SW for those computers have no data to backup their move ? I am sorry to say that this goes beyond any reasonable logic thinking , you could say they are making a mistake and expand on that , but it wont be for the lack of data or basic understanding of their consumers , thats for sure.

Let me ask you something , why does Apple move to their Arm CPU`s if its so obvious that all that is going to happen is they will lose market share ,lose sales and spend MORE on R&D ? what is the play here? and please think before you reply something snarky about how they dont understand , it will be embarrassing.

 

[ThatBuzzkiller]

You are so hostile , also you drone on and on about industry consensus that hardware should serve software , it is of course false and depends on who you are in the industry , the OEM`s love HW progression , they make money out of it , they for all the intended purposes would LOVE this change , if it provides better HW , the "X86" SW houses hate changes like this because it costs them MONEY to change over instead of making it.

OEMs wouldn't like the change at all because a platform's value is largely derived from it's software rather than hardware hence why hardware should be built for the software. What you see as maintaining compatibility isn't a weakness but it is strength behind the x86 architecture ...

x86 has stood the test of time and it even has power to transcend beyond politics that even a Chinese state backed joint venture like Zhaoxin is entering the race despite where the IP had originated ...

But you can just look at the iPhone , went into a billion devices a year market , with established market leaders (Nokia/Blackberry) and by building a better HW , they made it happen in the long run.
The iPhone had no SW at the start , it took time and iterations (first year iPhone was not crossing the %1 market share) , but the better HW won at the end and enabled a SW echo system around it.

Apple also didn't design most of their critical hardware components either for the original iPhone. Their CPU/memory were supplied from Samsung and their graphics processor was provided from Imagination Technologies ...

As far as having no software at the start that's not true because there was once a time where iOS provided lot's of value to customers but as Apple has started to obsess over hardware their software has degraded in quality over the years. Now iOS is in a constantly broken state and what's more is that after every update tons of programs from the App Store keeps breaking!

Apple has changed from software to hardware and it's not for the better either since every last one of their platforms they've created seems to die out every time. MacOS which was once a real competitor to Microsoft's Windows fell hard from grace and it's only a matter of time before Google's Android absolutely decimates iOS to a sub-10% market share ...

You are SO short sighted with your comments , do you truly believe we should encourage or accept that Intel/AMD and Nvidia/AMD should be the only game in town ? I applaud Intel for trying to disrupt the GPU market and Apple for the CPU market , and I am sure that if they make better HW , SW will follow , and for one reason and the best reason for business - MONEY.
You can see MS trying their best to be a player in the ARM space , and they will iterate on their version , they must , if not they WILL lose market share once a capable QC chips starts rolling out and Google Chromebook will be something more then a cheap browser , it happened in the phone market it can happen here again.

I didn't say that AMD/Intel/Nvidia should be the only vendors when it comes to CPU/GPUs and 'better' HW doesn't necessarily mean SW will follow suit as well since Intel's Itanium architecture started off with a more modern design but the software industry wasn't having it in the end ...

You actually trying to say that the way things are today should always remain this way into the future no matter what because "the industry" ? nope , thats not how tech works and it is not how its going to work into the future , the competition caught up with the subpar X86 CPU`s and its game on now , lets let them race , its time for Intel and AMD to put some die size into their CPU`s and give us consumers more bang for our buck , done are the days we are waiting ages for CPU to make progress on the back of node shrinks and new Uarch , its time to pay in die size and margins to be competitive!

That depends on the circumstances. x86 has seen a huge progression in terms of it's programming model with the introduction of AVX meanwhile Apple doesn't have any comparable offering and is in fact a regression of programming standards seeing as they keep adding more restrictions and abstractions ...

x86 hardware design supplements software design very well which favours the software industry while the same can't be said for Apple's CPU design since their too afraid of releasing a programmer's manual for them ...

x86 might be a duopoly but at least AMD/Intel are more open than Apple about releasing low-level details in their designs ...

"It's Apple that doesn't have any data to backup their move and their success is primarily based on hardware rather than software so why is it hard for you to accept this fundamental basis ?"
Its the above statement that actually seals the deal in regards to your crazy line of thought and your perception of what you know , you actually try and claim that the poster should accept your made up "fundamental basis" that Apple ,who sell computers for decades and write the SW for those computers have no data to backup their move ? I am sorry to say that this goes beyond any reasonable logic thinking , you could say they are making a mistake and expand on that , but it wont be for the lack of data or basic understanding of their consumers , thats for sure.

Let me ask you something , why does Apple move to their Arm CPU`s if its so obvious that all that is going to happen is they will lose market share ,lose sales and spend MORE on R&D ? what is the play here? and please think before you reply something snarky about how they dont understand , it will be embarrassing.

Retailing hardware is what Apple has done traditionally in the past. Only very recently have they sourced their own hardware components which marked a shift to their hardware strategy ...

Also I do not expect Apple to understand at all since they're a lost cause when they don't spend a dime on their software department. If Microsoft can be accused of taking a non-committed attitude towards Windows on ARM then the same can be said Apple as well since they're putting in a similar amount of effort too which is practically amounts to nearly nothing. Apple are mostly expecting macs to basically subsist off of the scraps from iOS instead of Windows/Intel but it's going to end up about as compelling as an iPad is ...

I wonder if Mac users feel more happy about having an inferior version of iOS over being an inferior alternative to Windows/Intel ?

 

[defferoo]

You can't build a sustainable platform by building software around hardware. It's industry consensus that hardware should serve software. If hardware was all that truly mattered then why should Apple ever stop with ARM ? Would it still be in Apple's best interest to drop ARM for a superior hardware architecture then ?

This statement is very telling of your perspective. The whole point of this transition is to enable more hardware and software innovation in the long-term. The reason Apple is so successful is that they marry hardware and software in a seamless fashion that is functional and intuitive. Hardware innovation drives software innovation and vice versa. It’s a beautiful thing really. Take the iPhone, without the right hardware (imagine a stylus on a plastic resistive touch screen), it would have flopped immediately, but without the software to take advantage of the hardware (multi-touch gesture support), they have nothing.

There will certainly be a somewhat painful transition period, but the payoff will be huge for years to come. Macs currently lag behind the iPhone and iPad in terms of innovation, taking control of the silicon enables Apple to take the destiny of the Mac into their own hands. Innovating in both hardware and software together will make their future products great.

There’s also little reason for them to move away from ARM now that they design their own chips, unless they feel that the ISA is missing something they need. The reason is that they can simply add whatever functionality they want to the chip and expose it via an API. This is the kind of thing that will set them apart from x86 machines.

 

[Richie Rich]

If that's true, and Matterhorn is v9, why hasn't ARM announced ARMv9 or at least released the specs? What do they gain by keeping them secret?

I guess ARMv9 is just new marketing label for ARMv8.6 and it's not that big change like going 32-bit to 64-bit. All parts like SVE2+TME was introduced already, SVE HW exists too (Fujitsu A64FX). Because Apple A13 has some AMX support already the next A14 might be SVE2 ARMv9 ready. We will see soon.

You can't build a sustainable platform by building software around hardware. It's industry consensus that hardware should serve software. If hardware was all that truly mattered then why should Apple ever stop with ARM ? Would it still be in Apple's best interest to drop ARM for a superior hardware architecture then ?

Steve Jobs was told that if you mean SW dev seriously then you have to develop your own HW. That's all what Apple does. And it works great apparently.

 

[name99]

If that's true, and Matterhorn is v9, why hasn't ARM announced ARMv9 or at least released the specs? What do they gain by keeping them secret?

Partner agreements? Anyone selling an ARMv8 phone today will be worried about having their phones Osborne'd, and will want time to put in place a strategy for
- selling down older inventory
- creating a story about how their particular v9 phone is better than all those other v9 phones.

But I agree that I find a 2021 date too ambitious (except for Apple). There'll need to be some OS work done, some compiler work done, and we aren't yet seeing evidence of that.
Apple can do that sort of stuff in secret; it's harder for ARM Ltd to do it all in secret without gcc/LLVM/Linux mods becoming visible.

 

[name99]

BTW Rosetta 2 doesn't support x86-64 VMs so good luck to any mac users out there doing virtualization!

Don't be so sure you understand what is going on here; clearly you don't.
https://www.parallels.com/blogs/apple-silicon-wwdc/
Note in particular "...makes it simple for businesses and individuals to use the applications and files from *any operating system* they need on their Macs"

Now how are they doing that? I have my hypotheses, but I'm sick of typing them up repeatedly, and having them ignored by people still living in the computing stone age. So I'll just leave you with that statement from Parallels.