Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
My $64,000 question for everyone is - If NVIDIA is the bees knees from a GPU standpoint, why did both Sony AND Microsoft choose AMD to power the PS5 and Xbox One X?
[automerge]1587751134[/automerge]

Apple does not view x86 as necessary anymore. It worked for a time and that time has passed.

Game consoles need GPUs and CPUs. NVIDIA does not do CPUs.

Apple may not view x86 as necessary (although there is not evidence of this) but is it even important? It's more important whether potential customers view x86 as necessary (and some of them definitely do).
 
Gotcha.

Yeah it'll be interesting to see if any of the other PC manufacturers hop aboard the ARM train.

If Apple finally found the secret to dumping Intel... will anyone else follow? Can anyone else do that?

Apple has been making their own processors for quite some time... but I don't see Lenovo, Dell and HP doing that. So maybe Qualcomm to the rescue?

I like the idea of Apple kickstarting the migration away from Intel. I'm just not sure if it will cause an industry-wide shift.

I could see the major PC manufacturers shifting to AMD much sooner than a shift to ARM.

Intel will throw around money to keep PC OEMs in line to the extent that they can. AMD has the cards stacked against it from a certain point of view, but they will keep making inroads and fighting it out for design wins as much as they can.

Microsoft can minimize Intels influence by using AMD CPUs, but that doesnt stop x86 from dominating our foreseeable future. Is x86 inherently bad? No, not really? But x86 is tied to Intel inextricably, so much so that I don’t think Intel can die or truly be dumped. Remember, they are only one process win away from being right back in the hunt. If they get 7nm right, then everyone needs to pay attention.

Bottom line, ARM/Arm-based PCs wont ever see the light of day beyond Apple. The PC OEMs are too scared of Intel to make a dramatic change.
 
  • Like
Reactions: Michael Scrip
Game consoles need GPUs and CPUs. NVIDIA does not do CPUs.

Apple may not view x86 as necessary (although there is not evidence of this) but is it even important? It's more important whether potential customers view x86 as necessary (and some of them definitely do).

Sony and MS could have had Intel or AMD custom design a CPU and used NVIDIA GPUs if they truly wanted to do that. That NVIDIA doesnt do CPUS is a moot point. Whether NVIDIA even careS about the console market is a moot point as well.


As @cmaier aptly observed, Apple has a billion or so devices running on ARM/Arm and about 75 million running on Intel. Guess who wins that argument. If x86 compatibility is the supposed end all, be all ultimate decision maker then why even have something more advanced than a 486 DX/50 in the year 2020? Its not like Intel has evolved past the 1990s. Being tied to a single ancient CPU architecture is just stupid. Its stupid for users and manufacturers. Good grief I miss the 1990s when there was 68K, PowerPC, MIPS, DEC Alpha, SPARC, PA-RISC, ARM, Clipper AND Intel. It was vibrant, a bit messy and chaotic and even quixotic, but it wasn’t the same old 14nm+++++++ excuses s***show song and dance we’ve been getting from Intel since Broadwell.

Now all I have is this x86 s*** with Intel droning on like the dead bloated cow they’ve been for the past 20 years, doing the absolute minimum work necessary, living on fat profits and stifling anything else as best they can. And all of you “x86 or Bust” people just enabling the same Jabba The Hutt cartel business practices. Bring on the ARM/Arm revolution, or grease fart or whatever it will be. Just something other than Intel. And AMD (Sorry, Dr Su, its nothing personal).
 
Why should Apple care about that when they could finally create they completely walled off garden?

Economics and return on investment. If Apple wants to keep widespread interest in developing on their platforms then they need to continue offering hardware that can do more than just create apps for iOS and iPadOS. Developers also need to create web apps, web sites, IoT apps, etc.

There are a number of frameworks that create multiplatform apps such as React Native, Flutter, Nativescript, Xamarin, etc. If Apple cut all of that off it would have a very negative impact on their market share in software development. Developers want choice and flexibility, not pigeon-holed environments.
 
  • Like
Reactions: Nugget
Bottom line, ARM/Arm-based PCs wont ever see the light of day beyond Apple. The PC OEMs are too scared of Intel to make a dramatic change.
Someone forgot to tell that to Microsoft. ;)
Surface Pro X is ARM based and runs full Windows 10, not a stripped version.
Can run x86 (32bit) apps, ARM 32bit/64bit apps. Only restriction is it currently cannot run x64 apps.

There are ARM based PCs out there using Qualcomm 8cx CPUs.
But again.. app compatibility and performance issues running emulated (x86) apps are still there.

Hopefully Apple will have worked out any performance issue before it decides to unleash this on the public.
 
  • Like
Reactions: MikeZTM
Someone forgot to tell that to Microsoft. ;)
Surface Pro X is ARM based and runs full Windows 10, not a stripped version.
Can run x86 (32bit) apps, ARM 32bit/64bit apps. Only restriction is it currently cannot run x64 apps.

There are ARM based PCs out there using Qualcomm 8cx CPUs.
But again.. app compatibility and performance issues running emulated (x86) apps are still there.

Hopefully Apple will have worked out any performance issue before it decides to unleash this on the public.

I watched Intel kill off DEC Alpha, PowerPC and MIPS support of WindowsNT, so while I give Microsoft props for going to the trouble of releasing the Surface Pro X, it’s one model and won’t sell in high enough volume to make Intel take notice or MS to commit to expanding ARM-based product to other form factors. I wish that wasn’t the case, but Intel will eventually call MS and the Pro X will simply disappear. MS doesn’t have the stomach for that fight.
 
I watched Intel kill off DEC Alpha, PowerPC and MIPS support of WindowsNT, so while I give Microsoft props for going to the trouble of releasing the Surface Pro X, it’s one model and won’t sell in high enough volume to make Intel take notice or MS to commit to expanding ARM-based product to other form factors. I wish that wasn’t the case, but Intel will eventually call MS and the Pro X will simply disappear. MS doesn’t have the stomach for that fight.
I think what will kill the Pro X faster than anything Intel could do, is the price.
$900 starting price with no pen and no keyboard. Given its limitations, MS should have included the pen and keyboard instead of asking another $260 for them.
 
  • Like
Reactions: AxiomaticRubric
I think what will kill the Pro X faster than anything Intel could do, is the price.
$900 starting price with no pen and no keyboard. Given its limitations, MS should have included the pen and keyboard instead of asking another $260 for them.
Well, the 12.9” 128GB iPad Pro is $999 with no pencil and no keyboard, so I think I know where MS was going, but they seemingly forgot the SP X also has no apps!🤷🏻‍♂️
 
Technically Power is a descendant of the old joint IBM-Apple chips, that Apple moved off of for x86.
...

Maybe in a alternative universe. IBM Power is descendent of the IBM 801 project which predated all of the RISC stuff that Hennessy and Patterson did.

Power3 is where PowerPPC merged in with Power. That was a significant change to Power, but Power existed before PowerPC.... that is why the 'PC" was tacked on the end the existing name to create the "new" name for the effort. Motorola threw some ideas and experience from the Motorola 88000 into PowerPC and Apple tossed in a few contributions, but the baseline IBM had already laid down.


IBM still ships it, still puts it in server class and rack boxes (and some of the Qualcomm stuff is from ex-Power folks, really).
There is still Motorola/Freescale/NXP PowerPC embedded processors sold. Just not close to the "personal computer" market.

Power after PowerPC wrapped up has some internal syngeries with IBMs Z-class mainframe CPUs ( where share some elements, but not exact implementations. ). IBM moved their AS/400 class servers to them also.\

Not just IBM shipping Power systems anymore either. They have somewhat open llcensed the chip for Linux based systems from other vendors.

 
  • Like
Reactions: rjohnstone
Maybe in a alternative universe. IBM Power is descendent of the IBM 801 project which predated all of the RISC stuff that Hennessy and Patterson did.

Power3 is where PowerPPC merged in with Power. That was a significant change to Power, but Power existed before PowerPC.... that is why the 'PC" was tacked on the end the existing name to create the "new" name for the effort. Motorola threw some ideas and experience from the Motorola 88000 into PowerPC and Apple tossed in a few contributions, but the baseline IBM had already laid down.



There is still Motorola/Freescale/NXP PowerPC embedded processors sold. Just not close to the "personal computer" market.

Power after PowerPC wrapped up has some internal syngeries with IBMs Z-class mainframe CPUs ( where share some elements, but not exact implementations. ). IBM moved their AS/400 class servers to them also.\

Not just IBM shipping Power systems anymore either. They have somewhat open llcensed the chip for Linux based systems from other vendors.

All of Power was terrible, though, because they started number bits with 0 as the highest-order bit.

I can’t begin to tell you what a pain in the ass it was to design a PowerPC because of that.
 
  • Like
Reactions: Zdigital2015
If Apple finally found the secret to dumping Intel... will anyone else follow? Can anyone else do that?

It has already started:

MS Surface Pro X: https://www.microsoft.com/en-gb/p/surface-pro-x/8vdnrp2m6hhc?activetab=overview
Lenovo Yoga 5G: https://www.lenovo.com/gb/en/laptops/yoga/yoga-c-series/Lenovo-Yoga-5G-/p/88YGC801370
Samsung Galaxy Book S: https://www.samsung.com/global/galaxy/galaxy-book-s/
...and several Chromebooks: http://www.linuxmadesimple.info/2019/08/all-chromebooks-with-arm-processors-in.html
(Bear in mind - current Chromebooks are no longer web-only machines - they can also run both Android and (regular) Linux apps)

...although they're not exactly setting the world on fire...

Thing is, when Lenovo or Samsung announce a new laptop, it is hardly front page news... but when Apple announce a new product, oh yes it is. If/when Tim stands up and announces that Apple will be dumping Intel processors, the world and their dog will know. It will be a massive publicity coup for ARM.

The other thing that is easy to forget is that both Intel and Microsoft have already lost one major battle: mobile phones and tablets may not be a total replacement for "proper" PCs but they have cut a swathe through the PC market - and the mobile market is completely dominated by ARM (and Unix-like operating systems). Nobody could have predicted that 10 years ago. Windows has also lost a substantial chunk of the server/web services/cloud computing to Linux (which is built on source-level compatibility and largely processor agnostic beyond the kernel level).

Apple has been making their own processors for quite some time... but I don't see Lenovo, Dell and HP doing that. So maybe Qualcomm to the rescue?

Qualcomm, Samsung (Exynos), Ampere (https://amperecomputing.com/) - even Amazon have developed an ARM chip (although that's probably an exclusive for their cloud computing service). The ARM club isn't cheap to join but because ARM will license anything from the instruction set itself though to a complete system-on-a-chip design, its far more accessible than the x86 club.

Or, of course, Intel. Fun fact: Intel have made ARM chips in the past (https://en.wikipedia.org/wiki/XScale) and could do again. In a crazy world where Microsoft produces software for Linux, who knows...

Of course Apple are the best set up to make their own processors because they have the huge iPhone/iPad market to supply...


I think what will kill the Pro X faster than anything Intel could do, is the price.

Its a Surface - the only range of computers that make Macs look over-specified and under-priced (but boring*).

I'm never quite sure whether Microsoft want people to actually buy their computers or just make them as Windows "reference platforms" to showcase Windows and inspire OEMs to make better PCs. Maybe Microsoft aren't sure, either - - the OEMs pay Microsofts rent by buying licenses, co MS can't afford to tick them off by competing with them too much.

I did have a Surface Book briefly and it was the closest thing I've seen to the Mac 'experience'. Then it died due to a major design fault ("Sleep of Death")... so, yeah, Mac experience :-> (sorry, couldn't leave that one hanging...)

(* Seriously, look at the Surface Studio 2 - makes the iMac look like a bucket of spare parts... until you compare the specs)
 
For me "the cloud" means "x86 docker containers" and the day I can't run and produce those is the day I have to begrudgingly move to a Linux machine on x86 hardware.

Cloud is also moving to ARM.
Amazon M6g prices are so attractive based on their SPECJVM score.

ARM could run ARM dockers so it shouldn't be a problem.

And BTW you probably already knows macOS does not support docker cgroups and is in fact running a virtual machine for that.
If your linux dist support multiple uarch there's a chance your docker script is already working on a arm docker server.
 
Cloud is also moving to ARM. Amazon M6g prices are so attractive based on their SPECJVM score.

I can't exactly tell my CTO that we need to upgrade our entire cloud infrastructure so that I can buy a new MacBook. ARM in the cloud exists, but it's still fringe at this time. Not to mention the ecosystem of existing Docker containers that all shamelessly swap and build upon one another via `FROM` lines. It's a big step backwards if you want to be on the forefront of that transition -- assuming it's a transition that actually happens. Perhaps the pricing is attractive because the demand isn't there.
 
  • Like
Reactions: ssgbryan
Game consoles need GPUs and CPUs. NVIDIA does not do CPUs.

Apple may not view x86 as necessary (although there is not evidence of this) but is it even important? It's more important whether potential customers view x86 as necessary (and some of them definitely do).

They view x86 as necessary does not change the fact that x86 isn't necessary.

It's just one CPU instruction set from 40 year ago.

Even Intel them self does not view x86 as necessary as they are moving from full x86 support starting from SaphireRapids (based on road map info from 4 year ago, at that time Intel plan to end full x86 support after icelake).

x86 is a huge burden for modern CPU. 40 year of backward compatibility is useless when our OS can not run 40 year old code already.
[automerge]1587769077[/automerge]
I can't exactly tell my CTO that we need to upgrade our entire cloud infrastructure so that I can buy a new MacBook. ARM in the cloud exists, but it's still fringe at this time. Not to mention the ecosystem of existing Docker containers that all shamelessly swap and build upon one another via `FROM` lines. It's a big step backwards if you want to be on the forefront of that transition -- assuming it's a transition that actually happens. Perhaps the pricing is attractive because the demand isn't there.

Exactly, but I see a huge opportunity here.
Startups can utilize this cheaper price to provide better service to their customer.
And for small company like us we just started to use Amazon Cloud and we are running Java code on it that is highly portable.

And I really do not think docker script will be that much different for a different uarch.
FROM support different platform and most distribution already support ARM as a target. It's just about time when more docker image add their arm64 flavor.

It could be a cloud first and then client change. Most people will start creating ARM docker file on a x86 machine.

And I can always doing docker job remotely with a test server instead of local machine so really do not think that's a huge problem.
 
Last edited:
Game consoles need GPUs and CPUs. NVIDIA does not do CPUs.

Apple may not view x86 as necessary (although there is not evidence of this) but is it even important? It's more important whether potential customers view x86 as necessary (and some of them definitely do).

Nintendo Switch runs on NVIDIA CPU.
Tegra also powering autopilot systems.

Multiplatform game still runs on switch.
 
My point is this: whether using multiple cores or Grand Central, do not overlook the importance of CPU frequency.

AMD’s recent success with Renoir is based not on the number of cores alone, but on how fast the CPUs process threads, particularly sustained speed. Intel CPUs may boost higher initially but they decrease in frequency much more quickly and run at lower sustained frequencies overall (compared to AMD) due to the heat they generate.

We will have to wait and see how Apple’s CPUs perform, but there is currently no benchmark which can directly compare x86 and ARM, not even GeekBench, so please don’t cite it. The lack of direct comparison is covered by other sites.

The biggest problem for Apple is going to be the loss of x86 compatibility. There will be no x86 emulator and there are many of us who need this for various reasons, not just developers.

There is benchmark that can directly compare x86 and ARM.
It's called SPEC.

It's a source code distribution that if you provide the compiler it will run on literally any CPU. Intel officially submit their score to SPEC database. This is a silicon industry standard benchmark. If you do not trust this number then I suggest you to stop talking about this topic and start writing your papers. You gonna earn a lot of money if you can destroy SPEC and bring up a new standard.

A13 is reaching 9900k level single core performance at merely 2.65Ghz by SPECint 2006.
That's almost double the IPC from Skylake and only 20% slower by SPECfp 2006.
Remember this is only an iPhone. If you plug a fan on this chip you can easily bump the frequency.

just a 5%-10% bump will leave Intel's best of the line single core performance behind.

iPad Pro A12X could runs Fortnite at 120fps + medium settings and please check which Intel CPU is needed to run Fortnite at 120fps with lowest settings.

We already seen how well iPad runs Civ6. Civ 6 even enabled PC mod on iPad.
We all know games are single thread bonded but Apple ARM chip will not be the slower part here.
It's Intel being the slower one and base on their 10nm failure they will stuck at this slow speed for the next few years.
 
Last edited:
Bet-hedging is what Microsoft did with Windows for ARM. Look what it got them so far—next to nothing.

If there is a slightest hint that Apple isn’t betting the farm on ARM Macs, Mac users will simply fight over the remaining stock of Intel Macs and then stay away in droves until Apple gives up and starts making Intel Macs again.

To avoid such a scenario, Tim Cook needs to get on stage and make clear that Intel Macs are going away within XX months and not coming back. We Mac users will still fight over the remaining Intel Mac stock, but when it’s gone, we’ll bitch and moan, try the ARM Macs, and hopefully like them enough to stick around in the Apple ecosystem.

This is especially the case because Intel is still a viable platform, unlike m68k and PowerPC in their respective eras. Customers will feel that Apple may take the easy road out and not require people to migrate. Apple needs to be abundantly clear that it’s ARM or the highway.

The tides of change must be forced.
 
Bet-hedging is what Microsoft did with Windows for ARM. Look what it got them so far—next to nothing.

If there is a slightest hint that Apple isn’t betting the farm on ARM Macs, Mac users will simply fight over the remaining stock of Intel Macs and then stay away in droves until Apple gives up and starts making Intel Macs again.

To avoid such a scenario, Tim Cook needs to get on stage and make clear that Intel Macs are going away within XX months and not coming back. We Mac users will still fight over the remaining Intel Mac stock, but when it’s gone, we’ll bitch and moan, try the ARM Macs, and hopefully like them enough to stick around in the Apple ecosystem.

This is especially the case because Intel is still a viable platform, unlike m68k and PowerPC in their respective eras. Customers will feel that Apple may take the easy road out and not require people to migrate. Apple needs to be abundantly clear that it’s ARM or the highway.

The tides of change must be forced.

Well said.

I'll admit I wasn't paying much attention during the PPC transition. But from what I gather... PPC had reached a dead-end and switching to Intel was Apple's obvious (only?) choice.

Intel has certainly had a lot of problems lately... but is Intel today as bad as PPC was back then?

Was anyone else using PPC after Apple left? I think game consoles did, right?
 
Bet-hedging is what Microsoft did with Windows for ARM. Look what it got them so far—next to nothing.

If there is a slightest hint that Apple isn’t betting the farm on ARM Macs, Mac users will simply fight over the remaining stock of Intel Macs and then stay away in droves until Apple gives up and starts making Intel Macs again.

To avoid such a scenario, Tim Cook needs to get on stage and make clear that Intel Macs are going away within XX months and not coming back. We Mac users will still fight over the remaining Intel Mac stock, but when it’s gone, we’ll bitch and moan, try the ARM Macs, and hopefully like them enough to stick around in the Apple ecosystem.

This is especially the case because Intel is still a viable platform, unlike m68k and PowerPC in their respective eras. Customers will feel that Apple may take the easy road out and not require people to migrate. Apple needs to be abundantly clear that it’s ARM or the highway.

The tides of change must be forced.
💯
 
Well said.

I'll admit I wasn't paying much attention during the PPC transition. But from what I gather... PPC had reached a dead-end and switching to Intel was Apple's obvious (only?) choice.

Intel has certainly had a lot of problems lately... but is Intel today as bad as PPC was back then?

Was anyone else using PPC after Apple left? I think game consoles did, right?

Intel today isn't as bad as PPC was.
IBM basically stopped the development of PPC and left sony/apple alone.

But if you refer to performance then yes, Intel is in bad shape. We almost go back to the Power Mac G5's water cooling solutions. All current high end Intel CPU requires water cooling to fully utilize their boost performance. It's just Apple don't care about their peak performance and favor overall design.

Even the worst Pentium 4 wasn't go beyond 200W but remember that was exactly the time Apple shift to Intel.
Apple should already know what Intel could offer in the next few years and if they decide they have to switch then it marks Intel in a really bad position now.
 
No. The great majority of applications written today are not written for a specific processor architecture. They're written to support certain operating systems, frameworks, libraries, etc. For most applications, the transition will most likely be as simple as recompiling the software. For example, I work on a large-scale software project intended to run on x86 servers however our demo platform is a bunch of ARM boards running a different operating system than what most of us develop on; we did virtually 0 work to support this, for the most part we just recompiled.

Developers have been writing cross-platform compatible software, using cross-compilers, etc for decades. Some folks here are blowing this way out of proportion (or they just have no idea how software is written today). Yes, there will be a transition period, however, for the great majority of Mac users this will be no big deal.

Another thing that will help this is that Macs are quite popular as developer platforms, including for software that's not intended to run on Macs in production. Those developers, like me, want to use a Mac and we want software to work on it. I can see a lot of developers jumping on this and working to make sure stuff works.

Yes, there will be issues like "how do I run my x86 OS in a VM" and whatnot but those will be solved with time. Aside from those issues, I think most people here are not nearly as tied to x86 as they think they are.

But if porting from x86 to ARM is generally that straightforward (referring to your first paragraph), why has MS, with all its understanding of the program, not been able to easily produce a version of Office that runs natively on its ARM-based Surface Pro X? Currently it's limited to running a 32-bit version of Office on an emulator. Would be interested to hear an explanation, plus your thoughts on the implications of this for seeing a native version of Office for ARM (not to mention Adobe CC for ARM) on the Mac.

I can understand how simpler software could be ported straightforwardly, but many of the most important productivity programs (like those in Office and CC) are big and complex. The installed files sizes of Word, Outlook, Excel, and Powerpoint, for instance, average nearly 2 GB each.
 
Last edited:
  • Like
Reactions: chikorita157
But if that's the case (referring to your first paragraph), why has MS, with all its resources, been unable to produce a version of Office that runs natively on its ARM-based Surface Pro X? Currently it's limited to running a 32-bit version of Office on an emulator. Would be interested to hear an explanation, plus your thoughts on the implications of this for seeing a native version of Office for ARM on the Mac.
It’s not that they can’t. It’s that they haven’t. Probably the market share is too small to convince them to focus on it.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.