Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

poorcody

macrumors 65816
Jul 23, 2013
1,316
1,546
The problem isn't cross-platform compilation. The problem is everything I listed in the two large posts above (one and two), such as driving away users by losing features like Windows and virtualization and all old existing Intel software and all hardware peripheral compatibility, and driving away/preventing new developers from even joining in the first place since ARM is a fringe platform. And perhaps above all; the risk of being stuck in a dead end again a la PowerPC.
I guess I'm not being clear with this... for 99% of developers, nothing changes whether the processor is ARM or Intel. They don't care if it's a "fringe" platform because nothing changes for them. They write against the OS/API -- not the processor. Same thing with hardware peripherals -- that is all abstracted to the OS for 99% of developers, and the abstraction is done by Apple (device drivers and compiler writers are the ones affected). Also you can't take a Windows binary and run it on a Mac -- it doesn't matter if they have the same processor. It doesn't work like that -- binaries are not compatible across operating systems. Running Windows and virtualization would require emulation though as you said.
[doublepost=1474412739][/doublepost]
That's single core performance. The Mac Pro has 12 effective cores and 24 VTs. For real world application this will blow any mobile chip. Also note that benchmark performance is no true indication of real world performance. Advanced technologies embedded in the Intel chips really make the difference.
Very true... I think the point is that getting an A10 or later core to match Intel's core is the daunting engineering task... they're probably not there yet, but it seems like they are getting much closer. Scaling up the number of cores is a relatively simple step once you get the core design competitive. Also, one assumes if they used an A-processor in the Mac, it would not be a mobile version but a desktop version that could run at higher speed and deal with more heat.
[doublepost=1474412915][/doublepost]
An A10 that runs x86 would be larger, run hotter, and thus slower.
Oh god yeah, that would just be an A10 with an x86 in it... that would be a disaster. Emulation would run outside the chip for the majority of processor instructions that hardly ever get used. But of course that might just be too slow to be realistic at this stage of performance.
[doublepost=1474413156][/doublepost]
Also note that benchmark performance is no true indication of real world performance. Advanced technologies embedded in the Intel chips really make the difference.
Good point, we don't really know if the performance is really in the league yet. I bet Apple has OS X running on one in the lab to judge just that point... maybe also to make Intel nervous when they are negotiating chip prices. :)
 

fatalogic

macrumors 6502
Aug 16, 2016
251
244
I seriously doubt it would happen but I would like someone to challenge Intel. They are clearly complacent because they have no competition. Someone needs to scare them so they can start really pushing the envelope again, I'm tired of these 5%-10% power bumps per generation.
 
  • Like
Reactions: Shirasaki

leman

macrumors Core
Oct 14, 2008
19,218
19,106
Actually, scrap my previous reply. The A10 is getting scary close to desktop performance. Tom's guide got a score of 170 in JetStream JS benchmark, while my 15" rMBP gets 200. Still a micro benchmark, but much more trustworthy IMO than Geekbench. It seems that Apple almost caught up with Core M performance. If they continue this trend, they might overtake low-power Intel designs in few years.
 

argentum47

macrumors regular
Jun 18, 2014
209
488
Tom's guide got a score of 170 in JetStream JS benchmark, while my 15" rMBP gets 200.
Interesting benchmark.

They are guaranteed to be running the same 'tasks' whatever the architecture is, so I guess, if the difference in the amount of CPU time spent on OS/browser/JS engine (whatever that's not part of the benchmark task) is both minimal, and if the JS engines on both platforms are fully utilizing whatever CPU instruction sets needed for optimization, this could be quite a credible measure of relative performance across different architectures.
 

leman

macrumors Core
Oct 14, 2008
19,218
19,106
Interesting benchmark.

They are guaranteed to be running the same 'tasks' whatever the architecture is, so I guess, if the difference in the amount of CPU time spent on OS/browser/JS engine (whatever that's not part of the benchmark task) is both minimal, and if the JS engines on both platforms are fully utilizing whatever CPU instruction sets needed for optimization, this could be quite a credible measure of relative performance across different architectures.

Of course, the are tons of factors that can influence the results here. Is the JS interpreted or JITed? What kind of JIT technology do they use? etc, etc. However, I like these benchmarks much better because they reflect a more complex scenario. There is a browser, JS code and the code that executes that JS code. So you'll probably get a much more complete measurement of what the machine can do rather than doing a numeric loop which probably does not have any practical significance.

P.S. Results between OS X and iOS should be meaningful, as both platforms use the same JS engine.
 

Freyqq

macrumors 601
Dec 13, 2004
4,038
181
Really, Apple is only having the most problems with Intel's schedule because they are using specialized high-performance iGPU chips that 99% of other PCs don't use.

Plus, as I keep saying, if Apple wanted to release an ARM based the Mac the 12" MacBook would have been the machine to do it. They didn't, which means no ARM chips in the near future.

12" iPad Pro was their attempt at an arm computer. It is marketed as a computer by Apple.
 

Closingracer

macrumors 601
Jul 13, 2010
4,308
1,840
Actually, scrap my previous reply. The A10 is getting scary close to desktop performance. Tom's guide got a score of 170 in JetStream JS benchmark, while my 15" rMBP gets 200. Still a micro benchmark, but much more trustworthy IMO than Geekbench. It seems that Apple almost caught up with Core M performance. If they continue this trend, they might overtake low-power Intel designs in few years.


Core M perhaps but won't ever compete with i5 or i7 that Apple uses.
 

maflynn

macrumors Haswell
May 3, 2009
73,552
43,522
Actually, scrap my previous reply. The A10 is getting scary close to desktop performance
In raw performance it is, but I wonder how it will hold up for highly threaded tasks, I'm not saying it will be slower, but rather I just don't know
 

leman

macrumors Core
Oct 14, 2008
19,218
19,106
Core M perhaps but won't ever compete with i5 or i7 that Apple uses.

In raw performance it is, but I wonder how it will hold up for highly threaded tasks, I'm not saying it will be slower, but rather I just don't know

I absolutely agree that its unlikely that A10 can maintain the performance over longer periods of time. Its just like with Core M and higher-powered Intel CPUs: on burst performance, Core M can hold its own, but once you need to do some real work, the gap widens.

Still, its quite impressive to see this kind of performance from A10. If this design can handle higher heat dissipation, Apple could have something very interesting on their hands.
 

Closingracer

macrumors 601
Jul 13, 2010
4,308
1,840
I absolutely agree that its unlikely that A10 can maintain the performance over longer periods of time. Its just like with Core M and higher-powered Intel CPUs: on burst performance, Core M can hold its own, but once you need to do some real work, the gap widens.

Still, its quite impressive to see this kind of performance from A10. If this design can handle higher heat dissipation, Apple could have something very interesting on their hands.


Not really if you look at Intel optimization and compatiblty with most software.
 
  • Like
Reactions: SteveJobzniak

leman

macrumors Core
Oct 14, 2008
19,218
19,106
Not really if you look at Intel optimization and compatiblty with most software.

I don't think that is a big problem. Apple already demonstrated that is is able to handle architectural transitions just fine. In the last years we had transition from PPC to x86, from x86 to x86-64 and from AArch32 to AArch64.

BTW, there is some new stuff in OS X kernel that suggests that Apple is at least playing around with an ARM Mac.
 
  • Like
Reactions: poorcody

Closingracer

macrumors 601
Jul 13, 2010
4,308
1,840
I don't think that is a big problem. Apple already demonstrated that is is able to handle architectural transitions just fine. In the last years we had transition from PPC to x86, from x86 to x86-64 and from AArch32 to AArch64.

BTW, there is some new stuff in OS X kernel that suggests that Apple is at least playing around with an ARM Mac.



Lol that is a big problem..... It's worse when Apple's A10 doesn't run half of what intels cpu runs which allows Intel to have one of the best cpu's in the consumer market. Power pc to Intel was a smart move but Apple would be stupid to go from intel to ARM. Apple can't dictate where the market should go unlike their phone. If Apple decides to move to ARM the countdown to extinction will start.

I for one is a user who prefers windows already and only got a Macbook Pro because of better battery life and can run the apps I desire. You take away compatiblty I would rather carry my charger and go back to Windows for school. I would bet there will be plenty more like me who would switch to Windows....or perhaps Linux should Apple pull a stunt like this.
 

Freyqq

macrumors 601
Dec 13, 2004
4,038
181
I don't think they would just replace the Intel chips with ARM. I think they would create a 13 and 15" iPad that docks to a keyboard an trackpad, like a Surface Book. When docked, it is running an OSX UI. When in tablet mode, it runs iOS as a UI. However, this is just a UI and both are sharing the same underlying code and file system. Apps would have 2 UI modes, touch and mouse/keyboard. Once they have a few revisions of that, they would probably stop updating the mac line for a couple years and then drop the whole line.

Replacing it all cold-turkey, without giving developers a couple years lead-time to get apps in place, would be a disaster. Either way, I don't think dropping Intel would be prudent, and I would not buy a mac without virtualbox windows support.
 

mfram

Contributor
Jan 23, 2010
1,311
353
San Diego, CA USA
I could see an argument to create Macs with ARM chips. What if you're a user that doesn't care about PC VMs or Windows support? I don't use either of those features. I would only be using native macOS applications. Having more battery life would a very nice thing.

But it wouldn't necessarily be that easy. For at least some length of time, you'd want to run applications compiled for Intel. Which means you'd need some kind of binary translation. It's possible, but would reduce performance even more. And there are probably a class of applications (games for example) that would want to run as Intel-only for porting purposes.

Nonetheless, a Mac designed for portability over performance (MacBook, maybe MacBook Air) with ARM-only support might be an option for some consumers. Give consumers the choice.
 

maflynn

macrumors Haswell
May 3, 2009
73,552
43,522
I could see an argument to create Macs with ARM chips. What if you're a user that doesn't care about PC VMs or Windows support? I don't use either of those features. I would only be using native macOS applications. Having more battery life would a very nice thing.
Microsoft tried that very same thing with Windows RT, and we know how that went over - a thud.

I see no difference if Apple tried that, either by having a cheaper ARM based model, or completely shifting over to ARM.

you'd want to run applications compiled for Intel. Which means you'd need some kind of binary translation
Its an emulation layer and it's almost always going to be slower then running it natively and there can be compatibility issues, i.e., some apps don't run as well.

As for other Apps, would Microsoft just transition over their weaker, less featured MS Office for iOS to run on ARM? I wouldn't be surprised, and of course other manufacturers would have to commit to investing money to re-work their apps, as I said before, changing platforms isn't as easy as just recompiling, even if the app was using Apple's APIs. There's always different gotchas that need to be addressed.
 
  • Like
Reactions: SteveJobzniak

Count Blah

macrumors 68040
Jan 6, 2004
3,192
2,748
US of A
I could see A chips being used in a hybrid iPad and Macbook product, that could switch in between iOS and Mac OS. But it would be scary to think about power users losing x86 for the higher end Macs.
Apple's moves, or lack there of, in the computer division, has shown us that Apple no longer cares about Power Users. They would rather the trendy and non-techincally savvy users, be their customers now. But even the trendy users, think marketing types, need to be able to switch back and forth between windows and OS X(refuse to say the new name for a couple years).

Removing that capability would see Apple's market share dive way back into the < 5% range.
[doublepost=1474726854][/doublepost]
Apple doesn't need to create new hardware to run OS X on an ARM based machine. They already have that: Imagine an iPad Pro running OS X. We already have the keyboard and mouse support, too. The mouse just only works in selected RDP apps (like Jump Desktop). With this, the iPad Pro could indeed replace a real notebook PC. If they keep compatibility with current iPad apps we would have a really great and practical SP4 Pro replacement without the Windows handicap.
And without the ability to run windows applications. Much like the 3.5mm jack, just because YOU don't use it, doesn't mean others don't as well.

I don't switch to windows bootcamp very often, but when I do, I REALLY need to. There is no way I'm buying an ARM based Mac, if I can't boot into windows when needed. And I guarantee you that I am not a unique case. I also advise quite a few family members concerning their computer purchases. And occasionally booting into Windows for work purposes, plays a part in their requirements. Apple moving to ARM, without running windows, is a non-sale for them.

Apple obviously doesn't care about the user base it built in the late 90s nor 00s. Apple cares about flash and trendy. Given that, I can easily see them making this move. But I can also see them losing even MORE customers due to alienation. But that does seems to be Apple's move these days.
 
  • Like
Reactions: SteveJobzniak

Fishrrman

macrumors Penryn
Feb 20, 2009
28,395
12,519
If Apple doesn't seem particularly interested in devoting the R&D into keeping its hardware line up-to-date, why would they invest much in a complete redesign of the -software-?

My guess is that they'll "keep on coastin'" with an Intel-based OS until they drop the MacOS line altogether in about 7-10 more years… (if not sooner)
 
  • Like
Reactions: SteveJobzniak

Count Blah

macrumors 68040
Jan 6, 2004
3,192
2,748
US of A
If Apple doesn't seem particularly interested in devoting the R&D into keeping its hardware line up-to-date, why would they invest much in a complete redesign of the -software-?

My guess is that they'll "keep on coastin'" with an Intel-based OS until they drop the MacOS line altogether in about 7-10 more years… (if not sooner)
iOS apps have to be written SOMEWHERE?!?!?!

With so many cross platform development tools, it would not shock me that Apple would abandon iOS development on it's own hardware.

Apple has been known to do extremely silly things like this in the past. Back in the late 90s, when apple acquired Next, Apple gave their developers windows machines to develop in WebObjects. Yes, their Sales and Consulting engineers would go to sales calls and to customer projects, toting windows machines. What was old, is new again.

SMH
 

leman

macrumors Core
Oct 14, 2008
19,218
19,106
Microsoft tried that very same thing with Windows RT, and we know how that went over - a thud.

The question is, how much of that was because of the ARM architecture and how much because of the design of Windows RT. What we are talking about here, is a full fledged OS X - just as it is now — just running on an ARM CPU.

Its an emulation layer and it's almost always going to be slower then running it natively and there can be compatibility issues, i.e., some apps don't run as well.

There is a decent chance that you don't even need en equation layer. The basic data size and alignment are the same between ARM64 and x86-64. A compiler from Intel binary to ARM binary should be able to produce a reasonable result.
 

maflynn

macrumors Haswell
May 3, 2009
73,552
43,522
The question is, how much of that was because of the ARM architecture and how much because of the design of Windows RT. What we are talking about here, is a full fledged OS X - just as it is now — just running on an ARM CPU.
My take is that there was a general confusion by the consumer. Also, Windows RT was a full fledged OS that was compiled to ARM, there were no native apps (other then Office) and an ARM based OS X will be the same thing. X86 apps won't run natively.
 
  • Like
Reactions: SteveJobzniak

poorcody

macrumors 65816
Jul 23, 2013
1,316
1,546
Also, Windows RT was a full fledged OS that was compiled to ARM, there were no native apps (other then Office) and an ARM based OS X will be the same thing.
But macOS has a huge advantage here: Windows apps use the Win32 API, which is heavily tied to the processor -- developers had to make changes to their apps to work on RT. macOS, on the other hand, has virtually no API-processor dependencies, so developers could literally just recompile their apps. That's what happened when OS X went from PPC to Intel.
 

maflynn

macrumors Haswell
May 3, 2009
73,552
43,522
so developers could literally just recompile their apps. That's what happened when OS X went from PPC to Intel.
I remember the transition from PPC to Intel, and I don't recall it being very fast or smooth. The transition to Intel took years.
 
  • Like
Reactions: SteveJobzniak

leman

macrumors Core
Oct 14, 2008
19,218
19,106
My take is that there was a general confusion by the consumer. Also, Windows RT was a full fledged OS that was compiled to ARM, there were no native apps (other then Office) and an ARM based OS X will be the same thing. X86 apps won't run natively.

But it was still a different flavour of OS, and the APIs it used were not exactly the same as the 'normal' windows was using. Thats exactly what I am saying: an ARM OS X would be OS X, but just running on ARM. Most apps can be recompiled with a single button press, others of course will need more deep modifications. Still, thats a very different situation than moving to what essentially is a different platform.
 

maflynn

macrumors Haswell
May 3, 2009
73,552
43,522
ARM OS X would be OS X, but just running on ARM.
That's what Windows RT - Windows running on ARM and that failed.

Most apps can be recompiled with a single button press, others of course will need more deep modifications.
Are you sure about that? Do you know that all source code is so easily compiled to a different platform?

Even if you're right, Developers would need to take the time and energy to validate and test. Time = money. You already stated that the transition from PPC to intel was simple and seamless but it wasn't.

Still, thats a very different situation than moving to what essentially is a different platform.
So you're saying ARM and Intel are essentially the same plarform :confused:
 
  • Like
Reactions: SteveJobzniak
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.