Discussion Quo vadis Apple Macs - Intel, AMD and/or ARM CPUs? ARM it is!

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

moinmoin

Diamond Member
Jun 1, 2017
4,952
7,663
136
Due to popular demand I thought somebody should start a proper thread on this pervasive topic. So why not do it myself? ;)

For nearly a decade now Apple has treated their line of Mac laptops, AIOs and Pro workstations more of a stepchild. Their iOS line of products have surpassed it in market size and profit. Their dedicated Mac hardware group was dissolved. Hardware and software updates has been lackluster.

But for Intel Apple clearly is still a major customer, still offering custom chips not to be had outside of Apple products. Clearly Intel is eager to at all costs keep Apple as a major showcase customer.

On the high end of performance Apple's few efforts to create technological impressive products using Intel parts increasingly fall flat. The 3rd gen of MacPros going up to 28 cores could have wowed the audience in earlier years, but when launched in 2019 it already faced 32 core Threadripper/Epyc parts, with 64 core updates of them already on the horizon. A similar fate appears to be coming for the laptops as well, with Ryzen Mobile 4000 besting comparable Intel solutions across the board, with run of the mill OEMs bound to surpass Apple products in battery life. A switch to AMD shouldn't even be a big step considering Apple already has a close work relationship with them, sourcing custom GPUs from them like they do with CPUs from Intel.

On the low end Apple is pushing iPadOS into becoming a workable mutitasking system, with decent keyboard and, most recently, mouse support. Considering the much bigger audience familiar with the iOS mobile interface and App Store, it may make sense to eventually offer a laptop form factor using the already tweaked iPadOS.

By the look of all things Apple Mac products are due to continue stagnating. But just like for Intel, the status quo for Mac products feels increasingly untenable.
 
  • Like
Reactions: Vattila

Glo.

Diamond Member
Apr 25, 2015
5,711
4,556
136
It does make sense, as it would be appropriate for the bulk of the Macs sold.
For the very reason I do not buy it.

And you know what fits that bill? iPad's SoC's.

In essence, people believe that Apple will tape out three SoCs with next round: A14, A14X and the SoC that would go into ARM Macs?

Nah, I do not buy it...
 

Ajay

Lifer
Jan 8, 2001
15,454
7,862
136
For the very reason I do not buy it.

And you know what fits that bill? iPad's SoC's.

In essence, people believe that Apple will tape out three SoCs with next round: A14, A14X and the SoC that would go into ARM Macs?

Nah, I do not buy it...
FWIW, they do have the money and talent to do so. Last I read, they had three teams running (I suspect one was implementation, bring up manufacturing with TSMC).
I'll be surprised if Apple comes out with higher end ARM Mac and with only Kuo knowing about it. With the number of ISVs involved, that project would leak like a sieve - despite Apple's KGB level of secrecy.
 

beginner99

Diamond Member
Jun 2, 2009
5,210
1,580
136
Good deflection. Unfortunately its not a speculation. Apple HAS to port FCPX and Logic Pro to ARM, first, if they want to move Macs to ARM from top to bottom.

Why? Because both of those Apps are inherent part of Apple ecosystem.

Its that simple.

While I agree with you that apple won't switch over completely any time soon (half year old macpro....), this argument doesn't make much sense. Why would they port a desktop app to a tablet? That would need a completely different UI. Makes no sense and they could have ARM builds of these apps running internally on test platforms.
 

mikegg

Golden Member
Jan 30, 2010
1,756
411
136
IMO,

I don't think that Apple will switch (as in replace everything) to ARM like they did with PowerPC.

There was no real alternative at the time, that's why they had to move to x86, but now with the "new" AMD and a "motivated" Intel, it would be a really bad idea to jump the ship.

Apple will certainly continue to evolve their ARM cores for iPhone/iPad and even try to add them into other products like a Macbook Air.

But an ARM iMac or Mini? I don't think so.
Apple does not want to rely on AMD & Intel, just like how they're not relying on Qualcomm or Samsung for mobile SoCs. Apple is fully capable of competing with AMD & Intel in performance.

More importantly, relying on AMD & Intel means PC makers will always be able to match Apple in features and performance. Apple is all about differentiating.

There is no guaranteed that AMD won't sh*t the bed in the future as well.

PS. I'm want Apple to switch to AMD but it's unlikely. Apple wants to unify all their apps and differentiate their offerings. Best way to do that is to ship their own ARM laptop/desktop CPUs, which they are extremely capable of doing.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,607
5,799
136
Apple may be very profitable, just their Airpods business is bigger than many Semi design houses out there.
However, only thing that I hear when people talk about Apple is iPhone or Mac, frankly, that is quite boring for such a massive company.
Apple could have been different under Steve, whom when I think of, I always recollect his Xerox story which is exactly what Apple is these days.

Just from the top of my head...

Behind the industry in AI, ML, HPC initiatives. Difficult to catch up with the likes of Google, Amazon, Facebook, MS etc who in lots of cases are working together, see ONNX
No widely adopted content delivery services, Ad delivery, social networking platform like Twitch, YouTube, Mixer, Twitter etc. FB and MS big winners here.
Behind the industry in digital transformation solutions.
Sitting out of the resurgence in cloud computing. Big gains for Alibaba, Tencent, Azure, AWS. Looking at those guys they made colossal revenues lately.
Sitting out of the Gaming/Cloud gaming industry (This industry is bigger than the movie industry)
Sitting out of the recent growth in Automotive Technology, Google made it big here.
Sitting out of Edge Computing.


Can someone help me list some things outside of iPhone/Mac which Apple have a major impact on the industry?
 
Last edited:

ThatBuzzkiller

Golden Member
Nov 14, 2014
1,120
260
136
While I agree with you that apple won't switch over completely any time soon (half year old macpro....), this argument doesn't make much sense. Why would they port a desktop app to a tablet? That would need a completely different UI. Makes no sense and they could have ARM builds of these apps running internally on test platforms.

Apple has no choice but to provide provide those desktop applications because in several cases they are absolutely necessary for certain demographics like professionals or for higher productivity groups.

How can people somehow entertain the thought that ARM will replace x86 for Macs when Xcode doesn't even run on iOS at all which are primarily for ARM based systems ? Also for an IDE, I haven't heard of many positive experiences with Xcode and based off of the anecdotes from the developers I know you could even say it's terrible. It's unacceptable to see that the only real option to develop for iOS is in that state. Another massively growing field is machine learning and data scientists use very popular frameworks like Tensorflow or PyTorch to train their neural network models but it's not available on iOS systems either. There is so much more that iOS lacks in terms of productivity applications like video editing software, numerical computing, scientific/engineering simulation, computer aided design, and professional rendering/architectural visualization.

Unless someone wanted to make their jobs harder nobody would dare contemplate replacing an x86 system for an ARM system. A worth of a device is not just based on it's entertainment factor which is ironic since Macs aren't even good for high-end gaming as well but it's also about offering more productivity too since at the end of the day there are people who actually use computers to do real work at which their careers depend on them.

Apple over the years has started to become more of a hardware company rather than a software company and this change in direction has had detrimental effects. Apple has become more out of touch with the architects, artists, editors, data scientists, engineers, programmers and other technical staff than they were before. If Apple does plan to go through with this it's no wonder why they are criticized for being anti-developer or anti-professional ...
 
  • Like
Reactions: Tlh97 and Carfax83

dacostafilipe

Senior member
Oct 10, 2013
772
244
116
Apple does not want to rely on AMD & Intel, just like how they're not relying on Qualcomm or Samsung for mobile SoCs. Apple is fully capable of competing with AMD & Intel in performance.

More importantly, relying on AMD & Intel means PC makers will always be able to match Apple in features and performance. Apple is all about differentiating.

There's no reason for Apple to create their own stuff if they only can "compete" with the alternatives. Apple does not produce their own SoC because they are on par with the other manufactures, no, it's because Apple just outperforms them every single time.

So, if they can produce an ARM SoC that can outperform the x86 guys in all areas with a big enough margin, then we can talk about a possible switch, but until then, I see no reason to believe in a near future iMac ARM version.

There is no guaranteed that AMD won't sh*t the bed in the future as well.

That's why having Intel and AMD compete against each-other is such a great news for Apple! They can move to the x86 guy with the best platform if needed. AMD continues to deliver? Move parts of the production to them. Intel somehow create a +25% IPC wonder-core? Blue team it is.

Changing hardware is easy. Software is the issue. I remember Rosetta!



PS: Obviously, everything is IMHO.
 

moinmoin

Diamond Member
Jun 1, 2017
4,952
7,663
136
Apple may be very profitable, just their Airpods business is bigger than many Semi design houses out there.
However, only thing that I hear when people talk about Apple is iPhone or Mac, frankly, that is quite boring for such a massive company.
Apple could have been different under Steve, whom when I think of, I always recollect his Xerox story which is exactly what Apple is these days.

Just from the top of my head...

Behind the industry in AI, ML, HPC initiatives. Difficult to catch up with the likes of Google, Amazon, Facebook, MS etc who in lots of cases are working together, see ONNX
No widely adopted content delivery services, Ad delivery, social networking platform like Twitch, YouTube, Mixer, Twitter etc. FB and MS big winners here.
Behind the industry in digital transformation solutions.
Sitting out of the resurgence in cloud computing. Big gains for Alibaba, Tencent, Azure, AWS. Looking at those guys they made colossal revenues lately.
Sitting out of the Gaming/Cloud gaming industry (This industry is bigger than the movie industry)
Sitting out of the recent growth in Automotive Technology, Google made it big here.
Sitting out of Edge Computing.


Can someone help me list some things outside of iPhone/Mac which Apple have a major impact on the industry?
Services already is Apple's second biggest category for net sales after iPhone, with strong growth projected. I think you're mistaken Apple's focus on consumer products with not having a hand in those areas, they don't sell servers anymore so of course they are neither involved in any development that builds on that. Consumer oriented AI and ML is part of their iPhone development (I don't care for this stuff at all, but Apple was instrumental in popularizing the use of fingerprint sensors, voice as well as face recognition in consumer products). Their iCloud service is one of the bigger customers of AWS and Azure, and Apple's effort to build its own datacenters doesn't appear to bear many fruits so far. Apple Arcade is together with Xbox Game Pass one of only two big Netflix like subscription based game services. Regarding Automotive Technology, how did Google in any way made it big there? Apple had its own car research efforts going nowhere for years so it appears it was recently shuttered. Edge Computing imo is just a fancy name for the regular wave back and forth between centralization (Cloud Computing) and localization (Client Computing).
 
  • Like
Reactions: Glo.

Richie Rich

Senior member
Jul 28, 2019
470
229
76
So, if they can produce an ARM SoC that can outperform the x86 guys in all areas with a big enough margin, then we can talk about a possible switch, but until then, I see no reason to believe in a near future iMac ARM version.
Take a look at almost two years old A12X, despite having only 4 big cores it's beating brand new AMD 6-core Renoir 4600U everywhere: Look at Geekbench5:
  • ST score: 4600U (1082 pts) vs. A12X (1111 pts)
  • MT score: 4600U (3975 pts) vs. A12X (4577 pts)

Due to having only 7W TDP, Apple can deploy within 15W TDP double core count and will have much higher MT performance even with full emulation hit (about 0.6-0.7). This is two year old chip. Compare the performance in 2018 against Zen1 - that's massive advantage for Apple. So performance wise Apple was ready in 2018. Apple is waiting for something else. Maybe SW, maybe new 2048-bit SVE2 uarch, who knows...

 

Glo.

Diamond Member
Apr 25, 2015
5,711
4,556
136
Take a look at almost two years old A12X, despite having only 4 big cores it's beating brand new AMD 6-core Renoir 4600U everywhere: Look at Geekbench5:
  • ST score: 4600U (1082 pts) vs. A12X (1111 pts)
  • MT score: 4600U (3975 pts) vs. A12X (4577 pts)

Due to having only 7W TDP, Apple can deploy within 15W TDP double core count and will have much higher MT performance even with full emulation hit (about 0.6-0.7). This is two year old chip. Compare the performance in 2018 against Zen1 - that's massive advantage for Apple. So performance wise Apple was ready in 2018. Apple is waiting for something else. Maybe SW, maybe new 2048-bit SVE2 uarch, who knows...

Have you taken into account differences in platforms? What if we will put iOS on AMD platform and it will outperform Apple ARM CPU?

That is what I don't understand. In all of those comparisons, why nobody takes into account how iOS is optimized, and how it yields performance?
 

dacostafilipe

Senior member
Oct 10, 2013
772
244
116
Take a look at almost two years old A12X, despite having only 4 big cores it's beating brand new AMD 6-core Renoir 4600U everywhere

Well, if you look at the more general purpose design of Zen2 and the amount of cores/caches/GPU cores, I would say that the 4600U looks a lot better here. Specially if you know that they use the same node!

A12X 122mm2 - Renoir 150mm2
 

DisEnchantment

Golden Member
Mar 3, 2017
1,607
5,799
136
Consumer oriented AI and ML is part of their iPhone development (I don't care for this stuff at all, but Apple was .
Sorry but consumer AI is a very tiny drop compared to what is going on in the cloud and enterprise ML. In terms of complexity, investment, scale, deployment, scope, scenarios, and whatever metrics you want to use.
I can just write a classification app with 5 mins of code in ML.NET and call it ML but that is not what the industry is all about.

Regarding Automotive Technology, how did Google in any way made it big there? Apple had its own car research efforts going nowhere for years so it appears it was recently shuttered. Edge Computing imo is just a fancy name for the regular wave back and forth between centralization (Cloud Computing) and localization (Client Computing).
Android Automotive and GAS is going to be de facto standard. Even GENIVI is trying to integrate with Android Automotive. GAS is the only way you can have a substantial app market place in Automotive, in China with Baidu apps you can only do Android. Then add LBS , offboard Navigation, you name it.
Classical power train is a dead end, hence things like the VW MEB platform and MS collaboration with VW. Google went the other route to gain pervasiveness in Auto Industry and it paid off.

As for your other points, like I said nothing major outside of their iPhone/Mac ecosystem (the iPhone did change the industry when it came out but that was 10 years ago). For example I can quickly list how MS/Google/FB/Amazon changed the landscape
AWS and Azure radically changed the way Web Deployment happens, and everyday is a better experience. Compared to just 10 years ago it is incredible how easy it is right to just create and app and deploy, then scale. Seamless DB integration, scaling and Service downtime monitoring, migration... From AKS, geo-replication, serverless architectures and what not.
MS/Google/FB etc with their Caffe, TensorFlow, ONNX, CNTK changing how data is consumed and processed and making it in such a way that it is accessible to everyone. Every Tom Dick and Harry can use faceswap and play with AI.
GAS changed the Auto landscape especially how fast you can deliver content to vehicles and merge the incumbent marketshare from OEMs with the market place and enabling use cases which in the past have only been available from top OEMs.
xCloud, Stadia, Streaming platforms (Twitch/Mixer/YT) are innovations on their own which changed the gaming industry
Multiple neural network algorightms which enabled widespread training of models to enable fancy use case like autonomous systems to be deployable on scale
...
 

Hitman928

Diamond Member
Apr 15, 2012
5,285
7,915
136
Take a look at almost two years old A12X, despite having only 4 big cores it's beating brand new AMD 6-core Renoir 4600U everywhere: Look at Geekbench5:
  • ST score: 4600U (1082 pts) vs. A12X (1111 pts)
  • MT score: 4600U (3975 pts) vs. A12X (4577 pts)

Due to having only 7W TDP, Apple can deploy within 15W TDP double core count and will have much higher MT performance even with full emulation hit (about 0.6-0.7). This is two year old chip. Compare the performance in 2018 against Zen1 - that's massive advantage for Apple. So performance wise Apple was ready in 2018. Apple is waiting for something else. Maybe SW, maybe new 2048-bit SVE2 uarch, who knows...


I'm pretty sure that MT score for the 4600u is wrong. The scaling is terrible and doesn't line up with the scores published for the 4700u or 4800u.


4700u Single: 1133 Multi: 6361


4800u Single: 1150 Multi: 7047
 
  • Like
Reactions: Tlh97

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Take a look at almost two years old A12X, despite having only 4 big cores it's beating brand new AMD 6-core Renoir 4600U everywhere: Look at Geekbench5:
  • ST score: 4600U (1082 pts) vs. A12X (1111 pts)
  • MT score: 4600U (3975 pts) vs. A12X (4577 pts)

Due to having only 7W TDP, Apple can deploy within 15W TDP double core count and will have much higher MT performance even with full emulation hit (about 0.6-0.7). This is two year old chip. Compare the performance in 2018 against Zen1 - that's massive advantage for Apple. So performance wise Apple was ready in 2018. Apple is waiting for something else. Maybe SW, maybe new 2048-bit SVE2 uarch, who knows...

I think it's a little disingenuous to compare what is as best I can tell an unsourced/unverified GB5 run on CPU-Monkey.com to a verified Geekbench index result. And also comparing a 4+4 core to a 6 core. Of course the 8 core chip will win, because for simple tasks like those in Geekbench, even small efficient cores help substantially.

Also, if you think Xcode's hardware-specific compile on Clang/LLVM is === to Clang/LLVM on Linux or Windows I have a really nice beachfront property in Mongolia I'd like to sell you.

Regardless, as it pertains to these self-proclaimed "real world" benchmarks, specifically Geekbench, here's an interesting breakdown:

But llvm or gcc is not even close to complex UI loads. If you think the llvm I$ miss rates or branch prediction numbers look bad, I have a bridge to sell you. Those are still really good low miss rates. They look high only because you compare to some silly trivial benchmark that just has one single loop.

Modern GUI toolkits really do nasty things to a CPU in a way that even a "complex" load like a compiler written in C++ doesn't even come close to.

A compiler may have some OS abstractions, and various abstractions for optimization passes and particular optimizations (and for hw descriptions etc), but on the whole it's fairly core system code that doesn't do anything actively odd.

A GUI app ends up having abstraction upon abstraction (often through several layers of toolkit libraries) and event loops with signals going hither and thither and moving a mouse pointer or touching the screen can cause millions of instructions to be executed with nary a loop in sight because you just have those things calling each others or causing other events to be created and then you have more abstractions to actually paint and update the end result...

I$ misses do matter, but most simple benchmarks don't even begin to scratch the surface.

By the way, that's not some random dude making the above quote. It's Linus Torvalds.

In the end, these benchmarks are fun to talk about but don't serve much real world purpose. Wake me up when we start getting real benchmarks for real apps comparing A12X and 4600U.
 

Doug S

Platinum Member
Feb 8, 2020
2,263
3,515
136
For the very reason I do not buy it.

And you know what fits that bill? iPad's SoC's.

In essence, people believe that Apple will tape out three SoCs with next round: A14, A14X and the SoC that would go into ARM Macs?

Nah, I do not buy it...


They could just do two, and use the A14X in both iPad and Mac.

Though Apple can easily afford to design a third chip line for the Mac. They do a second for the iPad, so why it is so hard to imagine they'd do a third for the Mac which sells ~20 million units a year? A lot of the cost would be the third mask set, it isn't like they'd be designing a whole new CPU core for the Mac.

Though I still believe they will use the A14X in the Mac. They just need to add a few bits for the Mac like ethernet controllers or whatever that wouldn't be used on an iPad (and of course some of the iPad blocks won't be used on Mac) but in a 5nm design their transistor budget is so ridiculous that a few mm^2 of "wasted" die space is almost irrelevant.

Heck, they could include 4 more big cores that are only used on the Mac for probably less than 10% die size increase and that could conceivably pay for itself via binning (i.e. dies with a bad CPU core get used in iPads) Based on the A12X vs A12Z GPU they are already thinking along these lines so it isn't so hard to imagine leveraging the Mac vs iPad difference to increase usable yield.
 

Ajay

Lifer
Jan 8, 2001
15,454
7,862
136
They could just do two, and use the A14X in both iPad and Mac.

Though Apple can easily afford to design a third chip line for the Mac. They do a second for the iPad, so why it is so hard to imagine they'd do a third for the Mac which sells ~20 million units a year? A lot of the cost would be the third mask set, it isn't like they'd be designing a whole new CPU core for the Mac.

Though I still believe they will use the A14X in the Mac. They just need to add a few bits for the Mac like ethernet controllers or whatever that wouldn't be used on an iPad (and of course some of the iPad blocks won't be used on Mac) but in a 5nm design their transistor budget is so ridiculous that a few mm^2 of "wasted" die space is almost irrelevant.

Heck, they could include 4 more big cores that are only used on the Mac for probably less than 10% die size increase and that could conceivably pay for itself via binning (i.e. dies with a bad CPU core get used in iPads) Based on the A12X vs A12Z GPU they are already thinking along these lines so it isn't so hard to imagine leveraging the Mac vs iPad difference to increase usable yield.
I think Apple could do a third SoC for MacBooks/iMacs - if that's the way they are going. If they are, they already hired on more engineers, I would suspect. While design time could be cut down by using shared standard cell libraries, that may not work if Apple chooses to make an ARM CPU using TSMC's HP process node. That would be a whole new ball of wax. More large cores would clearly be needed, and that would mean some significant changes to the chip (front end, backend, execution, caches, etc) to deliver sustained higher performance - IMHO. Well, everything in there is IMO.
 

Thala

Golden Member
Nov 12, 2014
1,355
653
136
Good deflection. Unfortunately its not a speculation. Apple HAS to port FCPX and Logic Pro to ARM, first, if they want to move Macs to ARM from top to bottom.

Why? Because both of those Apps are inherent part of Apple ecosystem.

Its that simple.

Look, you have to understand what porting to ARM means. Porting to ARM does not mean porting from MacOS to iOS. In fact within the same OS environemnt porting is little more than just recompiling - full stop. Porting from MacOS to iOS however is not trivial - independent of CPU architecture.
In conclusion you are dead wrong with your assumption that Apple would need to port to iOS first.

Or look at the follwing example. I did port dozens of Win32 apps to ARM64 - where i had to change none or less than very very tiny fraction of the code base. However porting the same app to say MacOS would be much much bigger effort even if i assume x86 for both targets.
As a matter of fact the compiler does the heavy lifting if you just want to port an app within the same operating environment for different CPU architecture targets.
 

Thala

Golden Member
Nov 12, 2014
1,355
653
136
Have you taken into account differences in platforms? What if we will put iOS on AMD platform and it will outperform Apple ARM CPU?

That is what I don't understand. In all of those comparisons, why nobody takes into account how iOS is optimized, and how it yields performance?

OS has no or very little impact on low level benchmarks like geekbench or SPEC. Hope you understand now.
 
  • Like
Reactions: Etain05

Hitman928

Diamond Member
Apr 15, 2012
5,285
7,915
136
OS has no or very little impact on low level benchmarks like geekbench or SPEC. Hope you understand now.

Geekbench scores better on Linux than on Windows. It's not a huge difference but it is significant enough to note. Phoronix found Linux to give ~10% higher scores than Windows. From my own tests I've seen 10 - 15% from Windows to Ubuntu and Suse. They use a different compiler between the OSs so there's a high likelihood of a performance delta just from that.

 

Thala

Golden Member
Nov 12, 2014
1,355
653
136
Geekbench scores better on Linux than on Windows. It's not a huge difference but it is significant enough to note. Phoronix found Linux to give ~10% higher scores than Windows. From my own tests I've seen 10 - 15% from Windows to Ubuntu and Suse. They use a different compiler between the OSs so there's a high likelihood of a performance delta just from that.


Indeed its from the compiler - as i said, the OS has minimal impact. Use the same compiler and you should have about the same scores. The ABI for AArch64 in Linux and Windows is slightly different as well - so the generated code might have different register allocation even when using the same compiler.