ARM MacBook Pro I don’t want an Overpriced Netbook.

[Bustermd]

Look,

Until the "Apple Silicon" Mac is in someones possession in actual real life use, we will never know the true performance. Its all up in the air at this point and we don't even have any confirmed specs. Benchmark comparisons between iphone and ipad SOCs to Intel Chips are not that helpful frankly as they doesn't automatically translate into better real life performance, which depends a lot on what apps are running and how they are optimized (or not) for the specific CPU. Also, you HAVE TO factor in thermals. There is a big difference running in a "walled garden" on a phone versus a laptop or desktop where there are many more variables, multitasking, and general much greater demands on the chip. It still remains to be seen if Apple Silicon will run cooler at the same level of processing power. To claim otherwise at this point is premature. To put things in perspective: my iphone 11 Pro gets uncomfortably hot in 10-15 minutes if I'm using it for a Zoom meeting with video enabled. It also gets noticeably warm when running wireless CarPlay and streaming from Spotify, to the point where I've had to take it out of my pocket. Point being graphically intense software or (relatively basic) multitasking will cause an A13 bionic to get toasty. So, a bigger and badder Apple Chip running more intensive software, especially if graphically intense, will realistically get hot as well. "How hot?" is the question and we don't have an answer at this point.

With all that being said, I'm excited for Apple Silicon, but I fully expect the first gen (maybe even the second gen) to be thoroughly underwhelming. I'm expecting a super thin and light Macbook type thing at first, emphasis on battery life and portability versus performance. Maybe a new Air or something in a familiar form factor to entice some consumers that don't know otherwise.

 

[richinaus]

"Critical support for Windows Boot Camp installations for gaming or support for an eGPU"? If Windows Boot Camp support is "critical" to you, then why are you here? Why would you not move to a Windows desktop or gaming laptop at this point?

No one is getting clowned, Apple is simply going in a different direction. You can either get on the ride or get off. I don't think Apple has been unclear about what they're doing. If your needs are truly dependent on Intel, then you need to make up your mind about moving to Windows 10 and commit.

This is 100% right.
Apple are moving in a different direction, and anyone reliant on windows apps / intel should be looking at their setups.

Anway, I am positive these new AS Macs are going to be great and are backing Apple to come out the gate with an awesome computer.

 

[slooksterPSV]

Look,

Until the "Apple Silicon" Mac is in someones possession in actual real life use, we will never know the true performance. Its all up in the air at this point and we don't even have any confirmed specs. Benchmark comparisons between iphone and ipad SOCs to Intel Chips are not that helpful frankly as they doesn't automatically translate into better real life performance, which depends a lot on what apps are running and how they are optimized (or not) for the specific CPU. Also, you HAVE TO factor in thermals. There is a big difference running in a "walled garden" on a phone versus a laptop or desktop where there are many more variables, multitasking, and general much greater demands on the chip. It still remains to be seen if Apple Silicon will run cooler at the same level of processing power. To claim otherwise at this point is premature. To put things in perspective: my iphone 11 Pro gets uncomfortably hot in 10-15 minutes if I'm using it for a Zoom meeting with video enabled. It also gets noticeably warm when running wireless CarPlay and streaming from Spotify, to the point where I've had to take it out of my pocket. Point being graphically intense software or (relatively basic) multitasking will cause an A13 bionic to get toasty. So, a bigger and badder Apple Chip running more intensive software, especially if graphically intense, will realistically get hot as well. "How hot?" is the question and we don't have an answer at this point.

With all that being said, I'm excited for Apple Silicon, but I fully expect the first gen (maybe even the second gen) to be thoroughly underwhelming. I'm expecting a super thin and light Macbook type thing at first, emphasis on battery life and portability versus performance. Maybe a new Air or something in a familiar form factor to entice some consumers that don't know otherwise.

While I don’t disagree I think that adding a fan or even a larger thermal conduit will help immensely. My phone only gets warm when doing updates and even then it’s not too hot. I would like to get a thermal camera to see though.

Even a fan cooling the cpu would make the thermals fantastic. The only complaint i have on the MBA is the fan isn’t connected to the cpu heat pipe. Idling at 90C isn’t nice either and the fan spinning doesn’t do much.

Overall I think thermal throttling won’t be as bad on the A series in a laptop as it may be on phones.

 

[richinaus]

While I don’t disagree I think that adding a fan or even a larger thermal conduit will help immensely. My phone only gets warm when doing updates and even then it’s not too hot. I would like to get a thermal camera to see though.

Even a fan cooling the cpu would make the thermals fantastic. The only complaint i have on the MBA is the fan isn’t connected to the cpu heat pipe. Idling at 90C isn’t nice either and the fan spinning doesn’t do much.

Overall I think thermal throttling won’t be as bad on the A series in a laptop as it may be on phones.

much more surface area on a laptop to dissipate the heat, in comparison to a small device like a phone.

The current iPad Pro 12“ is probably the best thing to compare against, and mine just gets warm sometimes. However, I don’t come anywhere near as close to pushing my ipad as I do a ’computer’.

I am hopeful that heat and loud fans are the 2 main things that are going to be reduced with AS.

 

[ignatius345]

@TheRealAlex
Is this your third thread on the same topic, it's hard to keep count?

It would be nice if you stopped though, or at least stopped regurgitating.

Might help if the MacRumors admins didn't promote this thread in a special banner at the top of the homepage...

 

[mdriftmeyer]

much more surface area on a laptop to dissipate the heat, in comparison to a small device like a phone.

The current iPad Pro 12“ is probably the best thing to compare against, and mine just gets warm sometimes. However, I don’t come anywhere near as close to pushing my ipad as I do a ’computer’.

I am hopeful that heat and loud fans are the 2 main things that are going to be reduced with AS.

You're not getting it. OS X Catalina with iMessage, Safari [6 tabs] running has 378 processes and 1,734 threads on my MacBook Pro 13" 2015. That's a bare basic use of the system.

No iPhone/iPad is running near the process load or thread load just to keep the system afloat. Everything that you take for granted on OS X has a hardware tie to it. Never mind the basic OS X running Apple Apps with noticeably slower performance [and it will] the third party software you cherish must still be ported and unless they move their entire code base by branching for am ARM solution it's going to be most likely in interpreted mode for a long time to come.

 

[slooksterPSV]

No iPhone/iPad is running near the process load or thread load just to keep the system afloat. Everything that you take for granted on OS X has a hardware tie to it. Never mind the basic OS X running Apple Apps with noticeably slower performance [and it will] the third party software you cherish must still be ported and unless they move their entire code base by branching for am ARM solution it's going to be most likely in interpreted mode for a long time to come.

Tech. iOS is macOS in the background. Just a more strict hardware set to deal with + cellular so things have to be lighter. All that aside, I thought Rosetta 2 would do some pre-translations the first time you open the app to optimize running it afterwards?

 

[entropi]

I hope I can buy something like the iGPU 2015 rMBP 15" - fairly big screen, no excess batterydrain & heat (no dGPU)...
(I like my mba 11" too tough - except for the screen - when the fan is on 1200 rpm it is really quiet and it can still handle some interesting things in Reason, Logic Pro and Ableton Live)

 

[bill-p]

Not necessarily disagreeing with your main point, but you keep mentioning the 5nm thing like it isn’t already happening. Why are you doing that? The A14 in the iPhone is on the 5nm process, in reliable mass production right now. As is the A14X destined for the next iPad Pro and a variant based on the A14X for the first ASi MB/MBA/MBP.

It is probably more about the SoC scaling and packaging that will be the challenge for the more powerful Macs that pack a dGPU in their current Intel forms than what process they are on.

Because the iPhone 12 has not happened yet? Last I checked, it's still not announced as of today. A14 and A14X are "educated" guesses at best.

 

[weems62]

Will a first Gen ARM based MacBook Pro Surpasss and Outperform the previous Intel based Version ? I’m gonna say No.

It May have Longer battery life, it will have WiFi 6 (intentionally left out of current gen) It May be less expensively But will it be “Faster”

Heck they can go from LPDDR4 to LPDDR5 RAM today by switching to ARM.

i feel like the link below we are getting clowned by the Swicth ti ARM, which won’t have critical support for Windows Boot Camp installations for Gaming or support for an eGPU.

I think its safe to say people are over hyping the apple silicon. Some people even think the a12z is as powerful as an i9 processor because of some benchmarks LOL. But we all know this isn't true. I would say the actual pure power of the a12z processor is on par with a ninth generation core i5, but does have much better efficiency, resulting in better thermals. I think the apple silicon will outperform the processors in the current entry level MacBook Pro 13 inch. (probably part of the reason they left the 8th generation processors in the lower end) They will outperform those pretty good, but not enough to justify switching from a PC to a Mac. Will probably be about the same as AMD's upcoming CPU's. The graphical performance will definitely be better than those 8th gen intel processors though, as it is currently horrible.

Now for efficiency. They will undoubtedly be more efficient than the current intel processors and probably better than the current AMD options. But they will seem better than the other two options because they will be in a Mac, where thermals have been limited for years. If your'e looking for something that you rely on to make money (if you're a real professional) then you should probably just go ahead and buy a 16 inch MacBook Pro if you want a MacBook, or just get something using windows.

This isn't just because of the initial performance limit though. The apple silicon Macs may be getting support for Adobe and Microsoft, but other things will most likely be very slow to the race. Even with apple using rosetta for emulation, it will be up to the developers of software to keep up, and I don't expect some to.

 

[SuperCachetes]

That is the ultimate goal of the switch.

I honestly think the real goal of the switch was much more indirectly related to what any of us potential users will see.

By unhitching from Intel's wagon (or albatross, depending on how cynical you want to be) Apple are free to set their own schedule for development and product release. They'll be able to react more quickly to changes in standards, technology, or customer demand. Consider some examples:

• Apple will be able to update laptops and desktops whenever they want. Hopefully there doesn't ever need to be another complaint about it being "X,XXX days between updates for expensive machines." No more "Apple is still charging $X,XXX for two-year-old Mac..." If they set the tick/tock of the clock, they can do meaningful Mac iterations every year if they choose. TSMC is on speed dial so to speak, while Intel seems to be aloof in comparison.
• When the next WiFi, Bluetooth, USB, or Thunderbolt standard is formalized, Apple won't have to worry about whether Intel's chipsets support it or when the Intel processors are released. Just won't matter since Apple is designing their own SoCs.
• If Apple release an ASi product and for whatever reason it has a serious internal flaw and/or it is savagely panned by reviewers and customer feedback, they can develop the fix (or replacement) SoC or product on their own time. No need to be betrothed to a future Intel chip generation.

So, yeah, I believe Apple when they say they want to make the best product/experience possible - and the battery life vs. performance discussion is an interesting one. But ultimately, kicking Intel to the curb allows Apple the freedom to really go at all of this hard-core, at their own pace. That is what excites me more than the numerical improvements in one metric or another...

 

[slooksterPSV]

I honestly think the real goal of the switch was much more indirectly related to what any of us potential users will see.

By unhitching from Intel's wagon (or albatross, depending on how cynical you want to be) Apple are free to set their own schedule for development and product release. They'll be able to react more quickly to changes in standards, technology, or customer demand. Consider some examples:

• Apple will be able to update laptops and desktops whenever they want. Hopefully there doesn't ever need to be another complaint about it being "X,XXX days between updates for expensive machines." No more "Apple is still charging $X,XXX for two-year-old Mac..." If they set the tick/tock of the clock, they can do meaningful Mac iterations every year if they choose. TSMC is on speed dial so to speak, while Intel seems to be aloof in comparison.
• When the next WiFi, Bluetooth, USB, or Thunderbolt standard is formalized, Apple won't have to worry about whether Intel's chipsets support it or when the Intel processors are released. Just won't matter since Apple is designing their own SoCs.
• If Apple release an ASi product and for whatever reason it has a serious internal flaw and/or it is savagely panned by reviewers and customer feedback, they can develop the fix (or replacement) SoC or product on their own time. No need to be betrothed to a future Intel chip generation.

So, yeah, I believe Apple when they say they want to make the best product/experience possible - and the battery life vs. performance discussion is an interesting one. But ultimately, kicking Intel to the curb allows Apple the freedom to really go at all of this hard-core, at their own pace. That is what excites me more than the numerical improvements in one metric or another...

You’re correct and you know who found the CVEs for a few of these exploits? Apple. Apple found flaws in Intels chips. Not some random researching team trying to exploit holes and weaknesses, nope, someone at Apple saw a flaw that compromised the security and integrity of their systems. I think that could be the straw that broke the camels back.

 

[wood1208]

Let me put in simple words. When Apple decided to make it's CPU/GPU to go into MAC than the decision was made in every respect to make the MAC better without Intel chips and also to control the cost, so Apple can offer at better price/performance/features point. Moreover, Apple can also dictate the regular upgrade cycle. I believe the thinking is to expand the MAC market to bring the services revenue along with it. Recent products with likable price like 2020 SE,Macbook Air suggests Apple is becoming aware of making products for price sensitive users while making the higher end products for those who want the best.

 

[aleni]

I’m thinking of apple could use a gpu module like the macpro for all the laptop lineups. So every laptops will come with its own integrated gpu like the ipad, and you just buy a simple ext gpu module from apple. Plug it in and boom, high gpu performance!

 

[ignatius345]

Let me put in simple words. When Apple decided to make it's CPU/GPU to go into MAC than the decision was made in every respect to make the MAC better without Intel chips and also to control the cost, so Apple can offer at better price/performance/features point. Moreover, Apple can also dictate the regular upgrade cycle. I believe the thinking is to expand the MAC market to bring the services revenue along with it. Recent products with likable price like 2020 SE,Macbook Air suggests Apple is becoming aware of making products for price sensitive users while making the higher end products for those who want the best.

It's not MAC. It's typed "Mac" -- short for "Macintosh". It always has been.

Why do you people capitalize it like some weird acronym? Because PC is capitalized? PC stands for "personal computer" -- what do you think MAC stands for? Multi Access Computer? Mighty Apple Computer?

Please, just stop typing it like this. It looks insane. It's usually what you see when you go to work in some PC-centric office where you get an email from clueless IT people telling you how to "access Outlook email on your MAC".

 

[Seoras]

Will a first Gen ARM based MacBook Pro Surpasss and Outperform the previous Intel based Version ? I’m gonna say No.

It May have Longer battery life, it will have WiFi 6 (intentionally left out of current gen) It May be less expensively But will it be “Faster”

Heck they can go from LPDDR4 to LPDDR5 RAM today by switching to ARM.

i feel like the link below we are getting clowned by the Swicth ti ARM, which won’t have critical support for Windows Boot Camp installations for Gaming or support for an eGPU.

I'm going to say yes.

My 2nd home PC was an Acorn Archimedes with an ARM2 in it which I had while doing my under-grad in CS/Elec at Uni.
Back then the thing that really made the ARM stand out, from the compiler tests and comparisons I did, was how much more efficient a RISC architecture like ARM was over the CISC architectures of Intel x86 and Moto 68k.
The ARM's C compiler could reduce a C function to about 1/4 of the number of instructions the other processors compilers used.
Which is counter intuitive. The whole point of having a complex instruction set (CISC) processor is that you can do clever things in one instruction right?
Admittedly there were others factors back then in ARMs favour like having 16 registers and conditional instruction execution which provided greater efficiencies than the CISC's had. They've since learned and caught up.
However the performance is really down to how well the compiler makes good of the instruction set and how efficient it is.
As with all things Apple now they have total vertical integration from silicon to application I expect to see greater efficiencies which should translate to better performance.
We've hit something of a ceiling over the last decade with processor clock speed. Now any gains are in GPUs which will top out too at some point. Vertical integration and efficiency is the next winning ticket which maybe another reason NVIDIA is interested in ARM.

I personally don't care about performance so long as it is on par and in the same ball park.
I'm more interested in the additional silicon goodies that come with Apple's silicon.
Do we get FaceID on Mac now? The hardware acceleration of CoreML AI models? Secure Enclave? etc, etc?
That, for me anyway, is a bigger selling point that what GeekBench reports.

 

[richinaus]

You're not getting it. OS X Catalina with iMessage, Safari [6 tabs] running has 378 processes and 1,734 threads on my MacBook Pro 13" 2015. That's a bare basic use of the system.

No iPhone/iPad is running near the process load or thread load just to keep the system afloat. Everything that you take for granted on OS X has a hardware tie to it. Never mind the basic OS X running Apple Apps with noticeably slower performance [and it will] the third party software you cherish must still be ported and unless they move their entire code base by branching for am ARM solution it's going to be most likely in interpreted mode for a long time to come.

You’re not getting it - I said it was the nearest thing to compare in terms of size and processor. Who said anything about tasks......

 

[involuntarheely]

I honestly think the real goal of the switch was much more indirectly related to what any of us potential users will see.

By unhitching from Intel's wagon (or albatross, depending on how cynical you want to be) Apple are free to set their own schedule for development and product release. They'll be able to react more quickly to changes in standards, technology, or customer demand. Consider some examples:

• Apple will be able to update laptops and desktops whenever they want. Hopefully there doesn't ever need to be another complaint about it being "X,XXX days between updates for expensive machines." No more "Apple is still charging $X,XXX for two-year-old Mac..." If they set the tick/tock of the clock, they can do meaningful Mac iterations every year if they choose. TSMC is on speed dial so to speak, while Intel seems to be aloof in comparison.
• When the next WiFi, Bluetooth, USB, or Thunderbolt standard is formalized, Apple won't have to worry about whether Intel's chipsets support it or when the Intel processors are released. Just won't matter since Apple is designing their own SoCs.
• If Apple release an ASi product and for whatever reason it has a serious internal flaw and/or it is savagely panned by reviewers and customer feedback, they can develop the fix (or replacement) SoC or product on their own time. No need to be betrothed to a future Intel chip generation.

So, yeah, I believe Apple when they say they want to make the best product/experience possible - and the battery life vs. performance discussion is an interesting one. But ultimately, kicking Intel to the curb allows Apple the freedom to really go at all of this hard-core, at their own pace. That is what excites me more than the numerical improvements in one metric or another...

while I want to believe what you're saying, what you're doing is just assuming Apple is going to do some things which are not really necessary.

to play the devil's advocate, Apple could also take the opportunity to upgrade less frequently because
- macs are not as high profit margin as iphone and ipads
- by moving to ARM there's less competition and comparisons with alternative computers will be more difficult
- the cost of developing a much more general purpose system like macOS (compared to ios/ipados) is high and the payoff is relatively low (can't see Apple forcing all app developers pay 30% on mac...)

so it seems to me Apple also has incentives to not update macs as frequently. I guess we'll have to see.

 

[Morgenland]

Apple never sells "Netbook"-stuff.

 

[Joelist]

This is getting a bit weird...

First, Apple Silicon uses the ARM instruction set - that is all it has in common with the rest of the ARM world. The SOC itself is a 100% Apple in house design. The GPU is a 100% in house design as are all of the other various blocks. So it has to be looked at as its own thing - and Apple Silicon is VERY high performance.

Second, at WWDC they were clear about the reason for the transition. They have things they want to deliver on Mac that they can only do with their own custom hardware. And yes I assume they want to bring over some of the goodness they already have created on the iPad Pro for example.

Third, Apple already clearly stated the Macs will have their own family of SOCs. Seeing as the A12 and A13 already are outperforming Intel and AMD mid to mid-high range CPUs on a per core basis there is every reason to expect the new Mac SOCs will be impressive even if all they do is up their core counts to something like 8 performance 4 low power.

Fourth, remember that this is not a conventional CPU arrangement. It is a SOC with a series of separate co-processor blocks (Neural, GPU and others). In this world more of the work typically done by the CPU is being handed off to specialized blocks.

Apple was also clear about all the steps they have taken and are taking to ease the transition. This includes Rosetta 2 which actually does pre-compiling to increase speed. Also the development tools are ready and all you need to do for a lot of apps is recompile. So a lot of the apprehension is premature.

 

[Yurk]

High Sierra 10.13.4 blocked TB1 and TB2 support for eGPUs, forcing users to upgrade to macs with TB3.
Mojave blocked Nvidia eGPUs and CUDA.
Catalina blocked 32 bit apps.
Big Sur on ARM dropped the last remaining way of using a Nvidia eGPU and CUDA, i.e. Bootcamp.
The mac mini cannot be downgraded from Mojave and the 16" mac cannot be downgraded from Catalina (and has serious noise / thermal / power throttle defects). That's the final nail in the coffin.

I can no longer use my $3300 Nvidia Titan V eGPU. All serious machine learning or HPC developers need CUDA. MXNet needs CUDA. Wolfram Mathematica machine learning needs CUDA. The performance of matrix multiplication on ARM is laughable compared to Intel processors. Intel MKL libraries are heavily optimized, much more than ARM BLAS or AMD BLAS libraries. Intel is not 3rd, they are the 1st in HPC performance, because they back their hardware with proper software library support. Intel knows that, that is why they are still #1 in CPU market share. Nvidia gets that. AMD does not get that. Apple drops valuable functionality with every update, and calls it an 'upgrade'. Their new macbook will be an i-phone with a keyboard.

I abandoned my 5 macs and upgraded to a Fischer Price laptop. It is just as capable of not supporting 32-bit applications and Nvidia CUDA, it never overheats, and moreover it has a replaceable battery.

 

[SuperCachetes]

while I want to believe what you're saying, what you're doing is just assuming Apple is going to do some things which are not really necessary.

to play the devil's advocate, Apple could also take the opportunity to upgrade less frequently because
- macs are not as high profit margin as iphone and ipads
- by moving to ARM there's less competition and comparisons with alternative computers will be more difficult
- the cost of developing a much more general purpose system like macOS (compared to ios/ipados) is high and the payoff is relatively low (can't see Apple forcing all app developers pay 30% on mac...)

so it seems to me Apple also has incentives to not update macs as frequently. I guess we'll have to see.

So your “devil’s advocate” position is that Apple would commit corporate suicide and let the entire Mac line - the legacy of the entire company going back to the ‘70’s - wither and die by attrition and neglect? They could do that now with their current Intel lineup with a lot less effort. Some would argue they’ve already been trying.

What I’m assuming, in a nutshell, is that Apple makes *some* money on laptops and desktops, and that they actually care about how customers view their computer lineup. Point to all the mistakes from the past that you like, from glacial update frequency to dubious hardware/feature experimentation... but Apple taking control of their own destiny in the realm of silicon has the potential to eliminate any problems caused by external roadblocks.

You are right about this Apple Silicon thing being a significant effort. I think you are way off the mark trying to reduce it to bean-counting just yet.

 

[involuntarheely]

High Sierra 10.13.4 blocked TB1 and TB2 support for eGPUs, forcing users to upgrade to macs with TB3.
Mojave blocked Nvidia eGPUs and CUDA.
Catalina blocked 32 bit apps.
Big Sur on ARM dropped the last remaining way of using a Nvidia eGPU and CUDA, i.e. Bootcamp.
The mac mini cannot be downgraded from Mojave and the 16" mac cannot be downgraded from Catalina (and has serious noise / thermal / power throttle defects). That's the final nail in the coffin.

I can no longer use my $3300 Nvidia Titan V eGPU. All serious machine learning or HPC developers need CUDA. MXNet needs CUDA. Wolfram Mathematica machine learning needs CUDA. The performance of matrix multiplication on ARM is laughable compared to Intel processors. Intel MKL libraries are heavily optimized, much more than ARM BLAS or AMD BLAS libraries. Intel is not 3rd, they are the 1st in HPC performance, because they back their software with proper software library support. Intel knows that, that is why they are still #1 in CPU market share. Nvidia gets that. AMD does not get that. Apple drops valuable functionality with every update, and calls it an 'upgrade'. Their new macbook will be an i-phone with a keyboard.

I abandoned my 5 macs and upgraded to a Fischer Price laptop. It is just as capable of not supporting 32-bit applications and Nvidia CUDA, it never overheats, and moreover it has a replaceable battery.

I agree on everything except for MKL for AMD and AMD Blas. with AMD processors you can use MKL_DEBUG=5 with MKL versions up to 2019 and pretend that the processor is actually Intel and it's going to work. last I checked both OpenBLAS and BLIS seemed to be very close if not the same as MKL in those scenarios (obliterating Intel processors in somewhat parallelizable code)

 

[involuntarheely]

So your “devil’s advocate” position is that Apple would commit corporate suicide and let the entire Mac line - the legacy of the entire company going back to the ‘70’s - wither and die by attrition and neglect? They could do that now with their current Intel lineup with a lot less effort. Some would argue they’ve already been trying.

What I’m assuming, in a nutshell, is that Apple makes *some* money on laptops and desktops, and that they actually care about how customers view their computer lineup. Point to all the mistakes from the past that you like, from glacial update frequency to dubious hardware/feature experimentation... but Apple taking control of their own destiny in the realm of silicon has the potential to eliminate any problems caused by external roadblocks.

You are right about this Apple Silicon thing being a significant effort. I think you are way off the mark trying to reduce it to bean-counting just yet.

All I'm saying is I don't know. part of me thinks Apple has not been focusing that much on macs recently and it seems unreasonable to think that suddenly they're shifting a lot of resources to this huge move when macs are not sold at the same volumes as iphones even though the prices are now comparable. in a stagnating market the move to ARM would shield Apple from competition even more (they are already somewhat shielded by leveraging the apple ecosystem and the natural monopoly on macos itself).

at the same time maybe all this time they have actually spent on developing ARM and macos for ARM. I can't deny that another tendency for apple recently has been to search for high value high volume products to sell since iphones sales are stagnating. with people working from home I'm guessing at least in the short term the demand for good and versatile computing devices has grown back to past levels so Apple could profit off of this new environment.

 

[Yurk]

I agree on everything except for MKL for AMD and AMD Blas. with AMD processors you can use MKL_DEBUG=5 with MKL versions up to 2019 and pretend that the processor is actually Intel and it's going to work. last I checked both OpenBLAS and BLIS seemed to be very close if not the same as MKL in those scenarios (obliterating Intel processors in somewhat parallelizable code)

I agree, except this tricks previous Intel MKL libraries, but not the most recent versions. There is some Intel sabotage going on, but also Intel optimizes their libraries for their processors. In particular, thanks to the most recent Intel MKL libraries, Mathematica 12 performes matrix multiplication 4 times faster than Mathematica 11 on my Xeon 8173M. Just thanks to optimizations on the Intel MKL libraries, Mathematica is faster on my 28 Intel Xeon than a 64 core AMD Epyc.