Apple Silicon: The Complete Guide

1: They will need to be, only this would justify such architecture migration;
2: Intel sucks (slow, bloated, huge consumption, expensive, obscure, etc.);
3: If Apple is that serious on migrating to Arm, they wouldn't do that if it sucks as much as Intel ones - Arm is known as being way more efficient than Intel in lots of aspects, and there is no reason on not making Arm CPU decent enough for true workstations, and perhaps creating a true revolution in the whole workstation/laptop market

Only a ten times change justifies it? The Intel migration was barely even two times, and that was largely because most Intel chips had more cores than the PowerPC equivalents.

There's quite a range between "the old CPU sucks" and "the new one is 10x as fast". What if it's 1.5x as fast? That already sounds pretty great.
[automerge]1598630323[/automerge]
on Arm, besides being a way more efficient cpu, you can put a lot more cores in the same room

People seriously overestimate how much code can be parallelized and thus benefits from many cores.

If you take, say, an AOT compiler or a video codec, sure, you can parallelize that a lot. But for many day-to-day things, not so much. Almost all JS code? Single-threaded. Typical algorithms in a line-of-business app? Single-threaded. Tons of things is I/O-bound, not CPU-bound: the CPU just waits for a network or disk operation. Some download task, some file transfer, some database query.
 
People seriously overestimate how much code can be parallelized and thus benefits from many cores.

The real advantage is heteogeneous compute units. 16 or 32 cores would be more than plenty for typical desktop/light server use, but then you can add on all sorts of special-purpose compute units.
 
Apple has been playing with ARM Macs in their lab for years... they obviously wouldn't make an official announcement unless it was at least as good as a comparable Intel Mac.

It's gonna be exciting to see what a Macintosh can do... with Apple's own specially-designed processor inside... and with their OS that is tailor-made for that processor.

It'll be a hardware/software solution that very few other companies can offer.
 
I think you will be disappointed then, 10x is quite unrealistic in my opinion. I'd be surprised if it's more than 3x faster.

No way Apple Silicon Macs would be 3x let alone 10x faster.

Comparison Apple will make in their keynote is how much faster it is compared to previous Mac generation. I would expect 2X. Real comparison would be latest gen Intel, no the Intel CPUs in the previous Macs as those would be 1+ year old. Even that 2X would put the Apple silicon way ahead of the competition.

Why not even faster? 2X would be enough to launch the product, thus they probably launch as soon as they get to 2X. That will be a huge market differentiator and also decrease the materials.
 
Dassault might stop laughing if Apple acquires enough CAD patents of their own to "convince Dassault of their best interest" in porting to the new platform. Look at their comparative R&D budgets.
Posssible, but I doubt Apple even cares. They don’t have to go after EVERY market with macOS or iOS, just enough markets to be profitable.
 
I find this argument about price always interesting.
As the die increases so Dows the cost. It's not a linear increase because the increase in failure is not linear.
Thinking that Apple will be able to compete on CPU price when they don't own the fab. Will need to absorb a higher failure rate, etc.; is just humorous.

The cost of that ARM processor is going to go up real high.
64 lanes of PCIe, embedded desktop graphics, larger caches, virtualization support, etc...
Opps, now you have a real processor.
Once again, people should check out Ampere and compare the cost of a real server class ARM processor that needs to go in something like a Mac Pro.

I for one, will not buy a Mac that is not Intel compatible because I use VM's in my job and if it can't run the tools I need I won't buy it. My company would immediately stop buying Macs if RedHat or CentOS won't run in a VM.

If Apple wants to kill the Mac, switch to ARM.
Didn’t you hear. Linux is already compatible. Lol.
 
I for one, will not buy a Mac that is not Intel compatible because I use VM's in my job and if it can't run the tools I need I won't buy it. My company would immediately stop buying Macs if RedHat or CentOS won't run in a VM.

Watch WWDC 2020 Special Event Keynote 1:41:50 where they show a Linux OS running on ARM. RedHat has had an ARM version since 2017 and CentOS since 2018, Mind telling us where you've been that you don't know this?

If Apple wants to kill the Mac, switch to ARM
Everybody is setting the groundwork to switch to ARM. The only thing Intel has is the large old software archive and if Microsoft's x86/x64 emulator is as good as Apple's then ARM PCs (which are cheaper then Intel PCs) become viable and it won't be long before either Intel itself produces an ARM chip that is decent or drops out of the CPU business.
 
Last edited:
Watch WWDC 2020 Special Event Keynote 1:41:50 where they show a Linux OS running on ARM. RedHat has had an ARM version since 2017 and CentOS since 2018,

Oh, for crying out loud, not this nonsense again.

Running an Arm version of Linux in a VM is of very limited utility for a developer who is working with production systems which are x86/amd64. You act as if we haven’t been all over this topic already in this thread.

Intel architecture absolutely dominates in the server world, and the rest of us in the dev and devops are just working within that reality. Arm’s not going to be a credible competitor to x86 for many, many years (if at all, it’s hardly a foregone conclusion).

Mind telling us where you've been that you don't know this?

Your snark here doesn’t serve your argument or elevate the quality of this discussion at all. Why do you persist in being so obnoxiously wrong about a subject you admittedly don’t know much about?
 
Oh, for crying out loud, not this nonsense again.

Running an Arm version of Linux in a VM is of very limited utility for a developer who is working with production systems which are x86/amd64. You act as if we haven’t been all over this topic already in this thread.

Mind telling us where you’ve been that you don’t know this?
I remember a time when Linux didn’t run on amd64, and a time when Linux on AMD64 was considered pretty useless. It’ll be fine.
 
I remember a time when Linux didn’t run on amd64, and a time when Linux on AMD64 was considered pretty useless. It’ll be fine.

Yeah, I don’t doubt if Arm ever gains any traction in the server room that I will have little difficulty replicating that environment on an Arm local dev machine. But it’s not going to happen the other way around.
 
Running an Arm version of Linux in a VM is of very limited utility for a developer who is working with production systems which are x86/amd64. You act as if we haven’t been all over this topic already in this thread.

I wonder how fast that will change. AWS has Graviton - which is cheaper for compute compared to amd64. Containers can be built for multi-cpu architectures - so the tools are getting easier. Server profits grow on power efficiency - I'm sure those cloud vendors will be happy to quicken the demise of power hungry CISC chips if they can squeeze more compute out of less juice.

10 to 1, I'll bet Apple will have their own cloud of ARM servers (iCloud?) to deploy stacks to just as soon as they can.
 
I wonder how fast that will change. AWS has Graviton - which is cheaper for compute compared to amd64. Containers can be built for multi-cpu architectures - so the tools are getting easier. Server profits grow on power efficiency - I'm sure those cloud vendors will be happy to quicken the demise of power hungry CISC chips if they can squeeze more compute out of less juice.

10 to 1, I'll bet Apple will have their own cloud of ARM servers (iCloud?) to deploy stacks to just as soon as they can.

Based on Data Center ARM-ification is Around the Corner ARM adoption is going to be slow for a while.

There is just a lot of inertia in terms of hardware and software preventing fast adoption. Heck, those reel to reel computers you see in old films were still a thing in a hand full of places clear into the the early 2000s. Ohio State University was still using punch cards into the 1980s and NMSU used old style mainframe drives from the 1970s as storage for free low speed internet into the early 2000s. (While people like to blame someone for throwing a hissy-fit about the government competing with cprivate businesses being the reason it was ended the reality was it was flaky as all get out. The main server was known as Dante and those who used it had a little jingle 'Dante died' every time the thing went down...which seemed to be at least once a month)

The point is we have seen that business will keep using hardware and software well past its rational use-by date because it would be insanely expensive to upgrade, their software may not run on anything newer, they literally can't get rid of it (common problem with governmental realty institutions) and so elect to keep using it rather then let it rot, and/or a sunk cost mentality.
 
Oh, for crying out loud, not this nonsense again.

Running an Arm version of Linux in a VM is of very limited utility for a developer who is working with production systems which are x86/amd64. You act as if we haven’t been all over this topic already in this thread.

A ton of software development is so high-level these days that neither the OS nor the arch matters. A web developer who uses a Docker container to isolate their web API doesn't care if it's macOS on x86-64, Linux on ARM64 or Windows on x86-32. It makes no difference.

Most iOS development has also always run on multiple archs: the Simulator runs on x86, and the actual published app (or debugged-on-device app) runs on ARM.

Intel architecture absolutely dominates in the server world, and the rest of us in the dev and devops are just working within that reality. Arm’s not going to be a credible competitor to x86 for many, many years (if at all, it’s hardly a foregone conclusion).

Be that as it may, it's just not that relevant. Most of the libraries you'll be using on a Linux server are OSS and, if not already available in ARM, probably a recompile away.

Your snark here doesn’t serve your argument or elevate the quality of this discussion at all. Why do you persist in being so obnoxiously wrong about a subject you admittedly don’t know much about?

They're being a bit rude, but I don't really see how "Apple shows Linux running on ARM" is an "obnoxiously wrong" response to "My company would immediately stop buying Macs if RedHat or CentOS won't run in a VM." — it's a correct thing to note that we already know for a fact that at least one Linux distro runs in a VM on a future Mac; they showed it.
 
A ton of software development is so high-level these days that neither the OS nor the arch matters. A web developer who uses a Docker container to isolate their web API doesn't care if it's macOS on x86-64, Linux on ARM64 or Windows on x86-32. It makes no difference.

I went a little overboard regarding this in "Could x86-64 emulation be possible on ARM Macs?" by saying in this day and age no-one should really be writing platform spacific software but this confirms that in some areas it doesn't amount t a hill of beans

I pointed out there were ARM packages from Docker in the "Could x86-64 emulation be possible on ARM Macs?" thread via the Getting started with Docker for Arm on Linux. article.

They're being a bit rude, but I don't really see how "Apple shows Linux running on ARM" is an "obnoxiously wrong" response to "My company would immediately stop buying Macs if RedHat or CentOS won't run in a VM." — it's a correct thing to note that we already know for a fact that at least one Linux distro runs in a VM on a future Mac; they showed it.

Some of this got kicked around in the "Could x86-64 emulation be possible on ARM Macs?" thread. Now if they said that had to fun x86 code I would have been more understanding but even there there Apple has a translator and Microsoft is going to have an x54 one for the masses by 2021 (Though i seem to remember someone saying developers get to play with an early build of it in November)

Apple is going to ARM and Microsoft wants to go there though the translator has got to be on par with Apple's if they want some PC makers to drop x86 in favor of ARM.

Besides it is not like Apple is going to stop selling Intel macs in 2021. As I pointed out in the "ARM based MacBook Pro discussion [merged]" thread:

Apple supported PowerPC code even with Intel macs from January 2006 to July 20, 2011 or to put it bluntly Apple did NOT stop support for Rosetta (ie PowerPC code) until 2011. 2009 was the end of the line for the hardware not the software. So Apple supported PowerPC code for nearly two years after it became clear that the PowerPC was effectively a road to nowhere.

If Apple does something similar then the last support for x86-x64 code will end sometime in 2026 more then enough time for recoding/recompiling to happen. And that is ignoring that unlike then there is the referb shop by Apple which sales Mac as old as three years. I have no idea if that will extend Rosetta 2's life or not.
 
Last edited:
Running an Arm version of Linux in a VM is of very limited utility for a developer who is working with production systems which are x86/amd64. You act as if we haven’t been all over this topic already in this thread.
Production software needs to be tested and performance tuned on production-equivalent server or Cloud infrastructure, not on developer laptops. If your container build scripts are so fragile that a change of ISA causes them to fail, you should worry about your manager replacing you.
 
Production software needs to be tested and performance tuned on production-equivalent server or Cloud infrastructure, not on developer laptops. If your container build scripts are so fragile that a change of ISA causes them to fail, you should worry about your manager replacing you.

Of course. Why do you assume this isn’t the case at my workplace? You do realize that it’s not a personal attack against you that most production infrastructure is x86. It’s not like the world is doing this just to spite you personally, or to spite Apple. It’s just the nature of the world as we live in it. There is absolutely no reason for you to act so defensive here.

As always, my position is very simple. The lowest risk path (and least expensive) will be for companies to continue to invest in development environments that are aligned with their production infrastructure, and not the other way around. The migration of macOS to Arm processors simply makes those devices inappropriate for developers at my company and our IT department will adjust the supported platforms as a result. We’re not just “holding it wrong.”

Honestly, I think Apple have just stopped caring about the devops userbase. macOS has been a fantastic platform for us since the release of MacOS X, but Apple’s current world view seems to be hyper-focused on video and audio production, Xcode developers, and general office app users. For those markets, Apple Silicon machines will be a great solution. Apple have every right to focus on those users even if it comes at the expense of unix-focused developers like myself. I’ll miss macOS, but Apple probably won’t miss me.

The benefits of Arm do not outweigh the risk and expense. I’m disappointed, but there’s nothing I can do about it. Suggesting that our code is fragile or that I deserve to be fired for this is untethered from reality and offensively inflammatory.
 
Last edited:
Of course. Why do you assume this isn’t the case at my workplace? You do realize that it’s not a personal attack against you that most production infrastructure is x86. It’s not like the world is doing this just to spite you personally, or to spite Apple. It’s just the nature of the world as we live in it. There is absolutely no reason for you to act so defensive here.

True but there is no promise that the future will stay x86.

As always, my position is very simple. The lowest risk path (and least expensive) will be for companies to continue to invest in development environments that are aligned with their production infrastructure, and not the other way around. The migration of macOS to Arm processors simply makes those devices inappropriate for developers at my company and our IT department will adjust the supported platforms as a result. We’re not just “holding it wrong.”

The problem with this mentality is the lowest risk path (and least expensive) tends to result in stagnation which can be deadly for a business.

"A manufacturer who sticks to old equipment cannot compete, and must fail. To survive, he must persu"ade people to risk savings in his business. He can then buy new equipment, increase production, and show a profit." - Yankee Dood It (1956)

My father worked for a company called Yaeger Typesetting Company which, when the original died, spit in two by his sons. One son continued the lowest risk path (and least expensive) hot metal Linotype side of the business while the other went with riskier option of computer soft metal.The Linotype side went out of business in the early 1980s (throwing my father out of work) while the computer side continued on but it couldn't easily adapt to the rapidly changing market of computer printing and as a result also went out of business in the late 1990s early 2000s.

"A lot of ATMs around the globe are still running Windows XP embedded, long after Microsoft ceased support with security and stability patches." - (2019 Dec 27) Bank ATM migration to Windows 10 – Deadline Jan-2020

Yes that is indeed the least expensive part but tell us how in the name of sanity can this behavior also be considered the "lowest risk path" rather then getting ATMs that run something that did not stop being updated in 2014?!

It like that political cartoon several years back regarding the FDA getting involved with drug executions. "You want it to be safe and effective?!"

Honestly, I think Apple have just stopped caring about the devops userbase. macOS has been a fantastic platform for us since the release of MacOS X, but Apple’s current world view seems to be hyper-focused on video and audio production, Xcode developers, and general office app users. For those markets, Apple Silicon machines will be a great solution. Apple have every right to focus on those users even if it comes at the expense of unix-focused developers like myself. I’ll miss macOS, but Apple probably won’t miss me.

Since the MacOS has had Single UNIX Specificationversion 3 (SUSv3) via Darwin since 2007 and odds are they have converted the DarwinOS to the AS chip and there are other UNIX ARM platforms I don't see the problem. Now per Server market takes a pause in 2019 – still dominated by Windows and x86 if you were talking about x86 Windows users then yea that is an issue.

The benefits of Arm do not outweigh the risk and expense. I’m disappointed, but there’s nothing I can do about it. Suggesting that our code is fragile or that I deserve to be fired for this is untethered from reality and offensively inflammatory.

Perhaps that is true for what you are doing but that doesn't mean that is true for the market in general. My father and the owner of the hot metal Linotype side of Yaeger Typesetting Company that that method and the equipment that supported it still had a future but anybody that looked at the big picture could tell you it was a road to nowhere. Just because hot metal Linotype had become the industry standard c. 1900 didn't mean it would stay the standard.

My father and that one son of Yaeger didn't understand that - they want to take lowest risk and least expensive path and stay with hot metal Linotype.

If the Forrester total economic impact report incorrect ARM for data-centers has the benefits of
  • 30% to 60% lower upfront infrastructure costs
  • 15% to 35% lower ongoing infrastructure costs
  • Cloud infrastructure cost savings of up to 80%
To rephrase Yankee Dood It: 'He who sticks to old equipment and old methods cannot compete, and must fail.
 
Last edited:
True but there is no promise that the future will stay x86.

I've never said otherwise. There's definitely no chance at all of the data center migrating to Arm within the service life of any developer machine I purchase this year, though.

"A manufacturer who sticks to old equipment cannot compete, and must fail. To survive, he must persu"ade people to risk savings in his business. He can then buy new equipment, increase production, and show a profit." - Yankee Dood It (1956)

You've got some pretty terrible analogies here. x86/amd64 hardware is hardly the equivalent of hot metal Linotype or Windows XP. Intel holds the lead in single-core performance today, and AMD continues to develop and release top of the line processors for both workstations as well as servers.

Yes that is indeed the least expensive part but tell us how in the name of sanity can this behavior also be considered the "lowest risk path" rather then getting ATMs that run something that did not stop being updated in 2014?!

This is such a comically bad analogy. Nobody is considering leaving the macOS ecosystem and moving to a platform that was EOL'd in 2014. Nobody. Why would you even bring this up?

To rephrase Yankee Dood It: 'He who sticks to old equipment and old methods cannot compete, and must fail.

Who is sticking to old equipment? No, I mean really --who are you aiming this comment at. I can't imagine it's for me.
 
You've got some pretty terrible analogies here. x86/amd64 hardware is hardly the equivalent of hot metal Linotype or Windows XP. Intel holds the lead in single-core performance today, and AMD continues to develop and release top of the line processors for both workstations as well as servers.

This is such a comically bad analogy. Nobody is considering leaving the macOS ecosystem and moving to a platform that was EOL'd in 2014. Nobody. Why would you even bring this up?

You clearly missed the point of the analogies. The point was that businesses tend toward sticking with old equipment and methods long past the time they should have been replaced or updated because it is the "least expensive" option.

I've never said otherwise. There's definitely no chance at all of the data center migrating to Arm within the service life of any developer machine I purchase this year, though.

As the ATM example shows that "service life", thanks to that "lowest risk path (and least expensive)" mentality you were on about, can be where the least expensive part causes the "lowest risk path" to go into dodo land. As I noted in a post a while ago one of the museums I surveyed for my thesis in the early 1990s was using an 8088 to catalog their collection

The reason was for them that it was supposedly the "lowest risk path (and least expensive)" option but when you looked at the big picture it wasn't. If anything ever happened to that computer they would have been up cripple creek without a battle because the database program didn't run on later machines and there was no way other than hand inputting (which costs) to transfer their some 10,000 item collection.

The saddest thing was that someone had offered to donate a 486 with a working license to a database program that still exists today but they didn't want to have the expanse of transfusing the data dn so turned it down. Which meant they would have to purchase a newer computer later down the line and spend more money doing the transfer.

Right now there is company that for a while embraced that "lowest risk path (and least expensive)" idea and now they are scrambling to catch up now that a real contender has appeared. Maybe you have heard of them. They are called Intel. :p

We have seen it time and time again - penny wise pound foolish. And odds are it will bite you in the butt as seen by Circuit City, Radio Shack, Sears, and many more.
 
Last edited:
As the ATM example shows that "service life", thanks to that "lowest risk path (and least expensive)" mentality you were on about, can be where the least expensive part causes the "lowest risk path" to go into dodo land. As I noted in a post a while ago one of the museums I surveyed for my thesis in the early 1990s was using an 8088 to catalog their collection

Windows XP Embedded is an end-of-life OS with no further security patches.

x86 is not an end-of-life CPU architecture. Deciding, today, to stay on x86 is a reasonable decision. Even deciding to move to x86 can make sense.

So, while I agree with your general point not to stubbornly stick with old tech, your analogy here isn't that great.
 
The new ultra slim form factor that the arm chips bring will probably be the big selling point

Imagine an iMac that’s as thin as the lcd screen ....
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Back
Top