This is nothing like the transition to Intel, which offered a huge benefit to the 95% of users who were trapped in the Windows ecosystem.
In the last 8 years or so, Apple's big ideas for the Mac have been complete duds (the trashcan Mac Pro, their entire laptop lineup from 2015 to 2020...)
Mac OS X recently dropped support for 32-bit software, and this should have freed up their OS engineers. Instead, the most recent release of Mac OS (10.15, Catalina) is widely viewed as one of the worst releases in recent memory.
I don't want an ARM Mac. I want Apple to refocus on the basics of Mac hardware and software.
Your argument doesn’t make much sense to me. I want them to focus in good software, too. But Apple likes to make all the key elements of their stuff, and this would give them significant independence.
Moving away from Intel sounds like a huge benefit to me due to the number of ways they have held back computers over the decades. I believe that ECC memory, for example, would be standard in all desktops and laptops today if Intel didn't use it for market segmentation. That doesn't even touch on all the security issues.
> I believe that ECC memory, for example, would be standard in all desktops and laptops today if Intel didn't use it for market segmentation.
This is an extraordinary claim. Don’t get me wrong, I love ECC, but it has a very real cost associated with it. Why would any alternative to Intel change the pricing dynamic?
We can't know, because choosing ECC or non-ECC based on price vs features isn't a choice most Intel consumers get to make. I do personally think the benefits would have proven themselves worth the cost over the previous couple of decades if The Market wasn't artificially constrained by an intentional limitation in the processors 99% of people use. Prices would come down with additional volume too.
The memory controller supports ECC if Apple cared about it they would’ve forced Intel’s hand a long time ago it would not be particularly difficult for Apple to request a custom SKU with ECC enabled.
> the number of ways they have held back computers over the decades.
Nonsense. There's a lot of valid reasons to be critical of Intel (largely stemming from general malicious business practices and greed), but suggesting they haven't been pushing computing forward and being market leaders and innovators is just plain untrue.
They were making 8c/16t Xeons back in 2010, and core counts there have continued to improve on the high end, and increment and propogate across their stack. It's taken till recently for most consumer software or usecases to really leverage more than 8 threads as well, so the demand on the consumer side wasn't really there. If it were, the AMD Phenom's likely would have been more successful given their multithreading performance advantage over their consumer intel contemporaries. Instead they delivered gains in performance per watt, lithography, IPC, idle performance, and instruction set/acceleration gains (AES-NI, AVX, QuickSync, etc), to say nothing of pushing the market forward in Memory and Storage as well. Clock speed gains were realized as well but pushing much passed 4Ghz starts pushing into limits of physics/thermodynamics given current fabrication tech/science.
ECC gives no benefit to most consumer applications, and actually would more likely be a detriment in most implementations, as it increases costs and limits performance while it's two primary benefits would largely go unnoticed if not entirely unused (error correction and improved total capacity). That's like suggesting you should make the entire frame of a car out of Titanium. Sure, it's technically better, but it's gains aren't worth it's costs and largely lost on most consumer buyers anyway, so why bother.
> That doesn't even touch on all the security issues.
No argument from there. Worth pointing out that exploits will often target the dominant marketshare holder of any given industry/segment (eg, Windows for OS, Android for mobile OS, Intel for x86 CPU's), but that doesn't discount the flaws that intel had implemented and their handling of said flaws/exploits. But that's not holding computing back so much as a matter of corruption/morality.
Intel is lagging behind AMD. If anything, Apple should jump ship for AMD and not have to screw-over their entire install-base and everyone's software and hardware investments. Going to ARM would be extremely stupid... an iPad isn't a laptop and vice-versa, and trying to force them to merge prematurely doesn't speak to any sense of product life or support.
You made a strawman regarding titanium. For as much as the recent Apple computers cost, a MacBook Pro should have ECC memory and plenty of ports, while a MacBook should not.
There is a possibility if this tweet is anything to go by, https://twitter.com/_rogame/status/1225381275617415168 and apparently it seems this is being talked about since Nov'19 when the codenames appeared first time in Mac betas.
There is something seriously wrong with their engineering vision when they can dedicate developers to adding new emoticons to iMessanger when they don't even have Time Machine backups to iCloud for MacOS.
I think the bigger priority should be stop adding features / taking away the ones we use for a UI refresh, and spend some serious time fixing their giant, growing backlog of bugs. macOS and iOS/iPadOS are progressively getting worse in too many ways to enumerate and some bugs have been around for many years. Furthermore, they don't appear to (or don't have time to) listen to you or I submitting a bug report, so most people don't bother anymore. It's a shame because their products have gradually become like American healthcare: more expensive and worse. I think Apple needs a new CEO who can bring prices down, and quality, durability, and innovation up.
MacOS has built-in file versioning since Lion. Whenever you save a file, it creates a version. You can browse and restore an old version using the program that created it.
Admittedly, this only works for applications that use the API. For anything in plain text you should probably be using a git repo though.
You have some good points, however Apple has allowed (even if temporarily) third-parties to exist on the periphery to fill-in solutions. TarSnap, SpiderOak One and BackBlaze all seem like viable options. Perhaps even Carbonite too.
I really enjoy working with the Mac and am very productive with it, but this would most likely represent the end of the road for me to use it in a professional capacity.
One of the tools that I require to do my job, a development environment for an ERP system, is a Windows application. It's a closed-source commercial application. The likelihood of it getting ported to MacOS is zero, so it's always going to require me to run Windows in some capacity in order to use it.
There are three ways I see this tool working on an ARM-based Mac:
* MacOS emulates enough of the Intel instruction set that VMware still runs. And doesn't kill it in the next version like they did the PowerPC emulation after the Intel switch.
* VMware runs but gives me a virtual ARM CPU which runs an ARM version of Windows which emulates an Intel CPU. I'm not sure what the ARM virtualization story is, so this may not even be a thing,
* I keep an Intel machine under my desk and use RDP. This is somewhat cheating and it's unlikely that work would be willing to pay for two machines.
On the flip side, Windows 10 seems to be getting really good, so switching actually wouldn't be that bad. WSL will basically give me all of the command-line tools that I need, Hyper-V can meet my virtualization needs (though I won't require it like I do on the Mac), and all of my other tools are available on Windows.
There is a fourth... but thats even more speculation.
AMD, which also is x86 compatible, had a division a while back to produce ARM/x86 processors in one chip... maybe this could be a solution to running both systems at once?
This article is mostly speculation (although based on rumors from reliable sources), but they do speculate that the "pro" level machines will stay intel for a long while.
Fat binaries with slices for each architecture, much as was done for 32+64 bit x86 and is still done for iOS. Porting x86-only code to arm can sometimes be a lot of work, but supporting both x86 and arm at the same time isn't any more work.
Actually, I think the best option would come from Microsoft. This is only speculation, but they could release an ARM version of Windows 10 with a built-in x86 translation layer.
It’s not like you won’t be able to buy an Intel Mac after the first ARM MacBook is released.
I was helping a friend with her 2014 MacBook Air, which can still run macOS 10.15 Catalina, which was released a few months ago. Keep it mind it shipped with macOS 10.9.2—that’s six major releases ago.
Macs tend to last a long time; I think it’ll be quite a few years before the lack of serviceable Intel Macs will be a problem for users who need to run something that’s Intel-only.
If this worked as well as the PowerPC emulation in MacOS did then this wouldn't be a bad option. The Windows-only tools that I need aren't terribly CPU-intensive, so even a modest performance hit would be tolerable.
There were a handful of emulators that did this for windows back in the days of powerpc Macs. Softwindows and virtual pc come to mind; virtual pc IP is now owned by Microsoft and theoretically part of hyper v.
I'm sure it won't be as slow as SoftWindows95 was on my 66MHz PPC601 around the first time I lived through predictions of the Mac's death due to an architecture change :)
Regardless, one of the tools that I require to do my job is Windows-only and will always be Windows-only, so I will need to use Windows in some capacity.
Really? My Mac doesn't show ads in the dock for Candy Crush and Minecraft and Netflix and whatever else, and I dont have to spend hours cleaning trash off a new Mac install.
MS is branding themselves as "low end" junk by doing that kind of thing.
Despite the headline this is still just speculation. But it is a pretty obvious bet at this point. The iPad pros are so fast, and have serious physical limitations compared to laptop and especially desktop computers.
I’m excited to see what Apple’s processor team could pull off with the higher budget, better heat dissipation, and much larger battery capacity that comes with a MacBook Pro housing.
Agreed. I've thought for quite a while that Apple should/could put multiple ARM processors in a laptop and turn them on/off depending on need and power supply. This is the best of both worlds, in my view. Doing lots of heavy computing (gaming, etc.)? Turn on all 6 ARM processors! Working from the library and remoting into your server? Only turn on 1 ARM processor.
I could even see being able to choose the number of processors when you buy the machine. Trying to save money? Get the single processor option. Need to compile Rust code daily? Get the 6 processor model. Or get a low end Mac Pro today and upgrade it tomorrow with processor cards. There are a lot of options when you control the complete system; Hardware, Firmware, and Software.
Saw a rumor today that the next iPad Pro will have a keyboard cover featuring a touchpad. No idea how credible that is, but it would be interesting to see if the iPad Pro and Macbook meet in the middle.
The thing that worries me, and this is possibly unnecessary, is that Catalina is paving the way for the appification of OS X, turning into a version of iOS on the desktop, complete with sandboxes and silo’d apps; more like an iPad with built-in keyboard than a general-purpose laptop.
Jobs did not want iOS devices to create content, only to consume it, and with Apple’s focus shifting to services, I worry that they will continue this push for ever more consumption, to the detriment of all else. We see it in their overtaking of and reported indifference to independent developers. We see it their push to create an all-encompassing media empire, like Disney, but with a watch and handhelds. We see it in their push to develop ever.more expensive hardware that doesn’t have commensurate increases in functionality.
I still use Apple and love it, but the hardware and software hasn’t yet moved away very far from what it started out as after the switch to OS X. But I can also see the end of that in my lifetime, perhaps in the next 15-20 years. By that that time, I can see development limited to a few approved vendors with deep pockets for expensive licensing, but the bulk of it coming from Apple.
I still use Apple and love it, but the hardware and software hasn’t yet moved away very far from what it started out as after the switch to OS X.
You do realize Mac OS X 10.0 was released in March 2001. Here are the specs of the early 2001 iMac: It featured a 500 MHz processor, 64 MB of RAM, a 20 GB hard drive and a 56 kbps modem,was available in Indigo, Blue Dalmatian or Flower Power, and was priced at $1199 U.S.
So yeah, we've come a long way.
I can see development limited to a few approved vendors with deep pockets for expensive licensing, but the bulk of it coming from Apple.
I doubt this will happen; there's no point. The recent release of Swift Playgrounds [1] for the Mac has the potential to create a new generation of Mac developers.
Apple wouldn't have created Catalyst to increase the number of apps for the Mac by giving a path to iOS developers to bring their apps to macOS.
Darwin isn't going anywhere; you'll still be able to run all of the Unix stuff, with the vast majority just needing to be compiled for ARM.
Not to be pedantic, but sandboxed apps on macOS goes back to the 10.5 days. I just checked on my Catalina machine; most GUI apps are sandboxed, but none of the Unix apps are. By definition, sandboxing has to work differently on the Mac than it does on iOS.
There's no doubt that Apple has to do better; the 10.15 Catalina release was way more buggy than we're used to seeing. Apple seems to have a lot of balls in the air, with macOS 10.16 coming in a few months at WWDC, plus new iOS, watchOS, tvOS and who knows what else.
If the transition to ARM is going to happen about a year from now, that means there are alpha versions of macOS running on prototype ARM Macs right now.
Hopefully, the Catalina release was a wakeup call for them to get their act together; guess we’ll know soon enough.
I don't think this will happen. I believe (but do not know) that a RISC emulation of a CISC processor is much harder than the reverse.
If AMD had died and Intel's top of the line enthusiast CPU was a 10 core; I wouldn't blame Apple. They would be leaving a lot of performance on the table that they could get from TSMC.
AMD has made the x86 space much more competitive and likely saved x86 in the long run in the server space. Same with gaming enthusiasts. Some of them dual boot to use Windows software/games while keeping the mac in that creative space. It would just be much easier to source AMD parts. Especially since both of them use TSMC.
It is a lot of software effort at a time when the quality control of Catalina is buggy mess. A lot of software work to emulate and optimize a new CPU. I am seeing, at least youtubers, consider other platforms. Snazzy labs recently moved to Linux because they couldn't afford a Mac Pro, risk a Hackintosh, or use Windows 10 because their 10 gig networking didn't work well on Win10.
If Apple does switch over they better have the CPU horsepower to justify it. Their brand new Mac Pro is only 18 cores and AMD's 3000 16 core can match it on Windows. AMD is very strong right now and Intel has had fire lit under them for the past year that may cause them to respond soon.
It's easier to emulate CISC with RISC than the other way around. CISC instructions break down into RISC instructions but detecting which RISC instructions can be added together into a complex CISC instruction is much harder (though I'd note that RISCV attempts to fuse certain instructions as an optional optimization).
Bad news if you develop in Docker on a Mac and then deploy in the cloud. You probably are taking great advantage of the fact that your binaries work fast both locally and on the majority of cloud servers. There are ARM servers but they are still fringe and seem to be unsupported by other fun platforms like Google Cloud Run and similar stuff.
It really would be quite a drag.
On second thought, most web dev tasks aren't very CPU intensive and would work under emulation pretty much fine. And we all expect code to run faster in prod than on dev anyways!
In the 1995 movie 'Hackers' there was a scene (https://www.youtube.com/watch?v=RL9yCWv7NS0) where they're talking about RISC being the future. The hype about RISC was strong at the time but a few years later that scene seemed dated. Now here we are in 2020 and ARM keeps taking over more and more of the world.
While not mentioned here, I have seen speculation that if this were to happen Apple could offer an intel co-processor that could be used as a compatibility layer for old intel apps.
Does anyone with better computer arch knowledge know if something like that would be reasonable from a price and performance perspective?
"But this week, reliable Apple supply-chain analyst Ming-Chi Kuo reported that Apple plans on releasing a Mac with an Apple-designed processor in the first half of 2021"
I'd buy one of these, but it'd probably be priced in line with the current Macbook lineup
Jesus Christ...They finally put the escape key back and potentially sorted the keyboard. While I'm sure there is a consumer market for such a thing, this would really make a mess things as a developer's laptop. The battery is already maxed out for what you could carry on an airplane. They need to use some of that space they saved by welding all the components to the mainboard and cool the i9 so it does not throttle itself. (Crazy talk to even hope for RAM slots... cause who wants a system you could expand...)
uhh.. most brew shit is built from source, for which prebuilt binaries are often available. You’d just brew as usual building from source or prebuilt arm packages would be provided as usual.
> Wouldn't switching to ARM break all currently installed brew packages
(Emphasis mine.) Most Homebrew programs are not installed from source; and they'll require recompilation for the new architecture by the Homebrew team and a reinstall of all packages (to fetch the new packages) to work.
https://news.ycombinator.com/item?id=22410918