Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Kuo: Apple to Launch Several Macs With Arm-Based Processors in 2021, USB4 Support Coming to Macs in 2022
- Thread starter MacRumors
- Start date
- Sort by reaction score
Why not? The question is whether they will reduce the cost of macs because of Arm inside. The response was “iphones are expensive.” My response is that they would be even more expensive except for the fact that Apple uses its own cpus.
Apple uses its own chips in iPhones and they cost as much as laptops. How much do you think their laptops will cost?
Yup. Even a billion dollars R&D is only $5/device when you amortize it across 200 million devices. But just getting away from Intel is a win; they’ll be able to be in control of the schedule.
I would think Apple will start to add more complicated instructions where it makes sense to do things in silicon vs. ARM code. It seems like there are quite a few operations (vector or otherwise) that could be accelerated with some well chosen instruction set extensions, but I’m certainly no expert.
Yet they don’t cost more than comparable Samsung phones.
You can’t compare apples to oranges. Appleas CPUs cost Apple 40-50% less than buying Intel CPUs.
[automerge]1585695548[/automerge]
Yep, I’ve long theorized that apple will increasingly deviate from Arm (at least many of the Arm extensions). No need to be compatible with anyone else.
For every $30-40 Apples saves in BOM cost they can reduce the selling price by $100.
Among all the claims of pros from moving to ARM, the cost reduction is the one that makes me laugh most. Yeah perhaps for Apple it's good thing, but for users, what's point in saving little bit in CPU which already constitute minor portion of laptop itself? I mean, each part wise, CPU price is probably the most expensive parts, but as whole, it's not that big portion. What's CPU cost? 300~500 depending on the model? Is iphone less expensive than top of the line Android? All the cost users have to bear due to architecture switch will surely surpass whatever benefits. You pretty much will have to switch most of apps you know that right?
Cpu is a big portion of the cost. At AMD we only had design wins for years in low cost machines because our price was lower than Intel. Makes a huge difference in price of the machine.
And, for the license Apple has, they are allowed to freely add non-published extensions to their implementation. I can’t find the article now, but I’ve read that they have already implemented non-published extensions in their current products.
This is the kind of thing you read from somebody mocking up an engineer, don't spend your time with whoever is this guy, he is not an CPU architect.
All this rumor it's clickbait, the arm Mac is on sale now, just it doesn't name on Mac, name it iPad pro
Apple never compete with itself, any arm Mac will be just an sub-par competitor to iPad.
Apple moving to arm will require then to design at least 3 CPU families just for macs, (low power high efficiency, middle power optionally efficient, and high power compute) with at least 2 tier, means at least 6 expensive SOC to develop each year, only lower end models maybe borrowed from iPad/iPhone, but not for iMac iMac Pro MacBook pro it will need custom silicon period.
So to support this adventure (fanboy fetiche) Apple would require to invest 3-6 times what they invest for iPhone/iPad, for a product line that just represent 10% of the income.
C'mon
Then apple has AMD (a corporation where apple holds some stock position) with it's x86-64 compatible Zen family (Ryzen, Threadripper, Epyc) a CPU build with the same process and at the same foundry Apple baked it's Ax and Tx, Zen is now the HALO CPU every enthusiast wants, Zen require a chipsets manufactured by... ASMedia Zen could be licensed just like arm, also apple could reassign exedent waffers to ensure product supply, Zen supports thunderbolt 3 USB4 (USB4 requires no Intel certification, and it's the same TB3), More sales, more reliable supply chain, trivial macOS update (and being tested now as beta macOS leaked) try to explain stock holders why you don't sell Mac with Zen and instead you spend 6x more than on lead product for an not so appealing product, which also rivals own settled product line (iPad Pro).
lets te time to shut up mouths here.
Last edited:
But Zen is not licensed just like ARM by Apple. One would expect such a deal to be in place if they were custom designing a Zen for release in 2020/2021.
It's safe to assume that the chips going into their laptops are going to be more powerful than what's in the top end iPhone. I don't understand the point of this response.
Sure, if Apple only saves $30-40 on that CPU.
If they can save $150-200, you can knock off $500.
Maths ftw 🙂
But to be free of Intel’s logjam at 10nm; being able to control the schedule; being on TSMC’s 5nm process node next year; adding custom silicon to extend the instruction set to accelerate key functions: priceless.
Apple plans to launch several Mac notebooks and desktop computers with its own custom designed Arm-based processors in 2021, analyst Ming-Chi Kuo said today in a research note obtained by MacRumors.
Kuo believes that Arm-based processors will significantly enhance the competitive advantage of the Mac lineup, allow Apple to refresh its Mac models without relying on Intel's processor roadmap, reduce processor costs by 40 to 60 percent, and provide Macs with more hardware differentiation from Windows PCs.
Earlier this month, Kuo said Apple's first Mac notebooks with Arm-based processors will launch in the fourth quarter of 2020 or the first quarter of 2021.
Kuo expects ASMedia Technology to become the exclusive supplier of USB controllers for Arm-based Macs, adding that the Taiwanese integrated circuit designer will benefit from Macs gaining support for USB4 in 2022.
USB4 converges the Thunderbolt and USB protocols as part of Intel's goal to make Thunderbolt available on a royalty-free basis, which should result in wider and cheaper availability of Thunderbolt accessories like docks and eGPUs.
As USB4 is based on Thunderbolt 3, it offers data transfer speeds up to 40 Gbps, which is twice as fast as the bandwidth of the latest USB 3.2 Gen 2x2 specification. USB4's underlying Thunderbolt 3 protocol also means the specification supports up to two 4K displays or one 5K display over a single cable.
The simplest way to view USB4 is as Thunderbolt 3, but royalty free for manufacturers. Intel will continue to offer Thunderbolt 3 on a standalone basis with a few advantages over USB4, including more support with reference designs and technical issues for manufacturers.
The new USB4 specification was published in September 2019, giving Apple plenty of time to implement it in time for 2022 Macs.
Article Link: Kuo: Apple to Launch Several Macs With Arm-Based Processors in 2021, USB4 Support Coming to Macs in 2022
Somethings to consider, the A-series chips support the Arch64 virtualization instructions and Windows 10 ARM support has come a long way. I think the entire industry is banking on Apple.
Great. So people can run Windows ARM via virtualisation, and then run all their apps via Win32 emulation. Sounds like a ****ing blast.
Do we know any solid BootCamp numbers?
There are around 100 millions Macs on Earth. I wonder how many of them use Windows via BootCamp?
Apple knows they will piss some people off. But I'm guessing Apple thinks it's worth the risk.
There are around 100 millions Macs on Earth. I wonder how many of them use Windows via BootCamp?
Apple knows they will piss some people off. But I'm guessing Apple thinks it's worth the risk.
Bootcamp isn't the only thing to consider here. I'd imagine these days more people virtualise either Windows or Linux than use Bootcamp directly.
Gotcha.
Then I will add 'virtualization' to my "How many Macs?" question...
I'm just curious, that's all.
If Apple is going down this route of making entirely new chips and architectures... they know it will affect some people. I was wondering how big that number really is.
If 80 million out of the 100 million won't be able to do their jobs anymore... then that would be absolutely horrible. People will be storming the Apple Campus with pitchforks.
But I don't think the number is that big.
Silly side question: is the number of people who will be affected by the loss or reduced performance of virtualization and/or BootCamp larger or smaller than the number of people who were affected by the removal of the headphone jack in iPhones?
Because we know which side of the headphone-jack debate Apple landed on...
Then I will add 'virtualization' to my "How many Macs?" question...
It's hard to know, but there's enough market for Oracle VMWare (free, with some paid extras), Parallels Desktop (paid, in multiple editions) and VMware Fusion (also paid, in multiple editions)
Anecdotally there's apparently plenty of people who "switch", and need to run one or two specialised Windows apps, and use a VM for it.
Amongst developers of anything that has a server-side component (which is a lot of things these days), it's very common to run a development version of the server locally, either because you're working on the code that runs on the server, or because it's easier to debug interactions between the client app and the server when it's all local.
Not silly but I don't think it's apt, I think a more apt comparison (to removing the headphone jack) there would be dropping 32bit support in Catalina. You can run 32bit apps, but you need to run them in a VM. You can use a 3.5mm headphone jack, but you need to plug in a lighting to 3.5mm adapter. It's slightly less convenient but it's very workable.
Dropping x86 for Arm, would be more like removing all ports, completely. It's still not a perfect analogy but it's much closer IMO.
The theorised Rosetta 2.0 for binary translation/emulation of x86 to Arm would then be like if they offered a desktop dock that has a female lightning port on it (like an iPhone has now) and communicates with the phone over bluetooth or something. It will work for some simple scenarios but it's very far from ideal.
I think that Renoir series of AMD mobile CPUs is quite interesting especially for models where Intel won't compete with a 10nm for another year or two.
Leaving aside the desktops (that's also a good question), that's particularly the 16-inch MacBook Pro. It won't get an exciting update with Comet Lake-H. It also won't get an exciting update with Rocket Lake-H either. It'll only be with Alder Lake-H almost two years from now that several things move forward: new memory controller, PCIe 4 (meaning, among other things, potentially twice the SSD bandwidth), probably USB4, maybe Thunderbolt 4, a big leap in integrated graphics (so maybe less need to have the AMD graphics running a lot), and of course faster CPU IPC.
On other MacBooks, i.e. the Air and the 13-inch Pros (maybe 14), I actually think Intel is doing OK now. Ice Lake-Y was a great help for the Air, and if Tiger Lake-Y does ship later this year, that'll be another nice leap (in particular due to Intel Xe) graphics. Comet Lake-U would be OK for the small Pros, because it would allow them to move to six cores for the first time; Ice Lake-U would be worse for CPU but much better for GPU (hard to say which choice Apple will make here, now that Ice Lake seems to have finally reached volume); Tiger Lake-U would be the best of both worlds hopefully by the end of this year.
In short: if Apple wants that engineering complexity, it's plausible that they'll (temporarily?) move to AMD CPUs on the 16-inch MacBook Pro, but not on other MacBooks.
Renoir Mobile still stuck on pcie3, while I agree on AMD Macs, this year don't come Sense to move the MacBook to AMD Zen, a modest update with intel Ice Lake it's indicated, a move to Zen require an new T3 controller and Mobo, current MBP 16 and MBP 14 share chipsets it's motherboard are just following different traces, updating MBP 16 to the latest Icelake also brings an MBP14 on Icelake, and macOS support it's almost trivial, a move to Renoir mobile it's more risky now.
I see this year AMD Zen debut on iMac and mini (wwdc/Ryzen 5/7+rx5400 GPU), and later an iMac Pro (Threadripper, wx5700/5900), it's very probable a gaming iMac (an specially souped up iMac likely Ryzen 9+rx5700), but don't discar a compact 3-4 liter headless "gaming mac" either a trashcan re-born or a cube or a mini Mac pro (w/o pcie slots).
If course everything I've wrote above it's just my imagination I have no insiders within Apple. 🤫
So is anything Intel has to offer for now.
MBP 16 and 13 have basically completely different motherboards. The 16 has a QMS380 PCH; the 13 has no separate PCH at all, because the -U CPU includes one. They share some chips like the T2, and the Thunderbolt controller, but… even just their shape is wildly different.
The MBP 16 is not going to Ice Lake. There is not, and likely won't be, an Ice Lake-H. It'll likely go to Comet Lake-H or skip that generation altogether. Then, unless Rocket Lake gets canceled in favor of Alder Lake, it'll use Rocket Lake-H first.