But will these Macs run Windows programs out of the box? Or existing for-Mac intel software? If they don't I would never buy one. I don't want to go to Windows and I don't want another 68K to Intel mess.
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.
TSMC to Remain Sole iPhone Chip Supplier, Could Provide ARM-Based Chips for Mac in 2020, Apple Car Chips in 2023
- Thread starter MacRumors
- Start date
- Sort by reaction score
That's what everyone said about Watch, HomePod and Apple's TV streaming service too. Besides, what is this party you're talking about? Tesla's model 3 that they cannot even seem to build? There are no self driving cars that will be mass produced for quite a few years so it's a little premature to declare someone late to a party that hasn't even begun yet.
There are many things they could do. For example they could add an additional super heterogenous core in addition to vortex and tempest cores. This core could do much fancier things with branch prediction, threading, etc. to provide high single (or dual) thread performance for the highest priority user process. They could increase cache line size/bus widths, etc. Many ways to scale up from A13X to A13M.
What I suspect Apple will do is commission Microsoft to port Windows 10 to A-Series. Windows is already available on ARM, so Apple will work with MS to tweak it for their devices. Porting client operating systems these days is not as overly complex as it was in the 90’s and early 2000’s (at least it seems that way to me).
I am sure Apple will work key devs like Adobe and Wolfram and many others to have their desktop ports ready. iOS versions will eventually supplant others.
Not really me. My oldest is a Senior Process Engineer at a Fab here in the US. I learn a lot from them.
Nope, they won't do that. They didn't work with Microsoft for bootcamp, either. And Apple absolutely doesn't care if their machines continue to run windows. They would happily give up every single customer for whom that is an issue.
Here's what I found on the web about cars colliding.
I don't know how to avoid a collision. Let me tell you a crap joke instead.
"There are multiple benefits to custom-designed Mac chips, including no delays because of Intel's manufacturing issues, better profits, more control over design, and differentiation from competitors' products."
Sure there are multiple benefits for Apple to use their own chips on their Mac devices, but better profits is the number one reason.
Sure there are multiple benefits for Apple to use their own chips on their Mac devices, but better profits is the number one reason.
I do wonder what % of Mac users use BootCamp because it is that contingent that could suffer the most if Windows ARM isn't a viable product. As far as virtualizers go, they would either have to be re-written to emulate x86 architecture or use an emulation layer provided by the new ARM based macOS, neither option would lead necessarily to good x86 performance, unless the underlying ARM hardware and emulation layer are together ridiculously fast. That's really my biggest fear of the ARM migration.
I don't use BootCamp and Windows doesn't get near my work environment but Linux does and I run that in a VM rather than via BootCamp. If ARM-based Linux can operate well in a VM and there's readily available ARM binaries then certainly that would go a long way to easing the transition for folks with similar environments.
I don't use BootCamp and Windows doesn't get near my work environment but Linux does and I run that in a VM rather than via BootCamp. If ARM-based Linux can operate well in a VM and there's readily available ARM binaries then certainly that would go a long way to easing the transition for folks with similar environments.
Well, we don't completely know what the A12X has yet, but I'm talking about something completely different. For now we'll just call it a B1 since M for Mac naming convention is already in use by the motion-coprocessor. Imagine something completely different designed by Apple's chip team wizards who know the requirements needed on a desktop and what to put into it. That's what I'm looking forward to, and that's why I've been wondering why the Mac Pro has taken this long.
I think they will always (or for a long time) continue to use Intel for their highest-end systems.
But for the sturm and drang (where price and battery-life are king), the ARM-based Macs will be a serious game-changer for the entire industry. Windows 10 64 bit has already been Ported to ARM; so there's the last excuse to stay on Intel G-O-N-E.
Intel's days of being fat and happy are about to come to a rude end, mark my words.
And there is literally NO ONE on the planet with NEARLY as much experience or skill in ARM development than Apple. Period.
They will likely have to RECOMPILE everything, unless there is a "Rosetta"-like Just In Time compiler, and/or another round of "Universal Binaries" (which would still require repackaging at the very least).
But Apple is the hands-down MASTER at seamless transitions; so I expect it will be just as uneventful as the 68k -> PPC -> Intel transitions in the past...
[doublepost=1539810184][/doublepost]
Apple has MUCH experience at the "platform change" game. If anyone can make it be a seamless experience, it is Apple.
Oh, and since 64 bit Windows 10 already runs on ARM, I would say BootCamp is a virtural (no pun) certainty.
While a Mac supply chain analyst expect this doesn't demonstrate much depth from the semiconductor business standpoint. There are several dubious parts here.
First, the iMac Pro and MBP touchbar versions all manage to ship with a "ARM" chip inside of them without displacing the main x86 chip. So Apple could put "ARM chips" into all of the Macs in 2020 and that would say very little about which chip macOS was running on top of ( x86 ).
As for the points:
' no delayss of Intel's manufacturing'. Errrrr, there are pragmatically two x86 suppliers for classic style PC systems. AMD and Intel. AMD CPUs are already on track to being on TSMC's fabs by 2019-2020. If supplier one of some product isn't working out the easiest pay for Apple to do is switch to supplier two.
'better profits'
The current Mac line up uses about 4-8 CPU+GPU configuration variations. The iPad Pro (while a close derivative to the iPhone SoC) uses one and only SoC across the whole product line. Doing a substantive custom chip design at 7, 5, lower nm is in the multiple $100M range. Chopped up into different configs the individual models in groups of 1-3 don't have anywhere near the volume the iPhone or even most iPads have. Low volume of custom 7, 5, lower nm chips means higher costs not lower ones. (and higher profits). The primary way Apple would get to better profits would only be to use exactly the same SoC in the iPhone ( or perhaps the iPad Pro ) in the "Macs".
Apple could 'split' the Mac product line Macbook/Mini vs MBP , iMac (Pro) , Mac Pro. That is control but to what benefit.... it is greater complexity.
'more control over the design' Does Apple really want control over a desktop and/or workstation CPU ? Stuffing an ARM chip into the current MacBook chassis is one thing. It has one and only one USB Type C port. Rumors seem to indicate that the iPad Pro will join it in that status. Apple could certainly do an "iBook" product. Bring the name back. slap iOS on it , and have an alternative mainstream clamshell design alternative to the iPad. That wouldn't be a Mac though.
for example, Apple insourced some of Dialog PMIC work but also left some work outside for Dialog for other future products. They don't have to keep everything internal. Just the stuff that matters and is high enough volume to pay for itself. Mac CPUs and GPUs for the most part doesn't meet that criteria as you look across the whole range.
'differentiation from competitors' products'
This only makes more sense if it is actually more of the same than in differentiation. If Microsofts "Always on" Windows on ARM products take off in high growth fashion then Apple would have to chase them. I suspect they will initially try to chase with the iPad Pro but if that doesn't work then perhaps with an new "iBook". if that didn't work then I could see them trying a MacBook derivative.
However, a 'different to be different' rolling out the T-series across the whole Mac line up would be very different from the MS Windows world without replacing the main CPU at all ( just a substantially boot and authentication process differentiated on security dimension . )
But different in that "can't cross boot into Windows" ... that will probably cause more damage than get them new users. Same if make cross boot into Linux harder. ( There are ARM builds for Windows and Linux but not as equally supported as the x86 ones. that's what 'harder' is about. You don't get Windows on ARM from a download from Microsoft's site. It is a secured, authenticated boot. )
Again more than rather loopy.. The only thing separating Class 5 automony from current cars is CPU 'horsepower' ( magically gotten to by lower process sizes )? Process shrink size and Class 5 are two wholly different dimensions of work. ( More compute with lower power requirements can help but stuffing it all into one chip isn't the primary issue. )
Cars should be primarily inferencing, not learning. If Apple's software doesn't "know" much it won't be class 5. And it won't matter that the chips are 5nm or not.
Last edited:
"Only TSMC's 3/5 nm process can meet Level 4 and Level 5 chip requirements."
Why? Smaller die sizes don't create magical new powers. Cars are big, there is plenty of room for all the computer chips they need at normal die sizes. Why would the die size have anything to do with whether the chip can handle level 5 automation?
Why? Smaller die sizes don't create magical new powers. Cars are big, there is plenty of room for all the computer chips they need at normal die sizes. Why would the die size have anything to do with whether the chip can handle level 5 automation?
Apple didn't build Rosetta. They just licensed it. That isn't experience. That is like someone has experiece with building an Operating System solution because they bought a Mac with macOS on it.
Apple doesn't have that expertise inhouse. At this point IBM isn't going to license another crack at Rosetta tech either. (IBM acquired it around the time Apple dropped it. )
Windows 10 ARM already runs x86 Applications. It doesn't do EMULATION (which IS slow!); but rather does Just In Time (re)compiling, which is kinda slow the first time a program launches while the compile happens, but after that, is essentially just as fast as ever (because it is now ARM-native).
This is the technique that Apple applied with GREAT success in their 68k -> PPC transition, and it was SO successful that it was many generations of MacOS (Classic) before the last 68k code was actually expunged, and pretty much no one ever even noticed.
I don't think Apple likes to sit around with a split architecture like that. They moved very quickly to transition every Mac to Intel in only 14 months, and shipped Leopard only a little over a year later, which was the last version to support PowerPC. This is why I'm hesitant to buy a new Mac desktop for use at home. I don't want to spend thousands of dollars on an iMac Pro or Mac Pro and have support dropped for them so quickly. I was lurking these forums around that time before later joining, and there were a lot of furious people with really expensive G5 gear. I'm hoping the transition will be smoother this time, and I think it will be for things like apps. But I just don't want to regret buying outdated Intel chips which may lose support a few years. The reason I want to buy a Mac Pro is so I can upgrade it somewhat over time and keep it for 6-8 years, but perhaps I should just buy a cheaper 5K iMac to make it 3-4 years until macOS drops support for Intel.
The sooner the world breaks its reliance on Windows the better. Its become a bad joke for many years now. If anything apple providing X86 and bootcamp in macs is more of a promoter of Windows than the macs themselves.
While bootcamp/intel macs have provided apple with sales by thinking users would use bootcamp as a stepping stone into the mac ecosystem, instead it created a segment of users who still continue to rely and upgrade windows. This can be seens by all the posts that threaten to leave apple if they go arm. Not sure why they buy macs in the first place.
Ripping off the bandaid hurts, but its better to do it fast. Companies like intel and microsoft have been sitting on their butts holding back the industry for too long.
While bootcamp/intel macs have provided apple with sales by thinking users would use bootcamp as a stepping stone into the mac ecosystem, instead it created a segment of users who still continue to rely and upgrade windows. This can be seens by all the posts that threaten to leave apple if they go arm. Not sure why they buy macs in the first place.
Ripping off the bandaid hurts, but its better to do it fast. Companies like intel and microsoft have been sitting on their butts holding back the industry for too long.
I know Apple didn't design or build Rosetta. That's part of the reason it wasn't that great. They just needed something that they could get to market-with quick, and Rosetta was available.
But they DID home-grow the earlier JIT-based 68k -> PPC "emulator", and that worked a TREAT!
[doublepost=1539812925][/doublepost]
I hear ya.
I bought my first new Mac (all my previous Macs were eBay-acquisitions) in 2005, a 1.8 DP G5 tower. WHAT a machine!!!
Too bad the Intel switch was announced about 6 months later...
Anyway, I think this situation is a bit different (no pun). There is still a significant performance-gap between the highest-end ARMs and the high-performance Intel chips. That will likely be true for another 5-8 years. And who knows what Intel will dream up when they start to feel the pressure from ARM-defection. Remember, nevermind Apple and macOS, Windows 10 64 bit already runs on ARM, with a Just In Time compiler that provides backwards compatibility to x86 applications. So, Intel is probably already pooping their pants. Don't expect them to go quietly in that good night.
That's why I think that Apple will keep its options open, platform-wise, for longer than usual this time around.
But since the computer industry is ALL about change, don't get analysis paralysis waiting for "things to stabilize"; because they never will. I had to come to that same place myself.
Well if actually look at what the first post of the thread actually talks about it is far more about what "Kuo said" than it is about TSMC. Kuo used TSMC to drive off into the "what Apple is going to do" swamp.... so the thread is going the same way.
That TSMC is the only supplier is kind of a 'duh' observation. Global Foundries dropped out of 7nm business.
"To ensure that GlobalFoundries remains competitive against Samsung Foundry and TSMC in the long run, Sunjay Jha obtained IP and development teams from IBM (along with two fabs and a lot of obligations), ... "
https://www.anandtech.com/show/13277/globalfoundries-stops-all-7nm-development
That leaves Samsung and TSMC as the choices ( Intel possibly if the cleaned up the huge mess they've made and get their fab outsourcing business back in shape. That will take several years. ). The core problem here is that 7nm and then 5nm is going to be hugely expensive. It is largely only going to work if almost everyone only goes to 1 or 2 places to get all the work done. That's been the trend in the industry. Went from dozen options down to just "maybe" three. And just one place to get the fab equipment from for those three. 5nm costs so much have to drop most of the money in an extremely narrow set of place by alot of players so can afford to buy it. it is a game of 'music chairs' where the fab vendors disappear as the song goes proceeds along. [ it isn't so much the one vendor is so much better than others ... it is that folks are pragmatically being taken off the board due to rising costs. ]
P.S. and don't need to pay an analyst lots of money for an 'insight' report on how the game of music chairs works. so it isn't about music chairs.
Last edited:
"The days of Windows mattering are done."
I disagree. 90% of client-side computing in the enterprise is done via Windows. As Apple's presence within the enterprise increases, maintaining compatibility with Windows will be important.
yes.
really want (need, actually) a highly efficient iOS-based MacBook Air type of machine running cross platform apps.
even Adobe is now on board with this (it seems).
i want a machine released latest 2019 September; with a big pre-announcement around 2019 March to get app developers on board.
this transition is not about Intel and what it can or cant do or whatever.
its about maximizing the potential of a unified platform across a variety of screen sizes.
in the end, the ultimate goal is to have paralleled ability across any/all apple device.
and the major decision the user makes is to decide what level of portability s/he needs:
jeans right side pocket; messenger bag; etc.
really want (need, actually) a highly efficient iOS-based MacBook Air type of machine running cross platform apps.
even Adobe is now on board with this (it seems).
i want a machine released latest 2019 September; with a big pre-announcement around 2019 March to get app developers on board.
this transition is not about Intel and what it can or cant do or whatever.
its about maximizing the potential of a unified platform across a variety of screen sizes.
in the end, the ultimate goal is to have paralleled ability across any/all apple device.
and the major decision the user makes is to decide what level of portability s/he needs:
jeans right side pocket; messenger bag; etc.
Last edited: