It's not about the chip size or make up. Intel could compete with Apple silicon, it's the operating system. X86 windows is the real problem. I've built windows desktops for years, and recently went back to Mac with a base M1 Mac Mini. I think the current CEO of Microsoft is looking towards windows in the cloud, with subscription. Microsoft realizes that Windows has become so bloated and windows 11 is not liked by their user base. Personally I may run Linux on my windows desktop to see how it is. I know people complain about Mac software, but I could put together a desktop in about a hour, and spend hours just trying to get all of the windows updates.
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.
Intel Looking to Mirror M1's Manufacturing Process for 'Meteor Lake' CPUs
- Thread starter MacRumors
- Start date
- Sort by reaction score
By 2023, Apple silicon will be more powerful. But still good to see the competition.
It's in a preview, but Parallels supports it ?
Attachments
They are the sole source of one of the principle 'ingredients' . It isn't the only principle 'ingredient'. While the ASML scanners are a couple $100M a pop, a full fledged fab with all the other tools needed is several billion. There are still very expensive pieces that ASML doesn't do.
And ASML will make more long term with lower risk to themselves when they have multiple customers competing to get access to their scanners. That keeps the scanner price up. They can't charge themselves super high prices for the scanners. And have to directly deal with making chips designs work for a much larger set of customers.
One of the reasons players are falling away in the game is that the cost of production is so 'crazy' high there is an every increasing need to spread the costs over more companies and more high margin products. That's another reason why ASML shouldn't declare themselves king . If most of the customers vote "thumps down" and move their money elsewhere ... they won't be king anymore.
[ Blocking the Chinese from getting ASML scanner will likely over the long create create a competitor. Any missing tech will get 'spied on' and thrown enough money and resources at it ... it is all just physics and engineering . ]
Also why it didn't make sense for Intel not to have a very serious fab service offering. Even at their historical margins and revenues it will be too expensive in another 2-3 iterations. Intel needs to be a little less greedy and help other folks make money so they can contribute to future Intel fab capital costs and R&D also. No way to dominantly sell everything profitable to everybody directly at 50+% margins over multiple decades at this point.
The GPUs haven't been on the bleeding cutting edge. Nvidia and AMD are about to TMSC N5/N4N/N4P after Apple has been on N5 for a while. Mediatek has a cellphone SoC that is N4. Intel's has N5 compute tile on the HPC compute gpGPU ( although not in volume) . Because Intel is doing relatively very small iGPU dies there is good chance they will be first on N3, but the large discrete dies will take substantively longer.
GPUs are getting closer to bleeding because the average selling price has crept up. But the refresh cycle is still relatively long to what the cellphones folks are doing. Nvidia and AMD will move up at end of this year but then they'll stay there for a while. In part, AMD/Nvidia probably don't want to make super high volumes of GPUs. [ even more so if GPU driven crypto collapses and there is a glut that washes over the resale market. ]
This all dGPUs on TSMC is a temporary thing. Samsung and AMD have agreement for the Samsung SoC iGPUs. AMD is going to be doing some work on Samsung process nodes. Samsung made a bet that they could finish transition to "gate all around" before everyone else. They may loose that bet , but if they don't AMD likely will be back (especially if they too go to disaggregated dies that are packaged up for a GPU package. )
At major fab technology infliction points TSMC has had hiccups in the past. They are not guaranteed to make the "gate all around" transition smoothly either.
Apple has not 'nixed' macOS Intel. In fact, they have stated that macOS Intel will be supported for 'many more years to come'. That is likely not 8-10 years into the future. But it is also extremely likely not 1-2. There are 100M Intel Macs out there. They still work reasonably well. This isn't like the 2006 transition. Two major different factors in play.
1. An installed base of approximately 100M versus of about 30-40M. It is 2x to 3x as big of an inertia factor this time. Sales of M-series Macs are quite healthy, but there is no stampede off of Intel macs. ( as usual Apple bragged about huge chunk of M-series buyers being "new to Mac" buyers. All sales to those folks does nothing to drop the Intel install base count). It will take several years to work that down. And longer than it did for the PPC or x86 transition.
2. The transition from PPC-> x86 happened in about 18 months. Apple has taken sustainably longer this time. September 2022 Apple will probably still Intel models still for active sale. As long as they are selling Intel models brand new there is no way they can turn around and kill in a "year or two' without a huge public debacle (and likely class action suit).
Apple might do a "sneak peak" of a Mac Pro at WWDC but pretty doubtful they will ship anything before late Fall.
"... Products are considered vintage when Apple stopped distributing them for sale more than 5 and less than 7 years ago. ..."
https://support.apple.com/en-us/HT201624
The countdown to 'de-support" for Apple starts when the stop selling something. They are still selling so the countdown clock hasn't even started. Cannot possibly be 'nixed' when the clock isn't even running.
3. Apple didn't own Rosetta on the first transition. Apple rebadged Transitive's tech . Which means they had to pay per machine sold. IBM bought Transitive right around the time Apple dropped macOS PPC. Apple wasn't going to get lower terms out of IBM ( not going to beat them up at the lawyer table. ).
This time Apple owns their Rosetta tech. It works and not likely they are gong to spend more R&D fixing bugs or extending the emulation coverage. So it costs not much to keep it around. In 10 years they may have left it in zombie status for so long that can't easily fix a critical bug even if they wanted to. So it won't stay around forever, but die in a year or two.. very probably not. ( they spent a ton of money making (there are augments even down to hardware level for this so it was in no way shape or form "cheap" . They need to make the money back as a feature for sale. )
Better at what? Around the PPC->x86 transition was the point at which Mac laptop sales started to dominate desktop sales.
Power6 and Power7 where better server chips. They were not better iMac or laptop chips. Those were and still are the bulk of Macs sales. The M1 mini is probably doing better nowl as a contributor to overall unit sales, but that that just skews the point even more.
Apple used their own I/O chip as a 'firmware dongle' on Macs. Apple did relatively little to promote the general PPC ecosystem breadth and growth. So no it wasn't surprising macOS didn't run elsewhere. That was a contributing factor. Apple didn't want to pay what it cost to be best in class for the segment their systems are/were in. So IBM skews Power to what paid the bills (their server boxes with Power and Z-series ). Money talks.
That's a fairly long time. That might run into major problems before the macOS Intel updates run out. When all the pre-T2 macs are put onto the obsolete list there is chance they'll hook booting to having a Apple T2 present. At that point in bad shape ( and may need some pretty hefty hackery). More likely most folks will just stop on that last non-T2-only version and declare it good enough to run forever.
The other big problem is likely lack of future GPU drivers. 5-6 year from now running 6-7 year old GPUs isn't going to be very attractive to most hackintosh users.
Last edited:
Windows 11 is partially not liked by some because it does get rid of some of bloat. It throws some 16 and 32-bit 1980's and 90's baggage overboard. It cut off older than 2016-17 era CPUs (that have weaker security and virtualization support). Pragmatically chops out legacy BIOS ( more strict UEFI without "get out of jail free " cards to lapse back into legacy BIOS properties. )
Same thing though on x86 hardware side though. Does x86 really new 4-6 different SIMD instruction subsets for modernly maintained and developed and optimized Windows 11 apps ? Probably not. Run 1990's vintage code in "raw" mode? Probably not.
I suspect that if x86-64 wants to remain relevant into the 2020-2040 era that there will be some "house cleaning" around the time that Windows 10 gets de-supported in 2025. At least for the CPU packages marketed into the WindowsPC market. [ can still have "I want to run code from the last century" CPU packages from AMD/Intel for a narrow subset of users , but they don't need to throw that baggage on the general PC market systems. ]
Hours for Windows updates has more to do with the inital image you start with than with Windows bloat.
Last edited:
they are not competing with apple, they are competing with AMD
Intel is considering manufacturing its 14th Gen Core "Meteor Lake" CPUs at least partly using TSMC's 5nm process to mirror Apple's M1 series of chips, DigiTimes reports.
Announced last year, Intel's Meteor Lake chips will be its first multi-chiplet design to integrate an application processor, graphics processing unit, and connection chips into a single Intel Foveros advanced package.
Intel initially said that the Meteor Lake CPUs would be manufactured with its own 7nm process, which it calls "Intel 4." At its most recent earnings call, Intel CEO Pat Gelsinger said that Meteor Lake would be the first Intel product made using Intel 4 and that prototypes had already successfully booted Windows, Chrome, and Linux.
Now, sources speaking to DigiTimes claim that Intel is considering placing orders for all of the chip blocks used in the Meteor Lake CPUs with TSMC, Apple's sole chip supplier. Rather than singularly rely on Intel's in-house 7nm process, the Meteor Lake chip blocks would be contracted out to be manufactured with TSMC's 5nm process, just like Apple's M1 chips for the Mac. The move would reportedly help avoid delays to the CPU's production and launch schedules.
The A14 Bionic, A15 Bionic, M1, M1 Pro, M1 Max, and M1 Ultra chips, are manufactured using TSMC's 5nm process, and Intel's decision could help Meteor Lake chips to better rival Apple's competitive custom silicon. TSMC is also said to have now received chips orders for the iPhone 14 lineup using its 5nm process.
The sources added that potential orders for Meteor Lake CPUs would be large enough to encourage TSMC to scale up its fabrication capacity for its 5nm chips by the end of the year. Intel's Meteor Lake CPUs are set to launch in 2023.
Article Link: Intel Looking to Mirror M1's Manufacturing Process for 'Meteor Lake' CPUs
All those arguments sound great but there was millions of installed base of PowerPC hardware and yet Apple threw them under the bus.
After several years.
Last release of mac OS X 10.5 ( Leopard) version number update was in August 2009.
Mac OS X Leopard - Wikipedia
Got security upgrades while 10.6 was the primary release. !0.7 didn't arrive until 2012.
Intel Development kits went out in 2005. 2012 - 2005 is 7 years. That is well past being some "short drag race to termination". there is about zero historical back up to the notion that it is being nixed in a couple of years.
Even is want to toss the security upgrade window it still 2009-2005 ---> 4 years. Which is still double "a couple of years".
Is it going to go far past a seven year window? No. If Apple only ships a "half sized" Mac Pro in very late 2022 and keeps selling the full sized MP 2019 into 2023 then it will probably longer than seven ; into the 8-9 year range.
The hardware and software of a Mac is generally viewed as a combined whole by Apple. macOS instance is licensed to the hardware. That is even more true now than back in the old days when Apple sold mac OS X DVD/USB. Users are just billed upfront for upgrade costs.
The other issue that folks gloss over from the last transition is that Apple was taking the mac OS X Mach Kernel 3 wide at time PPC , Intel , and Arm ( for iOS based on the same common base). There was pressure to dump one of those to get back to a "just two". Here there is no such pressure at all to drop down to "two".
P.S. I suppose there will be some hand-waving by some about how OS updates were on a longer cycle back in the old days. We get annual updates now , but there is little indication that is gong to make any material difference. If Apple paid PPC developers for a block of 4-7 years amount of time, they can do the same this time also and it won't relatively cost any more than it did then ( more stuff but larger base paying for more stuff). The costs and the sizes of the installed base matters. Apple monetizes the installed base with subscription software sales. Making a large fraction of those folks mad so they turn off their subscriptions as bolt to Windows is actually a negative hit on Apple's revenues. Apple has already charged for the upgrades. They can just recognize the revenue over time.
They can also cost control as the macOS Intel isn't going to get all of the "super nifty" new features over time that depend upon proprietary or new hardware. Costs very little to skip deploying features to the Intel branch.
Last edited:
So now Apple has invented ordering chips rom TSMC? Even though they were probably 500th TSMC customer (timeline wise).
Yep. And that’s when we were lucky enough to get an Intel chip. Sometimes all we got was “Intel delayed their new architecture, soooooo…. No updates this year! Enjoy!”
Nah, that’s not what I mean
I really like the use of the word Mirror in the context.
Don’t get me wrong, I hope CPU and GPU market has rough competition as it develops faster, and hopefully cooler and less power hungry hw.
Apple also “mirrored” tons of ideas during their time.
