To be honest; its not all that obvious. Apple moving to Apple Silicon does not preclude them from supporting Bootcamp with an ARM build of Windows – or Linux etc, etc. Instead, Apple have decided that only macOS will boot on the Apple Silicon Macs – that's an entirely political decision.
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 Delays 7-Nanometer Chips Until Late 2022 or Early 2023
- Thread starter MacRumors
- Start date
- Sort by reaction score
To be honest; its not all that obvious. Apple moving to Apple Silicon does not preclude them from supporting Bootcamp with an ARM build of Windows – or Linux etc, etc. Instead, Apple have decided that only macOS will boot on the Apple Silicon Macs – that's an entirely political decision.
Apple has decided on virtualization as the way forward for running other OS's. To me, that's the right move.
—Edited to add that maybe the reason behind not enabling bootcamp is that the arm version of windows is not ready for prime time.
Windows 10 on ARM Compatibility Check-In - Thurrott.com
It’s been two years since I last evaluated Windows 10 on ARM application and driver compatibility. What’s improved since then?www.thurrott.com
I agree with you that there is no reason – technically – why Apple couldn't support Bootcamp for an Arm Windows build. But they're not permitting this: an Apple Silicon Mac can only boot macOS – Apple are cryptographically enforcing this in Silicon as they do on iPhones and iPads. (Stated that in the WWDC 2020 videos).
I think that Apple are seeing that the days of Windows' monopoly are now numbered.
There are indeed many application-specific applications for Windows for which specialists will require Windows – however for 95%+ of computer users there is no longer a reason why a Mac can't meet their needs.
This definately wasn't the case 15 years ago when Apple moved to Intel – and bringing Windows compatibilty to the Mac was a definite win. Jobs was right selecting the Intel architecture back then.
But times change and with more and more of our activities being Internet/Cloud based I think that Apple is feeling that their Apple Silicon Macs – with all the custom-os-and-silicon-goodness that no one else on the planet can provide – will be able to stand against that old Wintel monopoly and carve out a far, far bigger market share than the Mac has ever had.
[automerge]1595619577[/automerge]
Indeed. I'm big user of VMware Fusion (not Bootcamp!) and I'm glad that's not being cut off by Apple as part of the Apple Silicon move.
Of course, for great performance an Arm-based guest VM is required. However, people are currently running x86 based VMs on iPads using QMEU – so I'm sure there will be some (non-Apple) solution for x86 virtualisation for Arm Macs. The Market provides.
Last edited:
Best to check out the WWDC 2020 Video "Explore the new system architecture of Apple Silicon Macs" at around 16:00 where the engineer discusses how Apple Si Macs only support secure boot into macOS. (Using cryptographic controls as per an iPad/iPhone.)
Right. Apple will be gone and non-Apple companies will use 7nm Intel silicon.
The quote was this:
“We’re not direct booting an alternate operating system,” says Craig Federighi, Apple’s senior vice president of software engineering. “Purely virtualization is the route. These hypervisors can be very efficient, so the need to direct boot shouldn’t really be the concern.”
And MS has made noises about currently only licensing to OEMs but nothing more to say at the moment... I think it's premature to say all the decisions are made and final at this point though. They clearly have an intent to support alternate OSs in some form.
Farther back in time ( 2 or more ) years ago that was for very "low end" stuff. Tryin to make sub $80 CPU and embedded or low end chipsets.
Oct 2018
Intel to Outsource Entry-level CPU and Chipset Fabrication to TSMC
Intel is facing a manufacturing crisis, in which demand has far outstripped supply, and the company is firing up all its silicon fabrication facilities to manufacture 14 nm products, mainly processors under the Core and Xeon brands. We've been hearing reports since early-September of Intel...
"... The company collaborated with TSMC and other partners in the past, so the announcement just confirms that the company will continue to do so in the future even after it expands its own capacities, if third-party foundries offer the right technologies for particular applications. ..."
Intel to Expand Production Capacities at Multiple Fabs
That drumbeat about a year ago when volume 10nm slid into 2020. Although one chipset when 'backward' ( 24nm?) on Intel to move out of the way for 14nm.
The drumbeat has been cranking up over the last 12 months or so of lots more stuff higher up the food chain.
January 2020
like the mainstream (cost sensitive ) version of the DG2 GPUs
Rumor: Intel would shift Xe GPU production towards TSMC at 6nm
It has been silent around Intel Xe for Desktop. The new news is that Intel might have its Xe graphics cards manufactured by TSMC, fabricated at 6nm. This is a report that must be viewed with skepticis...
www.guru3d.com
Intel has also acquired two AI chips companies ( going to 'retire' one ). Unlike the modem acquisition there isn't much good synergy at this point of trying to merge the other designs started externally onto Intel's Fab process to 'soak up' more production. External stuff that Intel has bought will probably stay external. Depending upon cost structure wouldn't be surprising if some of the FPGA stuff when external too ( If Intel can get a large price premium then can roll out on a medium yield process and just soak up the defects with profit. ) .
Eventually dumping very high volume celluar modems was also suppose to free up fab capacity also ( but that was more of a sell off than an outsource. ). They'll still have to make some when the iPhone 12 arrives but the numbers of 11's will drop. That will help Intel a bit in terms of capacity log jams for ( 14nm - 10nm ) .
Yeah, people for some reason people seem to think a smaller company can't buy a bigger company.
Things like leveraged buyouts happen.
Ask Avago and Hock Tan how he bought Broadcom.
Money is cheap.
that's the wrong connotation. Elements of the Ponte Veccchio package were going to be 3rd party before ( certainly the HBM ).
"...As an example, our data center GPU design, Ponte Vecchio, will now be released in late 2021 or early 2022, utilizing external and internal process technologies, combined with our world-leading packaging technologies. We now expect to see initial production shipments of our first Intel-based seven-nanometer product, a client CPU, in late '22 or early '23. ...
...
...
Yeah. Yeah. On Ponte Vecchio, originally, the architecture of Ponte Vecchio includes an I/O-based die connectivity, a GPU, and some memory tiles, all kind of packaged together. That's kind of the design of Ponte Vecchio. ... "
Intel (INTC) Q2 2020 Earnings Call Transcript | The Motley Fool
INTC earnings call for the period ending June 30, 2020.
www.fool.com
".... The only 7nm part that remains (roughly) on schedule at this point is Ponte Vecchio, Intel’s Xe-HPC GPU that will be going into the Aurora supercomputer. That is expected to ship in late 2021 or early 2022, and even then Intel is evaluating whether to move the manufacturing of some of Ponte Vecchio’s parts to third-party fabs. ..."
I think Intel has already gotten to 'tape out' stage with the GPU core chip already. It is probably pramgatically aleady too late to shift to another process. Even more so if still targeting 2021 of devillery of at least a few packages and rack units for early acceptance testing.
It may not be to late to shift either the cache memory ( lots more replication of same design. ) and or the package I/O chip ( again less diverse logic and easier validation ) to perhaps a 3rd party. Each Ponte Vecchio package has substantially more than just as die with GPU cores on it in the package.
if Intel was going to get a high price for the Ponte Vecchio it could be useful as a 7nm "pipe cleaner" for Intel. It isn't a high volume chip package. (i.e, need thousands of the packages not millions There are multiple orders of magnitude between those. ). Anandtech estimate that there Aurora has
"...
...
On the GPU side, all six of the GPUs per node will be Intel’s new 7nm Ponte Vecchio Xe GPU.
...
...
- We get a rounded value of 2400 total Aurora nodes (2394 based on assumptions).
Even if Intel wasn't getting 80-95+ % years they only need 14K working GPUs to deliver. If can crank out 30-40K on a medium yield process that would probably work out OK for them.
even if only getting 25 dies per wafer per day and four machines in parallel ( for 100 good dies per day) it would oonly take about 100 days to get to the 2400. That is doable even with "bad' yields.
Intel can't do that for commodity products where go down to Fry's /Microcenter / Best Buy and buy a CPU/GPU box off the shelf kinds of numbers. but Ponte Vecchino doesn't need anywhere near those kinds of numbers. Random Joe doesn't go down the street to buy a $400M super computer every day.
Now if the cache or I/O tile is also 7nm and those too have low yield rate than it gets more painful to stack up more 'wasted' wafer by-product.
For Apple ... even the lowly Mac Pro isn't that low in volume. ( 10's of thousands probably. So not even on the picture as being a competitive alternative. )
It is fairly likely that Intel will be cranking out 7nm in early-mid 2022. It just won't be "public" product. Super Computer and largest Web Services companies will get first crack in the "thousands" range of deliveries. Not even to visibly move the Intel corporate top line revenue needle but will be out there in the field for some.
The big issue for Intel in the more mainstream GPU market is that their answer has to be in the higher volume range to be anywhere close to competitive ( if they manage to get some traction further down in the workstation "computational" GPU market. )
Thanks for the info!that's the wrong connotation. Elements of the Ponte Veccchio package were going to be 3rd party before ( certainly the HBM ).
"...As an example, our data center GPU design, Ponte Vecchio, will now be released in late 2021 or early 2022, utilizing external and internal process technologies, combined with our world-leading packaging technologies. We now expect to see initial production shipments of our first Intel-based seven-nanometer product, a client CPU, in late '22 or early '23. ...
...
...
Yeah. Yeah. On Ponte Vecchio, originally, the architecture of Ponte Vecchio includes an I/O-based die connectivity, a GPU, and some memory tiles, all kind of packaged together. That's kind of the design of Ponte Vecchio. ... "
Intel (INTC) Q2 2020 Earnings Call Transcript | The Motley Fool
INTC earnings call for the period ending June 30, 2020.
".... The only 7nm part that remains (roughly) on schedule at this point is Ponte Vecchio, Intel’s Xe-HPC GPU that will be going into the Aurora supercomputer. That is expected to ship in late 2021 or early 2022, and even then Intel is evaluating whether to move the manufacturing of some of Ponte Vecchio’s parts to third-party fabs. ..."
I think Intel has already gotten to 'tape out' stage with the GPU core chip already. It is probably pramgatically aleady too late to shift to another process. Even more so if still targeting 2021 of devillery of at least a few packages and rack units for early acceptance testing.
It may not be to late to shift either the cache memory ( lots more replication of same design. ) and or the package I/O chip ( again less diverse logic and easier validation ) to perhaps a 3rd party. Each Ponte Vecchio package has substantially more than just as die with GPU cores on it in the package.
if Intel was going to get a high price for the Ponte Vecchio it could be useful as a 7nm "pipe cleaner" for Intel. It isn't a high volume chip package. (i.e, need thousands of the packages not millions There are multiple orders of magnitude between those. ). Anandtech estimate that there Aurora has
"...
...
On the GPU side, all six of the GPUs per node will be Intel’s new 7nm Ponte Vecchio Xe GPU.
...
...
..."
- We get a rounded value of 2400 total Aurora nodes (2394 based on assumptions).
Even if Intel wasn't getting 80-95+ % years they only need 14K working GPUs to deliver. If can crank out 30-40K on a medium yield process that would probably work out OK for them.
even if only getting 25 dies per wafer per day and four machines in parallel ( for 100 good dies per day) it would oonly take about 100 days to get to the 2400. That is doable even with "bad' yields.
Intel can't do that for commodity products where go down to Fry's /Microcenter / Best Buy and buy a CPU/GPU box off the shelf kinds of numbers. but Ponte Vecchino doesn't need anywhere near those kinds of numbers. Random Joe doesn't go down the street to buy a $400M super computer every day.
Now if the cache or I/O tile is also 7nm and those too have low yield rate than it gets more painful to stack up more 'wasted' wafer by-product.
For Apple ... even the lowly Mac Pro isn't that low in volume. ( 10's of thousands probably. So not even on the picture as being a competitive alternative. )
It is fairly likely that Intel will be cranking out 7nm in early-mid 2022. It just won't be "public" product. Super Computer and largest Web Services companies will get first crack in the "thousands" range of deliveries. Not even to visibly move the Intel corporate top line revenue needle but will be out there in the field for some.
The big issue for Intel in the more mainstream GPU market is that their answer has to be in the higher volume range to be anywhere close to competitive ( if they manage to get some traction further down in the workstation "computational" GPU market. )
As has already been mentioned, you can go lower than nm....into the picometers starting at 1,000pm and the cycle repeats all the way down before transitioning into femtometers.
For the consumer, iPad pro is bar none the best value for hardware of any machine currently available. When apple collapses the software barrier between their two hardware platforms, and increases its creator and gaming audiences, it's hard to imagine wintel has a chance in the consumer market.
In my view intel is behind AMD by 3 years, but that's within Intel's paradigm. Apple is moving on to a very exciting trajectory.
In my view intel is behind AMD by 3 years, but that's within Intel's paradigm. Apple is moving on to a very exciting trajectory.
The person who created this video is a Tech Analyst, specifically related to CPU, GPU, silicon related, and a has many contacts within all the major silicon design companies. This is a scathing analysis on Intel. No one is ever 100% but he does have a great track record. If he's right, things are worse for Intel than one is led to believe
Femtometres don't come into it as a Silicon atom is 210pm (0.21nm) so while yes there's still a way to go even beyond 1nm the end of the road is in sight for traditional process shrinks - probably at least a decade off yet, but it's there. 14Å (1.4nm) is currently the smallest node being actively worked on, and 5nm, 3nm and 2nm are pencilled in before then.
Just 10years ago, nobody thought 5nm was possible either. You are looking at the future with today’s eyes and today’s technology. Nobody said we needed to stay on silicon to achieve picometer or femtometer scales...and are these even needed anyway?
I’d expect a transition to Gallium Nitride long before we get to picometers. GaN is much more efficient in a number of areas so wouldn’t need to be on such small processes to see large gains in performance and power efficiency. GaN is still in it’s infancy at the moment but research may accelerate given the technical hurdles of increasingly smaller processes.
Last edited:
I can appreciate the technical details, and that some "pro" users have unique needs and software that intel is optimized for (looking at you adobe) but the issue will be the consumer experience.
My kids play fortnite with controllers on multiple ipad pro's at 120fps. Beautiful. They get slightly warm over a couple hours.
At the same time, I have an intel i7 u series that's twice the price and never gets out the gate (yes, it doesn't have a top discrete vid card), and my G14, 4900HS with decent discrete GPU, keeps up, but is hot, while huffing and puffing.
It's great what amd is doing within the more traditional windows environment, but there is a huge gap between what an ipad pro brings in terms of live-performance relative to competitors at much higher price points and I think that will be realized in the next couple of years as apple optimizes and scales up their silicon (i'm also looking at you ms surface pro).
My kids play fortnite with controllers on multiple ipad pro's at 120fps. Beautiful. They get slightly warm over a couple hours.
At the same time, I have an intel i7 u series that's twice the price and never gets out the gate (yes, it doesn't have a top discrete vid card), and my G14, 4900HS with decent discrete GPU, keeps up, but is hot, while huffing and puffing.
It's great what amd is doing within the more traditional windows environment, but there is a huge gap between what an ipad pro brings in terms of live-performance relative to competitors at much higher price points and I think that will be realized in the next couple of years as apple optimizes and scales up their silicon (i'm also looking at you ms surface pro).
Sure; that's why I said the end for traditional (as in what we know now) shrinks which is also what you were addressing - but in fact it's unlikely shrinking the process size will continue beyond Silicon using mythical 'other materials', other completely new methods will be devised - there's currently conceptual light based transistors, and 'memristors' to name but a couple - neither involve creating sub-210pm structures.
Apple has long lost their right to complain about Intel's raw performance. The hardware has always been there, Apple just decided not to use it.
I don’t think Apple ever publicly complained about Intel’s performance. And, Apple used their hardware, they had no choice to as their hardware ran on it. It‘s just that what Intel provided didn’t perform anything like what Intel had promised.
Nothing is forever. No company dominates the market forever. Intel's just another example of this rule, nobody stays at the top forever.
People say - rightly - that Apple moved to AS because they now can control the technology and are not dependent on someone like Intel to design their chips. But Apple doesn't have a fab. All they have is design. They have to rely on companies in Taiwan and South Korea. What happens when TSMC falters? They're stuck, and unless Samsung or some other FAB comes to the rescue. Wouldn't Apple like to avoid such a fate? The only way would be to build their own fabs a la Intel, but I think that's beyond their capability. Say what you want, Intel has decades of experience in manufacturing, and Apple has zero when it comes to chips (or much of anything, really as far as manufacturing).
Apple is a fabless designer. That can make them vulnerable, and they have no way around it, as they can't pull the same trick as they did with chip design, i.e. no way can they create their own fabs. Enjoy the ride, as long as TSMC is riding high... which as we know - is destined to not last... after all, no company dominates the market forever... they have a small window of localized dominance at the moment, but that window will close. Presumably they can keep ahead of the curve for at least long enough for Apple to transition completely to AS (i.e. two years), but beyond that...
Nothing is guaranteed. Who knows, maybe there'll be another transition in store for Apple, away from the sinking ship of AS and onto another raft, some young forward looking tech company with brilliant chips
People say - rightly - that Apple moved to AS because they now can control the technology and are not dependent on someone like Intel to design their chips. But Apple doesn't have a fab. All they have is design. They have to rely on companies in Taiwan and South Korea. What happens when TSMC falters? They're stuck, and unless Samsung or some other FAB comes to the rescue. Wouldn't Apple like to avoid such a fate? The only way would be to build their own fabs a la Intel, but I think that's beyond their capability. Say what you want, Intel has decades of experience in manufacturing, and Apple has zero when it comes to chips (or much of anything, really as far as manufacturing).
Apple is a fabless designer. That can make them vulnerable, and they have no way around it, as they can't pull the same trick as they did with chip design, i.e. no way can they create their own fabs. Enjoy the ride, as long as TSMC is riding high... which as we know - is destined to not last... after all, no company dominates the market forever... they have a small window of localized dominance at the moment, but that window will close. Presumably they can keep ahead of the curve for at least long enough for Apple to transition completely to AS (i.e. two years), but beyond that...
Nothing is guaranteed. Who knows, maybe there'll be another transition in store for Apple, away from the sinking ship of AS and onto another raft, some young forward looking tech company with brilliant chips