I bet they’re still bitter they canceled their ARM line up just as iOS and Android got big.
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
I bet they’re still bitter they canceled their ARM line up just as iOS and Android got big.
The fact is, that Apple's AS is going to be a low volume proposition - you can't lump in the iPad and the iPhone into that, those are the same family of chips but not the same chips. AS chips that are going to go into Apple laptops and desktops are a vanishingly small number compared to the kind of volume sold into the PC market by AMD and Intel.
Given that low volume, how can Apple possibly compete with Intel and AMD on cost? I don't know what the price breaks are on volume from the likes of TSMC and Samsung, but I suspect Apple is not eligible, and is paying top dollar.
In other words, those who think that somehow Apple putting their own chips into their laptops and desktops means they have lower costs are going to be severely disappointed. It costs a ton to do R&D, and then they'll pay top $ for the fabbing. If anything, I wouldn't be surprised if Apple didn't raise prices to cover their costs and still maitain their traditional profit margin.
You'll get your AS Macs, but you'll pay for it, and dearly.
Given that low volume, how can Apple possibly compete with Intel and AMD on cost? I don't know what the price breaks are on volume from the likes of TSMC and Samsung, but I suspect Apple is not eligible, and is paying top dollar.
In other words, those who think that somehow Apple putting their own chips into their laptops and desktops means they have lower costs are going to be severely disappointed. It costs a ton to do R&D, and then they'll pay top $ for the fabbing. If anything, I wouldn't be surprised if Apple didn't raise prices to cover their costs and still maitain their traditional profit margin.
You'll get your AS Macs, but you'll pay for it, and dearly.
Nice article about the situation at Intel.
www.extremetech.com
Mentions Qualcomm's ARM processor for Windows. (Lenovo is shipping it already.)
So looks like Apple is not alone in moving to ARM. Just that they chose to go it alone rather than depend on a partner like Qualcomm who they have had trouble with in the past.
We've Never Seen Intel Struggle Like This - ExtremeTech
Intel is breaking records -- only this time, not in such a good way.
www.extremetech.com
So looks like Apple is not alone in moving to ARM. Just that they chose to go it alone rather than depend on a partner like Qualcomm who they have had trouble with in the past.
I'm familiar with TSMC's wafer and mask costs along with chip development NRE and ROI trade-offs. Your high level points are valid but there are *numerous* real world technical and business details that are more relevant to determining Apple's costs than this. For example, for the TSMC 5nm CPU I'm designing now our volume and yield is completely decoupled from our cost. We could have a 1% yield and it doesn't matter. The bottom line is that from a cost and ROI perspective this will be an absolute slam dunk win for Apple.
This is where, for 95% of the market, you're wrong.
Most people have enough raw performance from a 10 year old desktop machine (given it has sufficient RAM and sufficiently fast storage). What matters today for the bulk of consumers is performance per watt.
The desktop market is dying/dead for most people. Sure, there's the workstation niche, but for most people they can get all the power they need from a 10 year old PC. What matters to these people is getting that level of performance (or better) in as little power as possible.
We'll see in 12-18 months anyway. Intel has been trying to build mobile processors for a decade or so and nobody uses them because they're crap. I do not expect this to change, and I expect Apple's leadership in this space to accelerate.
Joke's on you buddy. The iPhone has more processing power than most PCs (in terms of number shipped) sold from only a few years ago. Add input devices and a display and it is plenty for "most people". This isn't a thing today, but the future is a device that sort of size to drive AR. The desktop metaphor, for most workers will simply not be relevant inside 5-10 years.
Yup. Comes from being run by people with a vision outside of watching wiggly stock market lines and thinking "how do I make this cost less", or "how do I raise the stock price".
Last edited:
The bottom line is that from a cost and ROI perspective this will be an absolute slam dunk win for Apple.
I disagree. The only way Apple gains here is that they don't have to be on the same clock cycle race as general chip designers (Intel, AMD etc.), because Apple will design for HW and SW integration. When you can design in extremely specific routines that will make your software sing on these chips, you don't care about keeping up with Intel wrt. speed, as your software execution will still be faster even with lower speed chips. This is similar to the pointless arguments people have about how much RAM an iPhone has vs some Andrioid HW. It isn't apples to apples comparison - if you use the memory more efficiently, you can still gain functionality over more raw RAM numbers - an iPhone 1 GB RAM is not the same as an Android 1 GB RAM. Btw. I still remember how the same argument was made wrt. the PPC Apple hardware vs x86 - even as Apple fell behind by any measure.
What I expect is that from now on, Apple will be on their own trajectory, where spec comparisons between PCs and AS based computers again becomes apples and oranges. The only thing that will count is if you can do stuff faster and better and run cooler and longer - what the exact memory or clock speed numbers are will not matter. It'll make for some headaches for the benchmark designers, and make direct comparisons very difficult. This represents a danger to the consumer, because Apple may simply rest on their laurels and claim they're not falling behind, they're just "different" as they did toward the end of their PPC era.
But claiming that Apple will get the same prices from TSMC as high volume clients is unrealistic dreaming - it's not how things work. Apple is a very demanding, low volume client - the kind you only service for top dollar. And the R&D for AS will not be offset by huge volume either - it will have to be recouped by low volume Apple hardware, because nobody else is buying. Anyway, we'll see - I say we'll see no price cuts on Apple HW.
I disagree. The only way Apple gains here is that they don't have to be on the same clock cycle race as general chip designers (Intel, AMD etc.), because Apple will design for HW and SW integration. When you can design in extremely specific routines that will make your software sing on these chips, you don't care about keeping up with Intel wrt. speed, as your software execution will still be faster even with lower speed chips. This is similar to the pointless arguments people have about how much RAM an iPhone has vs some Andrioid HW. It isn't apples to apples comparison - if you use the memory more efficiently, you can still gain functionality over more raw RAM numbers - an iPhone 1 GB RAM is not the same as an Android 1 GB RAM. Btw. I still remember how the same argument was made wrt. the PPC Apple hardware vs x86 - even as Apple fell behind by any measure.
What I expect is that from now on, Apple will be on their own trajectory, where spec comparisons between PCs and AS based computers again becomes apples and oranges. The only thing that will count is if you can do stuff faster and better and run cooler and longer - what the exact memory or clock speed numbers are will not matter. It'll make for some headaches for the benchmark designers, and make direct comparisons very difficult. This represents a danger to the consumer, because Apple may simply rest on their laurels and claim they're not falling behind, they're just "different" as they did toward the end of their PPC era.
But claiming that Apple will get the same prices from TSMC as high volume clients is unrealistic dreaming - it's not how things work. Apple is a very demanding, low volume client - the kind you only service for top dollar. And the R&D for AS will not be offset by huge volume either - it will have to be recouped by low volume Apple hardware, because nobody else is buying. Anyway, we'll see - I say we'll see no price cuts on Apple HW.
You what?
Iphone, iPad, watch, Mac, iPod, AirPods, pencil, HomePod, appleTV...
Low volume?
Apple, across all their products ship a heap of parts. These are parts TSMC can manufacture cheaply because the majority of them are small dies, thus the defect rate will be great.
This is why AMD have gone to chiplets. To get the sizes down to a manageable level.
Even if apple are paying 2x the "low rate" for their own parts, it will be less than what intel were charging them for mediocre refreshes of Skylake.
Also, I'm not sure Apple are even interested in discount rate. They want the most advanced process they can buy from TSMC. They were first on 10nm process, one of the first (not sure if first) on 7nm, and will no doubt be first on 5nm and 3nm. They will be partnering with TSMC to help iron out the manufacturing process bugs with their smaller mobile parts that are easier to produce.
They're certainly ahead of the queue vs. AMD - who's parts are generally larger and riskier to try and produce on cutting edge process due to sheer physical size (die sizes like those in Vega 20 for example, and big navi when it arrives).
Having actually worked at most of these "high volume clients" you're citing, interfacing with TSMC (including other fabs, ASIC vendors, test and packaging houses, IP shops, contract manufacturers, etc), involved with both the business and technical developments, for over two decades I believe I have some idea about how things work
Apple's relationship with TSMC is a lot more complex than other clients. My apologies if it sounds contentious as I don't wish to be inflammatory. Unfortunately the info that you need is company confidential or protected under NDA's. It would never be posted on a public forum by anyone who values their career. People who do work in this field will understand my points. I'll leave it at that.
Well, I don't know what to tell you - when info is NDA'd and not open to discussion, people can make all sorts of claims. I'm not saying you're not telling the truth as you believe it to be (I don't know you), but things are complicated and people assume all kinds of things that are not true... like the other poster who took exception to my claim that Apple chips in their computers are low volume - his comeback was that Apple ships tons of phones, iPads and watches.... facepalm! You are not running the same lines for chips that go into watches as you'll do when it's going into f.ex. an iMac. The costs are amortized across the entire process, not just material costs - the point being that set up costs must be recouped for each individual line, and when the volume for f.ex. iMacs is low, then you are recouping the setup costs from a smaller volume! It doesn't help you that a separate setup for watch chips is higher volume. You can't lump it all together.
Anyhow, I'm not here to argue pointlessly when an argument can't be settled due to propietary knowledge - you know yours, I know mine, 'nuff said. I still say - it will cost Apple A LOT more for their chips going into desktop and laptop lines than it will cost AMD for their high volume chips. But you guys can believe whatever you want. We'll see how it shakes out in the final price - there WILL BE NO LOWER PRICE for Apple desktop and laptop HW. Won't take long to see who is right.
Cheers, and all the best!
Anyhow, I'm not here to argue pointlessly when an argument can't be settled due to propietary knowledge - you know yours, I know mine, 'nuff said. I still say - it will cost Apple A LOT more for their chips going into desktop and laptop lines than it will cost AMD for their high volume chips. But you guys can believe whatever you want. We'll see how it shakes out in the final price - there WILL BE NO LOWER PRICE for Apple desktop and laptop HW. Won't take long to see who is right.
Cheers, and all the best!
I'm skeptical of this article, but here you go:
www.hexus.net
Insiders talking to the Taiwanese newspaper have indicated the following:
Meanwhile, AMD will press ahead to try and take advantage of its chief opponent's misfortunes / missteps. AMD hopes to capture market share in desktops, laptops, and servers and its Zen 3 CPUs and RDNA 2 GPUs will be instrumental in this aim going into the new year. AMD's next gen parts will be made on TSMC's N7/N7+ processes, and it is expected to be TSMC's biggest 7nm customer in 2021. This is all the more impressive because TSMC's 7nm capacity for 2021 will be double what it was this year.
AMD and Intel do battle over TSMC capacity, says report
Meanwhile, Intel teased an Xe GPU event for 14th August, but seems to have changed its mind…
www.hexus.net
Insiders talking to the Taiwanese newspaper have indicated the following:
- Intel has reached an agreement with TSMC
- TSMC will begin mass production of Intel CPUs and/or GPUs next year
- Intel chips will be fabricated on TSMC's 7nm optimised version of its 6nm process. (I'm not sure if that means TSMC N7P, N7+, or N6.)
Meanwhile, AMD will press ahead to try and take advantage of its chief opponent's misfortunes / missteps. AMD hopes to capture market share in desktops, laptops, and servers and its Zen 3 CPUs and RDNA 2 GPUs will be instrumental in this aim going into the new year. AMD's next gen parts will be made on TSMC's N7/N7+ processes, and it is expected to be TSMC's biggest 7nm customer in 2021. This is all the more impressive because TSMC's 7nm capacity for 2021 will be double what it was this year.
What do they have to be bitter about? Everyone else should be bitter they missed the chance to use Atom!
(Intel still makes ARM processors in their Altera line of SoCs.)
Last edited:
As an interesting anecdote I worked at Altera on the design of an FPGA with integrated Arm core. This was a monolithic die as interposer wasn't available back then. Internally the project was codenamed Excalibur. After we completed the project we couldn't sell it to anyone. At the current process node the core simply removed too many logic elements which defeated the purpose of programmable logic. Things are different today with more options for integration.
So I'm in denial because we are talking about computers here? LoLYou're in denial.
Mobile is here and dwarfs the PC market. The PC is niche. And honkin' tower buyers uber niche.
Performance, Power and Efficiency are the way it's going whether you want it or not. Apple does not agree with you.
PC is tiny market. Dell can shift alot of low margin crap. It's small potatoes.
Azrael.
PC is niche? LoL, Apple is sub 10% of the PC market.
Performance/Watt continues to improve with X86 as well.
AMD's Renoir APUs are quite efficient and easy to cool. I suspect AMD's next generation will be even more efficient.
Intel on the other had is moving to new stacked designs and big/little CPUs, so the efficiency of their chips will also improve guaranteed.
Now that's an amusing thing to say. I guess anything can be said as long as it's in Apple's advantage.
Yeah Apple is going to design custom chips for a "dead market". LoL
No at all buddy. I see that when some users here don't like the facts(like Apple being a small player on the computer market which is relevant for the discussion regarding future Macs) they try to move goalposts.
No matter how many excuses you will try to make, Apple's slice of the PC market won't become bigger, Apple won't become more relevant on the PC market.
Xilinx has gone in big for SoC integration as well. ARM cores, GPUs, a whole host of embedded peripheral hard blocks. Best I can tell, the fixed circuits barely register on the die any more among all the fabric and memory.
I've been curious where Intel goes with Altera. They don't have a great track record with acquisitions... I'm not sure they care about the SoCs-- they seem to only keep ARM as long as contractually necessary, and I can't imagine an SoC with Atom will be much fun. I suspect they want the big chips for data center use, but it would be kind of exciting to see some fabric become part of the standard PC chipset.
Intel has a horrendous record with pretty much every acquisition. One of the co-founders of Nervana is a close friend of mine. Most of the key people bailed just after they were fully vested. I was at Altera just before they become Intel PSG. Going to Intel made sense because, like memories, FPGAs are both the drivers and beneficiaries of cutting edge process nodes. Leveraging Intel's then awesome process would have been a tremendous advantage. Nobody expected Intel to fall off the process wagon although this was self inflicted. Things went south when, as a monopolist, they decided to make redundant all their experienced engineers (ie 40+) and replace them with NCG's and RCG's. It's a classic private equity play that boosts short term profits in exchange for future innovation. Would have worked if there were no viable competitors. I was on the other side during this time, at AMD, and we just brought on Jim Keller to develop the first Ryzen.
Sorry, here's the corporate HR decoder ring
NCG = new college grad
RCG = recent college grad
Apple: we’d like you to make Mac CPUs
TSMC: well, I dunno. Not a lot of volume there.
Apple: oh look, Samsung could make our next iPhone CPU!
TSMC: well, I mean, uh, we’re just saying those aren’t the same manufacturing line
Apple: that sounds like a you problem
TSMC: well, I dunno. Not a lot of volume there.
Apple: oh look, Samsung could make our next iPhone CPU!
TSMC: well, I mean, uh, we’re just saying those aren’t the same manufacturing line
Apple: that sounds like a you problem
It's not just Apple. Microsoft are pushing Surface, Google is pushing low powered tablets and chrome books and everybody is working on AR. ALL of these things need great performance per watt so the batteries aren't prohibitive.
You may think that you "need" a high powered desktop, and maybe you do, but the huge majority of people can, do (or will do) work from web hosted cloud apps in the next decade (if they are not already doing so).
To re-iterate. I do not see this as an apple unique perspective. Apple are merely pushing in that direction more heavily than Microsoft is as they already have their own architecture and aren't so constrained with legacy support.
Microsoft are trying to get to the same place, but they're trapped by their customer requirement for x86 legacy compatibility. If they could, they'd be pushing in the same direction just as hard.
Google, Amazon and others will also be building ARM based machines that will massively outperform intel in power:watt for the foreseeable future.
And that's just the power:watt perspective. Never mind that intel's arch is so badly broken from a security perspective.
[automerge]1595922953[/automerge]
Exactly. The chips aren't the same exact line, but the company will use all of them as leverage for the overall agreement.
I'd wager they will be a lot more consolidated in future as well. Everybody is likely to go chiplet based in future to improve yields and manufacturing flexibility. AMD have done it. Intel are doing it. Apple will also do it.
Its much easier to stamp out a bunch of smaller dies and "glue" them together in various ways (to vary core count, GPU, other dedicated processors, etc.) than build single monolithic do-everything chips.
[automerge]1595923185[/automerge]
You're right. The traditional "pc market" is going away. To paraphrase... that's "where the puck has been" for the past 30 years. We're moving past that, like it or not. "Work" is going to be leaving the desk and getting closer to out in the field via AR, tablets, mobile.
Last edited:
Apple doesn't need to name drop Samsung. That's a ghetto NVDA Jen-Hsun move
Here's a hint. I have never seen TSMC create a customized process for a large customer to the level they did for AAPL. It's unheard of and I've been working with TSMC since close to when they were first founded. It would be more accurate to think of AAPL as a partner than a customer.You have a distorted and limited view of the computer market in general.
Both Microsoft and Google are very small players in the hardware computer market and I don't see how they are pushing for what you claim they are pushing. They are just promoting their hardware and software.
Microsoft's hardware isn't officially available in most markets world wide for example. Their Surface project is nice but it hardly had a real fundamental impact on the PC market.
Yeah they are trapped dominating the PC market with a huge software library backing up their OS.
And Microsoft is already promoting Windows S or whatever. The thing is most people still only want full Windows which can run any software they could imagine.
I doubt it. As Amazon's ARM Server parts were designed to be at parity in terms of features with X86 server parts they draw similar amounts of power for similar performance in certain cases and much lower performance in other cases.
Also intel's upcoming CPUs will use a different architecture so we will see how they will stack up in performance/watt and security. AMD's Zen CPUs have been excellent in this regard.
Yeah right, I've been hearing this for years. Generally from apple fans that can't stand the fact that Apple holds a small slice of the traditional PC market after decades of presence in said market.
I see you have an obsession with AR, which also has been the best thing since sliced bread from the time apple introduced it in 2017.
[automerge]1595926194[/automerge]
You say it like TSMC did something out of kindness or friendship or whatever.Apple doesn't need to name drop Samsung. That's a ghetto NVDA Jen-Hsun move
Last edited:
Size isn't everything. The decisions Apple, Microsoft and Google make in their hardware products are closely looked at by the competition.
Plus, Google is hardly a small player in the education segment.
Well I meant both in terms or size and impact. The competition is more concerned than anything to satisfy the demand, give people what they want.
Google's Chromebooks are cheap so that's why they are somehow successful in the educational segment. I remember some Apple execs made some salty comments regarding the educational segment choosing chromebooks over ipads.