M4+ Chip Generation - Speculation Megathread [MERGED]

[Tagbert]

Poor analogy. A much more apt analogy would involve a customer ordering pizza from dominos. There is no trade secret or “skill” with chip design just like there’s no skill in “designing” a pizza from dominos. Whatever pizza you tell dominos to make can be easily made by someone else if they wanted to. Likewise in chip development, the hardest part is the manufacturing process, as that determines the yield curve. Apple just has to decide how much money they want to spend and where in the yield curve they want to be. The performance profile can be easily matched by anyone if they wanted. Your large pepperoni pizza from dominos isn’t special. I could’ve done the same thing.


What Apple brings to the table is economies of scale through their large sales volume. They’re not bringing much technical expertise, if at all. It doesn’t take any skill whatsoever to determine how large you can make your core, how many you can fit in the SoC and what clock speed they can run at. These are incredibly elementary tasks that a 3rd grader can do. The primary limitation is the manufacturing process.

Are you really basing your argument on the idea that there is no knowledge and skill involved in chip design?
Wow! That is far from the mark.

 

[Tagbert]

Is there a reason ‘High IQ Person’ is being replied to seriously? They’re clearly unserious.

It’s a bit like staring at the car wreck as you drive by. It’s hard to turn away.

 

[senttoschool]

Sorry, I should have made it clear that I think A18 will also be on N3P. So I agree A18/M4. TSMC is saying "2H 2024" for N3P high-volume production, which is identical to what they said for N4P (A16) and N3 (A17/M3). To my mind, it's virtually certain that A18 will be on N3P. Apple and the iPhone will once again be the first consumer product on the new node. This has been happening for almost a decade now (since A8/A8X in 2014), it's not hard to see coming. N3E is the exception, not the rule.

It's not 100% confirmed that it is using N4P. Second half mass production is probably too late for the A series usually. It needs to be the first half of the year. For example, N3B volume production started late 2022/early 2023.

The mass production of iPhone chips usually starts in Spring.

As for Occam's Razor, I'll admit it is too early to make assumptions, but this is the speculation thread, after all! I'd suggest the simplest explanation is that Apple's roadmap always had the Mx and Mx Pro/Max coming out at the same time (i.e., within a month of one another). This pattern also goes all the way back to the early days of Apple Silicon, with the A6/A6X in 2012. A14/M1 also fits. There is only one exception (other than A15/M2), the A10/A10X, but that involved a process node change (with the A10X iPad Pro being the first consumer devices on TSMC 10nm), the only time that (switching process nodes midstream) has happened. So most likely the gap between M1 and M1 Pro/Max was on the roadmap for the transition before the pandemic, while the gap between M2 and M2 Pro/Max was not.

We can't use iPad SoCs as evidence for cadence, in my opinion.

The reason is because Apple Silicon goes into far more devices and it's far more important today than it was 10 years ago.

Count how much of Apple's business relies on Apple Silicon today than when they were making X chips. Count how many more employees Apple has today than before. Everything relies on Apple Silicon. An Apple exec also came out and said they want to release the M series every year. You can check my post history for this source. I've written about this extensively.

Regardless, more than a decade of Apple silicon history argues that the M3 and M3 Pro/Max launching at the same time is likely to be the rule going forward. So that explains your 10-month Pro/Max course correction, putting the M-series launch cadence back on track (after it went off-track for M2). Mx and Mx Pro/Max will come out at the same time as long as Apple continues to use monolithic dies, I'll guess wildly that will be until about 2030, when heterolithic integration seems (to me) inevitable, resulting in changes to the architectural (and, yes, marketing) approach. That's probably the biggest, most closely-held of trade secrets, when and how Apple plans to make that jump. It's also the biggest unknown in terms of science and innovation, there are challenges looming at "1nm" and beyond, not just for TSMC, but for everyone.

M3, M3 Pro/Max came out one month after the A17 Pro, using the same core designs. Therefore, Apple designed the M series in tandem. They've shown twice now that the A series and M series designs release within 1 month of each other (A15 & M1, A17 Pro & M3, M3 Pro/Max). It's reasonable to assume that Apple has the capabilities and intentions to release the M series on a yearly cadence and that the M2 generation couldn't come out on time due to a rare combination of events.

Therefore, it's more reasonable to assume that the M series is on a yearly cadence - same as the A series.

 

[MayaUser]

Would be funny if TSMC gets stuck and Apple silicon stagnates. What will the Apple fans say when they see no performance or efficiency gains for 5 years straight? Will they finally wake up and realize that Apple's "design" is worthless marketing gimmick and only lithography matters for chip performance?

So TSMC and Apple must be stuck 5 years for others to catch up...you contradict yourself
Think about if the others get stuck and Apple/TSMC doesnt...what difference will be
Are you even reading what you write before you press "post reply" mr high iq person?
Im guessing Elvis has left the chat

 

[MayaUser]

Therefore, it's more reasonable to assume that the M series is on a yearly cadence - same as the A series.

So based on that, we can expect some M4 to be release/announce in October/November 2024

 

[senttoschool]

So based on that, we can expect some M4 to be release/announce in October/November 2024

Yes, that's my prediction. Of course, I could be wrong. We don't know for sure. But there is strong evidence.

 

[leman]

Poor analogy. A much more apt analogy would involve a customer ordering pizza from dominos. There is no trade secret or “skill” with chip design just like there’s no skill in “designing” a pizza from dominos. Whatever pizza you tell dominos to make can be easily made by someone else if they wanted to. Likewise in chip development, the hardest part is the manufacturing process, as that determines the yield curve.

Buddy, you should consider changing your handle.

 

[tenthousandthings]

It's not 100% confirmed that it is using N4P. Second half mass production is probably too late for the A series usually. It needs to be the first half of the year. For example, N3B volume production started late 2022/early 2023.

Yes, I'm sorry I messed this up, you are right that TSMC initially used "2H 2022" (same as what they used for N4P) for N3, not "2H 2023" as I recalled. So the A16 is my only leg to stand on.

FWIW, I haven't heard of anyone disputing TechInsights' conclusion (based on physical analysis) of N4P for A16.

 

[tenthousandthings]

An Apple exec also came out and said they want to release the M series every year. You can check my post history for this source. I've written about this extensively.

I couldn't find this in the first few pages of your post history, so you'll have to point me toward it. I know of two clear instances where Apple executives have said in formal interviews that Apple wants every Mac get every generation of the M-series silicon. The first was Anand Shimpi, in an interview about the M2 Pro/Max launch with Andru Edwards in February 2023. I did a transcript of his part of that interview. Here's the relevant quote:

"... But really the thing that we see, that the iPhone and the iPad have enjoyed over the years, is this idea that every generation gets the latest of our IPs, the latest CPU IP, the latest GPU, media engine, neural engine, and so on and so forth, and so now the Mac gets to be on that cadence too. If you look at how we’ve evolved things on the phone and iPad, those IPs tend to get more efficient over time. There is this relationship, if the fundamental chassis doesn’t change, any additional performance you draw, you deliver has to be done more efficiently, and so this is the first time the MacBook Pro gets to enjoy that and be on that same sort of cycle. ..."

That is not the same as saying they want to release the M-series every year. He does not say that. The other instance I would have to track down, but it was in an Indian publication and the executive mentions all shipping Macs when he says the something similar. Everything so far points to a cycle averaging 18 months, with odd-numbered generations in October and even-numbered generations in June.

 

[tenthousandthings]

M3, M3 Pro/Max came out one month after the A17 Pro, using the same core designs. Therefore, Apple designed the M series in tandem. They've shown twice now that the A series and M series designs release within 1 month of each other (A15 & M1, A17 Pro & M3, M3 Pro/Max). It's reasonable to assume that Apple has the capabilities and intentions to release the M series on a yearly cadence and that the M2 generation couldn't come out on time due to a rare combination of events.

Therefore, it's more reasonable to assume that the M series is on a yearly cadence - same as the A series.

I get what you're saying. I think, yes, it's reasonable to assume Apple has that capability, but, no, it's not reasonable to assume Apple has that intention. The facts to date do not support the latter.

 

[deconstruct60]

I think there is a low chance of this happening. N3E will be used for A18 Pro. Therefore, it'd be easier to build the M chips on the same node as A18 Pro.

And yet there was no TSMC N4 M-series SoC produced. Just because Apple ships an A-series doesn't mean they "have to" ship a M-series.

Individual iPhones are sold for 3 whole years at a time. It gets to be the 'lead' iPhone , but they it is sold as the n-1 and then n-2 iPhone. The A-series are also "hand me down" SoCs for entry iPad , AppleTV , etc. There is way too much attention paid to when Apple starts to sell the SoCs and not on how many of them they sell ( which length of deployment to 'new' devices contributes to).

Through M2 and M3 iterations the M-series hasn't hit any exact 12 month mark at all.

Apple doesn't "have to" wait until the following An+1 comes out to release. Nor does the A-series always "have to" come out first. Major contributing factor is when the TSMC fab process is ready and "what it is". N3E is mainly just "bigger and cheaper". The Pro and Max chips are already relatively (to A-series) bigger. N3E is mainly going to get something 'even bigger' for those dies. Apple isn't charging discount low, low prices for the Pro/Max so 'cheaper' may not matter all that much.

Also, it's not apparent that there is an 18-month cadence. In fact, M3 Pro/Max came out only 10 months after M2 Pro/Max. I'm not suggesting that Apple has a 10 month cadence. Occam's Razor suggests it's a 12 month cadence that follows the iPhone cadence since they both use the same architecture and occasionally, this cadence will deviate due to supply issues, node timings, pandemics.

That isn't "Occam's Razor". The A-series was making the investment back years before the M-series showed up. The A-series can operate independently on release schedule. It did years before. It did for TSMC N4.

Your supposed "Razor" is that there is some huge encumbered coupling to the non technical "every September" timing .... And there isn't. If the TSMC N3P fab process is bad for the "every September" but good for a Mac Spring release then Apple could go with that. (moving Mac SoC 'peak volume' away from 'iPhone' SoC peak volume would spread the load out over more of the year. With TSMC 'arriving' in 2Hxx with new product that is 'bad' timing for phone but it isn't necessarily 'bad' for Mac. Mac isn't obsessive-compulsive-disease committed to every September. ) . That is inline with the "keep it simple" notion of Occam's Razor.

The M3 Pro/Max very likely had more to do with Intel cancelling/deferring huge orders of N3B than anything about the rigid iPhone release schedule. If TSMC went to Apple and said "can you use up these 10,000 extra wafers we have" ... which is the fastest way to do that. Make approx 100mm^2 A17Pro or approx 120mm^2 M3 or make over 400mm^2 Max dies. The last will suck up wafers 4x as fast as the other two.

Is Intel going to screw up their wafer orders every 12 months? Probably not.


The iPads when all of 2023 with zero updates. No, rigid 12 month cycle there either.

The M-series on exact same iteration cycle as A-series is more "monkey see, monkey do" than Occam's Razor. It makes about zero technical sense if looking at substantive fab cycle iteration times at TSMC. Nor particularly economically, Apple trying kill off 400mm^2 dies as fast as possible. ( Apple making TSMC 'eat' costs bad dies during most of 2023 , pretty good chance was a bit of a 'trade'. And Studio/Mac Pro drifting for an extended period of time won't be surprising either. )

 

[deconstruct60]

I'm not in a position to disagree, but, according to Anandtech, SRAM density will benefit from N3P's overall optical shrink (adjustments to the optical performance of the scanners), so, if I'm right about the process node, perhaps its role could be increased in the M4 GPU designs?

It will benefit, but probably still not at N3B levels. In the article, they note that N3B SRAM is a 5% shrink from N5. N3E goes back to N5 ( so 'up' 5%). N3P chip density is 1.04x of N3E. Pretty sure that "chip density" is a composite score of a 'standard arm core implementation' (so includes some SRAM and logic). If it is only clawing back 4%, then pretty likely it is still a net negative. It is lower net negative , but still not in the zone where can pour on SRAM like ketchup without sliding 'backwards'. It is just an incrementally less steep slippery slope.


Pretty decent chance the A18 / A18Pro are N3E and Apple plays some 'name games' with the same die with some cores/features turned off to segregate the "Pro" version. Like the A16. And, IMHO, just skip it for M4. N3P seems likely to be what Apple was hoping to get with N3B. N3E is just a stop-gap that doesn't do much for the M-series. There is enough life-cycle overall volume for the A-series over >3 years to make re-spin to N3E with some tweaks/fixes worth it. Where the M-series, in aggregate, doesn't really have that.

Previously there were rumors that Intel's cancelled N3B orders were going to turn into N3E orders. If Apple was going to be the only major buyer ever of N3B , it would make some sense to rush off of N3B as fast as possible ( TSMC would want to turn it 'off' sooner rather than later). However, recent leaks indicate that Intel is going to consume N3B. So more doubtful Apple is in a 'panic' to get off of N3B quickly. At least for the Mac SoCs.
The A17Pro isn't as good a match to the "hand me down" targets of entry iPad and the like so getting to a more long lived node (e.g., N3E ) that is also 'cheaper to make' would make sense. Apple needs a 'non-Pro' A17-ish SoC for that.

N3P gets some performance uplift ( which will help selling M4-on-N3P as an upgrade over M1-M2 ) and is likely a bit more affordable than N3B. Should be a longer lasting node also ( for when the plain M4 gets into the "hand me down" zone).

 

[senttoschool]

I get what you're saying. I think, yes, it's reasonable to assume Apple has that capability, but, no, it's not reasonable to assume Apple has that intention. The facts to date do not support the latter.

The fact is, in 2/3 generations, the M series followed the A series by 1 month.

The only generation where it didn't was during the middle of covid, with huge supply chain issues, massive demand, and the new Air design being late.

 

[senttoschool]

I couldn't find this in the first few pages of your post history, so you'll have to point me toward it. I know of two clear instances where Apple executives have said in formal interviews that Apple wants every Mac get every generation of the M-series silicon. The first was Anand Shimpi, in an interview about the M2 Pro/Max launch with Andru Edwards in February 2023. I did a transcript of his part of that interview. Here's the relevant quote:

"... But really the thing that we see, that the iPhone and the iPad have enjoyed over the years, is this idea that every generation gets the latest of our IPs, the latest CPU IP, the latest GPU, media engine, neural engine, and so on and so forth, and so now the Mac gets to be on that cadence too. If you look at how we’ve evolved things on the phone and iPad, those IPs tend to get more efficient over time. There is this relationship, if the fundamental chassis doesn’t change, any additional performance you draw, you deliver has to be done more efficiently, and so this is the first time the MacBook Pro gets to enjoy that and be on that same sort of cycle. ..."

That is not the same as saying they want to release the M-series every year. He does not say that. The other instance I would have to track down, but it was in an Italian publication and the executive mentions all shipping Macs when he says the same thing. Everything so far points to a cycle averaging 18 months, with odd-numbered generations in October and even-numbered generations in June.

“The M2 family was really now about maintaining that leadership position by pushing, again, to the limits of technology. We don’t leave things on the table,” says Millet. “We don’t take a 20% bump and figure out how to spread it over three years…figure out how to eke out incremental gains. We take it all in one year; we just hit it really hard. That’s not what happens in the rest of the industry or historically.”

Apple execs on M2 chips, winning gamers and when to buy a Mac | TechCrunch

Apple's M series chips were incredibly well telegraphed when they arrived in late 2020. Apple had been designing its own silicon since the A4 appeared in

techcrunch.com

We're not going to get Apple to come out and explicitly confirm they will release an M chip once a year. But all interviews indicate a 12 month update cadence - same as the A series.

 

[deconstruct60]

“The M2 family was really now about maintaining that leadership position by pushing, again, to the limits of technology. We don’t leave things on the table,” says Millet. “We don’t take a 20% bump and figure out how to spread it over three years…figure out how to eke out incremental gains. We take it all in one year; we just hit it really hard. That’s not what happens in the rest of the industry or historically.”

Apple execs on M2 chips, winning gamers and when to buy a Mac | TechCrunch

Apple's M series chips were incredibly well telegraphed when they arrived in late 2020. Apple had been designing its own silicon since the A4 appeared in

techcrunch.com

We're not going to get Apple to come out and explicitly confirm they will release an M chip once a year. But all interviews indicate a 12 month update cadence - same as the A series.

That is fundamentally flawed in three parts. First, that the A series is as broad ( number die versions) as the M series. At one point the A-series had two dies. An and AnX. AnX did NOT arrive every year (e.g. A10X -> A12X -> A14X/Z ) . There was no Apple dogma there that the AnX has to drop every single year they did a plain An die. They didn't 'dribble out' the AnX , but they also didn't pull the trigger every year either. The 'plain' M-series took over the role that the AnX had in the iPad Pro line up.

Second, it is a bit a joke in the A-series sector. Last several of years Apple has done either binned down dies for iPhone Pro or segmented the new die to only the Pro and dribbled it down to the plain iPhone they next year.

Third, did you see a 20% jump in Single Thread drag racing? That 20% is cherry picking. It is a SoC. So there is subsantive "kitchen sink" of stuff in there. Certain subsystems get a push in a narrow function area , but it isn't an 20% overall on every function push. Lots of this hand waving being done by Apple here is throwing 'shade' on packages that are not a broadly targeted so have fewer subsystems to iterate through generational change.


The Watch SoC has drifted for multiple years with no updates. Apple doesn't have a deep resume of doing 4-5 dies per year on a constant basis. When they make moves , they don't hold back. However, they don't always make moves every year. In fact, "not figuring out how to eke out incremental gains" isn't necessary is not a obsessive compulsive about an non technical, arbitrary 'every 12 months' schedule. Drop very good increments when have them. Technology innovation don't necessarily come every 12 months. Bugs fixes , clean ups sure... major jumps aren't regular. ( Moore's Law was 18 months. )

Apple is churning the leading edge iPhone die at 12 months in part because it is a single die. Last two years aimed at an even smaller product smaller product base ( just iPhone Pro). [ We'll see if get back to covering regular iPhone this year. But if do, they are likely back to just a single die again. ]

 

[WilliApple]

Lol were talking about M4 already?

Should I just wait until M4 instead of getting M3 😂

 

[hovscorpion12]

Might as well wait for the M8 series. Going to be mind-blowing

BMW knows 😎

 

[Chuckeee]

Apple execs on M2 chips, winning gamers and when to buy a Mac | TechCrunch

Apple's M series chips were incredibly well telegraphed when they arrived in late 2020. Apple had been designing its own silicon since the A4 appeared in

techcrunch.com

We're not going to get Apple to come out and explicitly confirm they will release an M chip once a year. But all interviews indicate a 12 month update cadence - same as the A series.

I think Apple is strongly driven to deliver annual iPhone updates by purely marketing force. This is not necessarily the same as annual processors updates, although historically that has been the general trend. For economic purposes, Apple's required to have new iPhone models available for each holiday season.

This is certainly not the case for Apple’s other products, iPads, MacBooks, desktops, watches, AppleTV, etc.. All of these devices updates but there is not a strong market force driving regular annual updates.

We need to remember Apple is now primarily a phone company with a strong ambition to be a service company. Production of Mx chips is not their primary business but is just part of the fallout of making phones. So does annual Mx updates aid phone sales? That Is the question. I suspect annual computer updates is no longer necessary to support iPhone sale. Although occasional updates to show apple technology dominance is still necessary.

It now seems AI is the latest & greatest marketing shiny object 🪩. So Apple will want to release products with AI features so they don’t appear to be [even more] technology deficient. I suspect this can’t be done purely with a software update so M4 processors with AI capabilities will probably occur sooner rather than later, To be included in new Mac computers.

 

[senttoschool]

I think Apple is strongly driven to deliver annual iPhone updates by purely marketing force. This is not necessarily the same as annual processors updates, although historically that has been the general trend. For economic purposes, Apple's required to have new iPhone models available for each holiday season.

This is certainly not the case for Apple’s other products, iPads, MacBooks, desktops, watches, AppleTV, etc.. All of these devices updates but there is not a strong market force driving regular annual updates.

We need to remember Apple is now primarily a phone company with a strong ambition to be a service company. Production of Mx chips is not their primary business but is just part of the fallout of making phones. So does annual Mx updates aid phone sales? That Is the question. I suspect annual computer updates is no longer necessary to support iPhone sale. Although occasional updates to show apple technology dominance is still necessary.

It now seems AI is the latest & greatest marketing shiny object 🪩. So Apple will want to release products with AI features so they don’t appear to be [even more] technology deficient. I suspect this can’t be done purely with a software update so M4 processors with AI capabilities will probably occur sooner rather than later, To be included in new Mac computers.

I didn't say it was necessary.

All I'm saying is that it seems likely based on evidence so far. In every interview, Apple execs have hinted at yearly updates for Apple Silicon, same as iPhones. In 2/3 generations, that was true. The only generation it wasn't true was M2 and that was most likely due to covid, work from home, huge supply chain issues, huge demand, and new Macbook Air design being late.

 

[MayaUser]

Lol were talking about M4 already?

Should I just wait until M4 instead of getting M3 😂

Why not? We are 6 months away from getting a glimpse of what m4 will be capable of

 

[TAJones99]

The M4 will make all your wildest dreams come true!

Weird Science style lol

 

[MayaUser]

High IQ persons replies ages...not so well

Would be funny if TSMC gets stuck and Apple silicon stagnates. What will the Apple fans say when they see no performance or efficiency gains for 5 years straight? Will they finally wake up and realize that Apple's "design" is worthless marketing gimmick and only lithography matters for chip performance?

this, as i said and expected didnt ages well at all

• iPhone 16 Pro (2024): "A18" (3nm, N3E)
• "iPhone 17 Pro" (2025): "A19" (2nm, N2)
• "iPhone 18 Pro" (2026): "A20" (2nm, N2P)
• "iPhone 19 Pro" (2027): "A21" (1.4nm, A14)

Good luck in the future, and lets see the others how "stuck" will be behind TSMC/Apple .Will they finally wake up..

 

[High IQ Person]

High IQ persons replies ages...not so well

this, as i said and expected didnt ages well at all

• iPhone 16 Pro (2024): "A18" (3nm, N3E)
• "iPhone 17 Pro" (2025): "A19" (2nm, N2)
• "iPhone 18 Pro" (2026): "A20" (2nm, N2P)
• "iPhone 19 Pro" (2027): "A21" (1.4nm, A14)

Good luck in the future, and lets see the others how "stuck" will be behind TSMC/Apple .Will they finally wake up..

I foresee years of delays, yield issues, and poor efficiency.

 

[leman]

Would be funny if TSMC gets stuck and Apple silicon stagnates. What will the Apple fans say when they see no performance or efficiency gains for 5 years straight?

Apple had three generations of chips using pretty much the same lithography and they managed to deliver performance and efficiency gains nevertheless. Not sure what you are trying to say. Sure, Apple relies on TSMC’s roadmap to implement their chips. If all foundries get stuck I’m sure the industry will figure something out.

 

[High IQ Person]

Apple had three generations of chips using pretty much the same lithography and they managed to deliver performance and efficiency gains nevertheless. Not sure what you are trying to say. Sure, Apple relies on TSMC’s roadmap to implement their chips. If all foundries get stuck I’m sure the industry will figure something out.

N5 to N4 to N4P were not the same processes.