Apple A14 - 5 nm, 11.8 billion transistors

[Eug]

A14 - First announced for iPad Air.

 

[IntelUser2000]

The reason Apple doesn't cram huge batteries into their phones is that they determined from data and research that users would rather have other features than a bigger battery.

iPhones don't blow up in the users' hands either.

 

[teejee]

I did not say that. Apple as a trendsetter certainly has a huge impact on what people want though. And people seem to change their mind and claim that what is provided is what they want, even though until recently they claimed otherwise.
Like, rounded backs, until few weeks ago, they were great, because it makes the phone seem slimmer and more comfortable to grab, more natural feel, blah blah. Now, blocky design is more comfortable.

My issue with apple is that they deliberately postpone updates simply because they can. And this is alright, it's the position they have earned with their success. Still sucks though.




I'm not saying they SHOULD. I WISH they did. I wish everybody did.

That being said, I am surprised that with all this vertical integration they are so proud of, they are unable to cram bigger batteries in the same body.
How is it possible that Samsung is able to engineer a phone that is smaller, with a bigger screen, lighter AND with bigger battery? Yes, battery life is even worse there, but this is too be expected, because of inherent inefficiencies of android/hardware.
The reason surely is that they are holding back because.. They can. They keep this ace in the sleeve.

It's the opposite with SOCs. They had a great run these past few years.

Try this with a Samsung:

In short, an iphone 12 is not possible to bend with normal hand strength. I prefer a solid phone instead of en extra hour of screen time.

A smartphone is a compromise, if you add more battery you need to remove something else taking space and weight.

 

[MarkizSchnitzel]

Try this with a Samsung:

In short, an iphone 12 is not possible to bend with normal hand strength. I prefer a solid phone instead of en extra hour of screen time.

A smartphone is a compromise, if you add more battery you need to remove something else taking space and weight.

What good is that for? I have never ever heard of anyone in my vicinity, having bent their phone. There were a few online stories only.

But that is not even mutually exclusive. A negligibly thicker device would be able to safely have a bigger battery.

 

[MarkizSchnitzel]

Yes, Apple is "holding back" their flagship product because they want to keep an "ace" in their sleeve.

That makes a ton of sense. NOT. What's with these silly conspiracy theories?

The reason Apple doesn't cram huge batteries into their phones is that they determined from data and research that users would rather have other features than a bigger battery.

Why would it be mutually exclusive?
OF COURSE they are holding back. EVERYBODY is. Because fast progress has stagnated for years now. They need to justify the NEXT model, next year. Only update what research tells you is the absolute minimum needed to sell sufficiently. Hence, no high refresh rate. They claim it was wireless charging OR high refresh rate. What do two have in common to be mutually exclusive?
They have too small a battery and can't afford HRR screen is the real reason. Of which cause is that most people don't really know what HRR is.

 

[teejee]

Why would it be mutually exclusive?
OF COURSE they are holding back. EVERYBODY is. Because fast progress has stagnated for years now. They need to justify the NEXT model, next year. Only update what research tells you is the absolute minimum needed to sell sufficiently. Hence, no high refresh rate. They claim it was wireless charging OR high refresh rate. What do two have in common to be mutually exclusive?
They have too small a battery and can't afford HRR screen is the real reason. Of which cause is that most people don't really know what HRR is.

Apple will likely do HRR by implementing true variable refresh rate to save power. This is not possible with normal OLED displays. They need LPTO displays for this. This is only available in Samsung Note 20 ultra that was launched two months ago. So it is not surprising that Samsung cannot (or don’t want to) deliver those displays in ”Apple volumes” already for iphone 12.

Samsung's Note20 Ultra Variable Refresh Rate Display Explained

www.anandtech.com

 

[jpiniero]

Apple A14 SoC Put Under the Microscope; Die Size, and Transistor Density Calculated

Apple has established itself as a master of silicon integrated circuit design and has proven over the years that its processors deliver the best results, generation after generation. If we take a look at the performance numbers of the latest A14 Bionic, you can conclude that its performance is...

www.techpowerup.com

Says chip is 88 mm2 for 134M transistors/mm2.

 

[Doug S]

I did not say that. Apple as a trendsetter certainly has a huge impact on what people want though. And people seem to change their mind and claim that what is provided is what they want, even though until recently they claimed otherwise.
Like, rounded backs, until few weeks ago, they were great, because it makes the phone seem slimmer and more comfortable to grab, more natural feel, blah blah. Now, blocky design is more comfortable.

My issue with apple is that they deliberately postpone updates simply because they can. And this is alright, it's the position they have earned with their success. Still sucks though.




I'm not saying they SHOULD. I WISH they did. I wish everybody did.

That being said, I am surprised that with all this vertical integration they are so proud of, they are unable to cram bigger batteries in the same body.
How is it possible that Samsung is able to engineer a phone that is smaller, with a bigger screen, lighter AND with bigger battery? Yes, battery life is even worse there, but this is too be expected, because of inherent inefficiencies of android/hardware.
The reason surely is that they are holding back because.. They can. They keep this ace in the sleeve.

It's the opposite with SOCs. They had a great run these past few years.

They aren't comparable in all ways, they use different materials for the frame, have different hardware (e.g. magnets) and so forth.

Your argument seems to be that because Samsung crammed a bigger battery in their phone that Apple should use one of similar size, even though you are admit that Samsung's battery life is worse. Is your goal battery life or battery size?

I don't understand the obsession some people have with batteries. Are you using your phone so heavily that it drains its battery in less than a day? And your use is so constant you don't ever have an opportunity to top it up at some point during the day?

Everyone has some little niche thing they are obsessed with, that most people don't care about. Mine is large displays on a laptop. I insist on a 17" display, but I want it light so the DTRs, gaming laptops or workstations are not acceptable to me. As a result I have a very small number of options. But I'm not complaining "hey LG found a way to make a 17" laptop that's only three pounds, so why aren't all 17" laptops only three pounds so I can get what I want from the OEM I want?" I realize, as should everyone saying Apple should have bigger batteries despite already ranking near the top in battery life, that what I want is not what the bulk of the market wants.

 

[Mopetar]

That's approaching levels of density only seen in neutron stars. It's definitely more dense than my cousin who thought he needed to study for his blood test at any rate.

 

[MarkizSchnitzel]

@Doug S
I am sorry to have hijacked the topic with irrelevant discussion.
I will say this still.
My points are this:
1) apple is capable of designing an iphone to be the same dimensions and structural stability but with a bigger battery. Or at the very worst an almost imperceivable increase in thickness. They have cream of the crop engineers at every step of the process.
The only reason they don't do it is excessive bean counting. Or, looking at their stock prices, maybe not excessive, maybe just the right amount.
2) any phone made thicker by 1 mm is not suddenly too thick. Otherwise, people would not be putting ridiculous cases, and phones would be designed of high quality plastic.
3) it is senseless, to me, to EVER say that battery is "enough". It is NEVER enough. Not until it is perpetual. Anything else can be only sufficient. Who here would not like e.g. a phone that lasts for say 1 week?
4) as with anything in life, high end is never cost effective. For phones, with very short lifespans, more then most other areas. A tradesmen can buy all the tools he will ever need for the next 15 years for the price of 1 high end iphone/galaxy.
5) most of the high end bleeding edge is completely wasted on most people. You don't need A14 for messaging and similar low level stuff.

 

[Doug S]

3) it is senseless, to me, to EVER say that battery is "enough". It is NEVER enough. Not until it is perpetual. Anything else can be only sufficient. Who here would not like e.g. a phone that lasts for say 1 week?

What's the value add for a battery that lasts a week? Going on a trip and not having to bring a charger?

I think any battery life above a day is useless, and I'd rather have a lighter phone. I charge my iPhone 11 pro max every other day, so FOR MY USE if the battery was less and it was lighter I think that would be good. If I only charged it every third or fourth day that would add ZERO value as far as I'm concerned.

The idea that no battery is ever enough unless it is perpetual is ludicrous. Charging is hardly a terrible chore, it isn't like you have to hop on a bike and generate the electricity yourself.

 

[Mopetar]

@Doug S
I am sorry to have hijacked the topic with irrelevant discussion.
I will say this still.
My points are this:
1) apple is capable of designing an iphone to be the same dimensions and structural stability but with a bigger battery. Or at the very worst an almost imperceivable increase in thickness. They have cream of the crop engineers at every step of the process.
The only reason they don't do it is excessive bean counting. Or, looking at their stock prices, maybe not excessive, maybe just the right amount.
2) any phone made thicker by 1 mm is not suddenly too thick. Otherwise, people would not be putting ridiculous cases, and phones would be designed of high quality plastic.

Your argument in picture form:

There's no iPhone that's going to satisfy everyone and if everyone got to add on their "one little thing" to make it perfect, it would wind up being a mess. You think your request sounds reasonable because it's "Just an extra 1mm" but there's someone else who wants another 1mm for more batter, and a person who wants a bigger battery still and it's just another 1mm, but now it's too thick for you and you have a complaint. Pretty soon someone just needs a SCSI port on their iPhone. It can probably fit in between the Parallel and HDMI micro ports since we made the device even taller to accomodate the people who needed a tablet sized screen.

The cream of the crop engineers know what listening to every little request gets you and why its better to just ignore them all and make a great product that can be adapted to suit the needs of people who fall outside of the 99% of users for some particular purpose.

 

[Jan Olšan]

"nobody needs more battery life"
"battery life is a Homer car"

Inb4 people gushing out in ecstasy at the revolutionary quality of life breakthrough if Apple actually does decide to make the battery thicker for more capacity. The extra weight absolutely worth it!

Wake up, it's quite silly to believe that every single product design and spec decision is carefully tuned to make the best device ever for the maximum number of users and not at all, never ever by factors such as "this saves $$ on every device we sell".

 

[Roland00Address]

The cream of the crop engineers know what listening to every little request gets you and why its better to just ignore them all and make a great product that can be adapted to suit the needs of people who fall outside of the 99% of users for some particular purpose.

This is true some of the time but not true all of the time.

Remember 3.5" phones? Apple stuck with the smaller phone sizes stubbornly even though all the data at the time showed people wanted larger phones even though people were not sure what was the final resting place for how big they wanted the phones to be.

• 3.5” diagonal (03 by 02) 5.65 inches² the original size iPhone original to iphone 4s
• 4.0” diagonal (16 by 09) 6.84 inches² aka the taller iPhone 5
• 4.7” diagonal (16 by 09) 9.44 inches² aka the taller and larger iPhone 6
• 5.5” diagonal (16 by 09) 12.93 inches² aka the taller and larger iPhone 6 plus

And since then we got a far more variety of sizes. But analysts show that android was getting US market-share until the iphone 6 generation. Likely due mainly to screen size and not other specs or features. Same thing with the “higher end” world markets that are not the United States. Though Apple’s market-share in global lower and midrange markets is a more complicated story for Apple phones are generally more expensive with the cheapest MSRP being $400 (US).

But remember the Galaxy S3 (2012) it was a 4.8" 1080p display competing against the iphone 4s (released 5 months earlier) with its 3.5" screen. 5.65 inches² compared to 9.84 inches². And when the iphone 5 was released 6 months after the S3 it was a 4.0" 16 by 09 screen compared to a 4.8" (6.84 inches² compared to 9.84 inches²) Those 3.0 inches squared were a big thing, apple fans were so happy to move from 5.65 to 6.84 inches² but many customers still wanted more.

(moving to more current phones)

• The iPhone X moved to rounded corners and notch so it is not directly comparable. It is taller (13x6, aka 19.5 by 09) 5.8” diagonal. (Not going to do area comparisons for rounded corner and notch makes it more complicated.
• Today's 2020 models of new phones iphone 12 is 5.4” to 6.7” with the notch and a 4.7” “classic” iphone SE 2nd generation released in 2020 with the same soc as the iphone 11.

------

Generally I think apple usually gets the balance of factors close to right, better than most OEMs able to keep a wide range of products with a small range of skus, but this is not always the case. (There is a reason why Apple is a 2 trillion dollar company instead of a 30 billion dollar company.)

And it is not just me who have this opinion (describing myself I am an apple fan who rarely buys their products for I adore value.)

It is also the opposite extreme of apple fans, you know the Accidental Tech Podcast hyper-fans and so on . These hyper-fan often points out that Apple is slow to adopt several things even if it makes no sense to do so. (Now this is more macbook / macOS stuff and not the more cutting edge iPhone / iOS stuff. Some of this can be blamed on slow intel but even before 14nm debacle it also applied to their laptops with apple being 6 months to 3 years out of date of the latest silicon.)

 

[smalM]

Apple A14 SoC Put Under the Microscope; Die Size, and Transistor Density Calculated

Apple has established itself as a master of silicon integrated circuit design and has proven over the years that its processors deliver the best results, generation after generation. If we take a look at the performance numbers of the latest A14 Bionic, you can conclude that its performance is...

www.techpowerup.com

Says chip is 88 mm2 for 134M transistors/mm2.

I think, Techpowerup is a little bit off with the interpretation and we are looking at mostly the same layout of NPU, big CPUs, small CPUs, SLC, and GPU as on the A13 die. I/O got pushed around a little bit, memory interface stayed the same.
Hopefully, we see a real die shot soon.

Transistor density is where it was to be expected, 171 MT/mm² is for logic only and Apple doesn't push the limits of a node in their designs.

 

[name99]

This is true some of the time but not true all of the time.

Remember 3.5" phones? Apple stuck with the smaller phone sizes stubbornly even though all the data at the time showed people wanted larger phones even though people were not sure what was the final resting place for how big they wanted the phones to be.

• 3.5” diagonal (03 by 02) 5.65 inches² the original size iPhone original to iphone 4s
• 4.0” diagonal (16 by 09) 6.84 inches² aka the taller iPhone 5
• 4.7” diagonal (16 by 09) 9.44 inches² aka the taller and larger iPhone 6
• 5.5” diagonal (16 by 09) 12.93 inches² aka the taller and larger iPhone 6 plus

And since then we got a far more variety of sizes. But analysts show that android was getting US market-share until the iphone 6 generation. Likely due mainly to screen size and not other specs or features. Same thing with the “higher end” world markets that are not the United States. Though Apple’s market-share in global lower and midrange markets is a more complicated story for Apple phones are generally more expensive with the cheapest MSRP being $400 (US).

But remember the Galaxy S3 (2012) it was a 4.8" 1080p display competing against the iphone 4s (released 5 months earlier) with its 3.5" screen. 5.65 inches² compared to 9.84 inches². And when the iphone 5 was released 6 months after the S3 it was a 4.0" 16 by 09 screen compared to a 4.8" (6.84 inches² compared to 9.84 inches²) Those 3.0 inches squared were a big thing, apple fans were so happy to move from 5.65 to 6.84 inches² but many customers still wanted more.

(moving to more current phones)

• The iPhone X moved to rounded corners and notch so it is not directly comparable. It is taller (13x6, aka 19.5 by 09) 5.8” diagonal. (Not going to do area comparisons for rounded corner and notch makes it more complicated.
• Today's 2020 models of new phones iphone 12 is 5.4” to 6.7” with the notch and a 4.7” “classic” iphone SE 2nd generation released in 2020 with the same soc as the iphone 11.

------

Generally I think apple usually gets the balance of factors close to right, better than most OEMs able to keep a wide range of products with a small range of skus, but this is not always the case. (There is a reason why Apple is a 2 trillion dollar company instead of a 30 billion dollar company.)

And it is not just me who have this opinion (describing myself I am an apple fan who rarely buys their products for I adore value.)

It is also the opposite extreme of apple fans, you know the Accidental Tech Podcast hyper-fans and so on . These hyper-fan often points out that Apple is slow to adopt several things even if it makes no sense to do so. (Now this is more macbook / macOS stuff and not the more cutting edge iPhone / iOS stuff. Some of this can be blamed on slow intel but even before 14nm debacle it also applied to their laptops with apple being 6 months to 3 years out of date of the latest silicon.)
[/QUOTE]

Calling this behavior "stubborn" is perhaps not correct. A constant limitation in what Apple can do is volume (and the company policy that they don't want to diverge SKUs too much along various dimensions).

In other words it is quite possible that Apple did not move to larger screens not because they didn't want to, but because no-one could produce the volume of screens they required at the quality level they required.
We've certainly seen this in other transitions. I'm guessing their not moving to LPDDR5 last year was absolutely driven by volume, this year may still be constrained by volume. (It will be interesting to see if they adopt LPDDR5 in lower volume products like the iPad Pro's and the new Macs.)

 

[AmericanLocomotive]

What's the value add for a battery that lasts a week? Going on a trip and not having to bring a charger?

I think any battery life above a day is useless, and I'd rather have a lighter phone. I charge my iPhone 11 pro max every other day, so FOR MY USE if the battery was less and it was lighter I think that would be good. If I only charged it every third or fourth day that would add ZERO value as far as I'm concerned.

The idea that no battery is ever enough unless it is perpetual is ludicrous. Charging is hardly a terrible chore, it isn't like you have to hop on a bike and generate the electricity yourself.

If the only thing you use your phone for is checking text messages a few times a day and other basic things, than yeah, you battery can last a while. But for people that play games on their phones, or do a lot of heavy web browsing, they can easily deplete the battery in a day or less. You don't always have an opportunity to charge it, either. A battery that lasts 2-3 days of "normal" use would be a boon to people who like to travel, and may go long periods without access to a charger. Long night out and crashed at a friends house? No problem, because you have tons of battery life. Car died in the middle of nowhere at night? No problem, plenty of battery life.

You have Apple's heaviest phone, with Apple's largest battery, and you're saying you'd rather have a "lighter" phone with less battery? What? In what reality does the "weight" of a phone matter? You will NOT notice an extra 50-60 grams in your pocket. Want a lighter phone? Get something that doesn't use heavy and fragile glass for the back. Nor would you notice an extra 1mm of thickness that'd allow for 30% more battery. Especially since you already have one of Apple's largest phones.

Furthermore, nearly 80% of smartphone owners put some kind of case on their phone. So the whole "I'd rather have a slimmer, lighter device" argument goes right out the window when the vast majority of people are equipping their phone with a giant, heavy case anyways.

 

[Roland00Address]

Calling this behavior "stubborn" is perhaps not correct. A constant limitation in what Apple can do is volume (and the company policy that they don't want to diverge SKUs too much along various dimensions).

In other words it is quite possible that Apple did not move to larger screens not because they didn't want to, but because no-one could produce the volume of screens they required at the quality level they required.
We've certainly seen this in other transitions. I'm guessing their not moving to LPDDR5 last year was absolutely driven by volume, this year may still be constrained by volume. (It will be interesting to see if they adopt LPDDR5 in lower volume products like the iPad Pro's and the new Macs.)

Okay I hear you, and this is a valid point.

But the people complaining about Apple not getting feature or spec X right (such as wanting more battery) are not going to be persuaded by that if they are already comparing Apple products to competitors such as Samsung and so on and those competitors got feature or spec X right.

We've certainly seen this in other transitions. I'm guessing their not moving to LPDDR5 last year was absolutely driven by volume, this year may still be constrained by volume. (It will be interesting to see if they adopt LPDDR5 in lower volume products like the iPad Pro's and the new Macs.)

Is anyone shipping LPDDR5 products right now? I thought the spec was only announced in 2019 like the first 3 months or so, and we are only getting memory companies announced they have completed their process and can start selling those products in the last 30 days.

 

[name99]

Okay I hear you, and this is a valid point.

But the people complaining about Apple not getting the X right (such as wanting more battery) are not going to be persuaded by that if they are already comparing Apple products to competitors such as Samsung and so on.


Is anyone shipping LPDDR5 products right now? I thought the spec was only announced in 2019 like the first 3 months or so, and we are only getting memory companies announced they have completed their process and can start selling those products in the last 30 days.

Samsung "announced" they had chips in mid 2018:
https://www.anandtech.com/show/13084/samsung-announces-first-lpddr5-64gbps-data-rates
Apparently SS has gone all-in on Intel style announcements where they go through 5 different announcements of something-or-other before you actually buy it, 2+ years after the first announcement.

More legit is Micron in Feb 2020
https://www.anandtech.com/show/15480/micron-shipping-lpddr5-dram
claiming volume production, but, like I said, how large are the volumes?

There are at least a few flagship phones (like ZTE) with LPDD5 but I don't know how widespread.

 

[Geranium]

Is anyone shipping LPDDR5 products right now? I thought the spec was only announced in 2019 like the first 3 months or so, and we are only getting memory companies announced they have completed their process and can start selling those products in the last 30 days.

Galaxy Note 20/20 Ultra, S20/Ultra/FE, Z Flip, Z Fold 2, Xiaomi Mi 10/10T/10T Pro +some others uses LPDDR5. But Phone with LPDDR5 doesnt exist cause Apple not using it.

 

[Lodix]

Almost any smartphone over $500 this year has LPDDR5 RAM inside. So millions of devices.

 

[name99]

Almost any smartphone over $500 this year has LPDDR5 RAM inside. So millions of devices.

The difference is millions vs 200..250 M for iPhones...

Now could (or should) Apple eg make iPhone Pro's more Pro by giving just them LPDDR5 for this year? It's hard to know. There's real value in having the devices be so similar as far as developers are concerned, and in terms of people not think that the iPhone is substantially less than the iPhone Pro. On the other hand there's also value to people paying Pro prices getting an all-round better basket of goods.

My guess is this debate goes on within Apple, and at some point maybe we will start to see such distinctions, the way there are deeper techy distinctions between an iPad Pro or MB Pro than just an iPad or MacBook.

 

[Eug]

The difference is millions vs 200..250 M for iPhones...

Now could (or should) Apple eg make iPhone Pro's more Pro by giving just them LPDDR5 for this year? It's hard to know. There's real value in having the devices be so similar as far as developers are concerned, and in terms of people not think that the iPhone is substantially less than the iPhone Pro. On the other hand there's also value to people paying Pro prices getting an all-round better basket of goods.

Apple never mentions RAM in their iPhone presentations or in their iPhone specifications.

 

[name99]

Apple never mentions RAM in their iPhone presentations or in their iPhone specifications.

That's not the only factor. Pro's now have 6GB as opposed to 4 for the non-Pro models, but they also don't mention that in the keynotes.

The real issue is - do they, at some point, want to establish a SPEED differential between the Pro models and the non-pro models, to go along with the other differentials (better camera, better screen, now more RAM, ...)? Once that decision is accepted, then a better speed grade of RAM is simplky a means to an end.

I think the time has passed for this particular decision. Next year presumably will be LPDD5 across the board, no obvious reason not to.
But when LPDDR6 rolls around, and the decision has to be made?

 

[IntelUser2000]

A14's cache doesn't scale well:

Apple A14 Die Annotation and Analysis – Terrifying Implications For The Industry

Despite TSMC’s claims of a 1.35x shrink on SRAM, Apple’s system cache has only shrunk 1.22x. This has far reaching implications for the industry. Designers cannot use increased LLC sizes as a crutc…

semianalysis.com

Despite TSMC’s claims of a 1.35x shrink on SRAM from N7 to N5, Apple’s 16MB system cache has only shrunk 1.19x.

This is what I mean by being a half-node. 5/7 meant 5 squared / 7 squared for an approximate 50% reduction in size in traditional terms. Now, 5/7 means 5/7. You'll need the 3nm node to get a full node reduction in density and power reduction over 7nm.

According to that article, caches are showing even lower gains, at 19-20% range.

 

[name99]

A14's cache doesn't scale well:

Apple A14 Die Annotation and Analysis – Terrifying Implications For The Industry

Despite TSMC’s claims of a 1.35x shrink on SRAM, Apple’s system cache has only shrunk 1.22x. This has far reaching implications for the industry. Designers cannot use increased LLC sizes as a crutc…

semianalysis.com

This is what I mean by being a half-node. 5/7 meant 5 squared / 7 squared for an approximate 50% reduction in size in traditional terms. Now, 5/7 means 5/7. You'll need the 3nm node to get a full node reduction in density and power reduction over 7nm.

According to that article, caches are showing even lower gains, at 19-20% range.

I wouldn't build too ambitious a case on that. As I have said repeatedly, everything about the A14 screams "placeholder". It's basically the bare minimum set of changes they needed to put in place to get something appropriate for Macs.

This isn't a criticism. Apple's pace since the A7 has been astonishing, and they managed to get the A14 and its suite of HW out more or less on time even in spite of covid. But even the greatest design team in human history isn't omnipotent. A lot of things have to come together to transition from an A12X/A13 to the sort of Xeon class CPU's required for Mac Pros. (Obviously support for many cores, support for lots of IO, some sort of story for dGPU, some sort of story for lots of [variable sized?] RAM, a start on the sort of Gen-Z/Optane tier of persistent RAM, a hypervisor story, SVE as an alternative to AVX-512, etc etc.)

It's only reasonable (and basically common sense, the same sort of cautious approach that has worked so well for TSMC) to split the job into at least two stages (A14, then A15) and to make as few changes to the already working and well understood parts of the A13 base as possible.

We see this in multiple ways. The way A14 IPC is essentially unchanged, so unlike every previous core. They appear to have basically taken what they had, ported it to 5nm (in the fastest, safest, easiest way possible), enjoyed the 15% speed boost, and called it a day.
There do appear to be SOME internal changes, but again only those necessary for the larger project -- hypervisor support was added, and apparently a 4th NEON unit. (Do we get SVE, likely 2x256 on top of those 4 NEON units? My hope is yes, and that this has so far not been announced as a combination of XCode not being ready yet, and having something big and new to add at the Nov 10 event.)

So with all this background, my guess is that it was simply considered not important this time round to optimize the L3 (and much of the rest of the chip) to anything like the level we've seen before. Take the existing cells, shrink them as much as TSMC says is safe without changing *anything* (timing, voltage levels, ...) and call it a day.

It will be interesting to see how the somewhat less of a repeat of an A13 chips (A14T, and especially Lifuka) compare in this respect. I'd expect them to be slightly denser, insofar as whatever is genuinely new (not a copy/paste from 7nm) will make more appropriate use of 5nm capabilities. But once again those chips have been put together in a hurry; the optimizations have not yet come.

The A14 is the Mac's A6.
It will take the A15, the Mac's A7, to get excited at what's really becoming possible, and then one or two optimization rounds before we really start flying.
Honestly I think if TSMC delays 3nm one year, so that we see A15 and A16 all on 5nm variants, Apple won't press the point at all.