I think in practice the performance increase of A16 will be much more than on paper specially because of the 50% memory bandwidth increase. Also new n4 will cause less thermal problem and the general performance will be noticeable very well. Thanks for awsome video

Very insightful content as always. Overall from your commentary I can gather that there's still a tangible improvement over the A15, given that the LPDDR5 and optimized process node reinvigorate the "new" chip, especially in the power refficiency department. Still settling into your new home?

Everest is the new name for the 2× performance cores, and Sawtooth is the 4× efficiency core of the A16 Bionic.

Brilliant. Very insightful. Thanks

It seems like that M3 will be based on A17 instead of A16. This way it completes 16-20month refresh cycle that AX / M chips are using (A8X, A10X, A12X, A14X / M1), while A9X and M2 are bunch of stopgap chips that aren’t much faster than their predecessors. M3 based on A16 will offer very little gain in performance, even less than A15 from A14. At the same time M3 and A17 are rumored to give significant gains over the previous A15/M2 which is impossible had M3 is based on A16.

It's always the efficiency gained from N4 process node Just a bit higher clockspeed and high bandwidth for the cpu Soo it's actually the same chip

Really enjoying this new channel. Thanks for the in-depth analysis! I was wondering if A16 was worth it over A14 in the iPhone 12 Pro Max. Looks like it might be worth waiting for iPhone 15.

It's new because it's in the new iPhone, duh. Revolutionary! Innovative! Cutting edge!

great video!

We miss you at Kieser Training, wish you all the best :)

Although it may be a huge risk with other companies/brands, if the opportunity was there I could see one possible situation to explain it. They may have made a unusually large performance jump via design. They may have decided to use the fact that they always sell new IPhones to go ahead and build that chip in a HUGE production run and "BINd it" down for the previous gen and then used the next highest BINs as the "next". The amount of money that would save and make them would be incredible! They could use the first run to get data and take their time with software to optimize the ____ out of the "2nd gen". It makes sense. A majority of IPhones are not sold with any consideration into hardware performance other than a few obvious key features like camera and......I dont know of anything else moat people care about besides maybe battery life and 5G. It is only a matter of time until that is fully exploited with every model. Hardware construction quality has definitely gone way down in the past 10+ years imo, so the cuts to maintain those margins indefinitely will be coming from other places. If they did do that and made a ton of money, investors are going to expect it regularly. Anyway, that my cynical take on the issue. It could be countless other things just as likely and many even more likely than that, but it's what comes to my mind and I felt like jabberin about it. Nice video as always! I loved that you actually showed the process needed to look past some evasive Marketing techniques we missed even when actively looking already. It's kinda greasy imo. I don't appreciate that kinda thing. I don't get all mad about it, but I almost take offence at that kind of Ad "campaign". The ones where they are clearly going out of their way to hide or omit that common expectations people might have come to have about a product are not present in said product being marketed. Especially when itnwas their marketing that imposed and maintained those expectations in the first place. It's also not hard to imagine that people would probably feel at least a little annoyed if/when they noticed. Which kinda tell me they don't care and are banking the volume of the majority that don't notice/care. Not just Apple of couse and this example isn't quite as bad as most others, but it kinda shows that it is becoming a mainstream form of marketing that use to be reserved for Ads airing weeknights on Cable TV at 3:15AM. Everything feels shady anymore to me. It probably is, but I can still be disappointed about it right :-D

Good analysis. Nothing I'd disagree with at all!

I don't know about the phones, but on the M1/2 macs the display engine has it's own CPU. They are very similar to the iphone SoCs, so that's probably what they mean.

I am upgrading from my XS Max to a 14 Pro Max 1TB black. The reason I am upgrading is because of the crash alert safety feature that has the ability to save human lives in the event of a crash by summoning help faster and the Satellite SOS feature that again has a potential to save human lives if ever needed. I am not really concerned about the chip speed so it’s all good. That’s my two cents anyway.

Architectural change is hard, very hard and time consuming. Even mighty intel with massive engineering resource can afford fresh architectural change once in few years. Evolution from old Skylake to next Icelake took 4 years. So it is not surprising or disappointing that Apple re-used old architectures of yesteryear, which outperforms at least 2 years ahead of its closest competitors like Qualcomm or Arm Inc. I have no doubt that Apple Silicon will maintain such 2 years performance lead over competitors in foreseeable future.

Great video after all that being said seems like Apple will discontinue the 14 pro line up

This is just a crazy thought…. What if the limits of lithography only allow gains that can measured at .x nm? So imagine that the 5 nm process doesn’t include the 5.36,5.44,5.49 nm wide transistors? Technically, any transistor that is up to 5.49 nm size could be counted as a 5nm size transistor. It’s just a thought.

Nice!

It’s basically a rebranded A15 with LPDDR5, yeah it’s a little better but won’t be noticeable. Besides 4NM is more branding than anything else it’s frankly an optimized version of N5P, which was an optimized form of the original 5NM from TSMC. The “real” node changes will come with 3NM which should see a slightly noticeable perf/watt bump.

2:35 - You said all is just "clock" related, but not architecture... than see A14 vs A15.. same difference and was different architecture + double the system cache.

I've been very impressed with the performance with the A16 bionic, it never runs hot no matter how much I push the GPU. I really wish the M2 Pro and M2 Max chips were on this n4 architecture.

Honestly I made the jump from an Iphone 7 plus so for me personally, its a big step and much needed for myself.

With the A16, "the elephant birthed a rat" to quote a Brazilian saying. Zero IPC gains, minimal performance/efficiency gains purely from the new node and even that is pretty small as TSMC drags their feet with their real next-gen node. Realistically, expecting more wouldn't be fair since no CPU architecture makes noticeable progress every year; the normal pace from leaders like Intel and AMD is 18-24 months between incremental updates, and they can only handle larger changes by having two completely separate design teams since major new archs have a design phase longer than 2 years. But it's Apple that created this expectation of significant CPU improvements every year, since they want people to upgrade devices, in particular the iPhone, every year. And the Apple fanboys ate up that hype since the original Apple Silicon SoCs, predicting imminent doom of the x86 and also RTX and Radeon. Another thing. It's (relatively) easy to make fast progress with full node jumps since you have much bigger transistor & thermal budgets to add features and optimizations: more cache, more cores, more execution units, bigger traces for uops or branch prediction etc. But it's much harder to make noticeable progress within the same or almost-same node, like AMD for example did multiple times in recent iterations (Zen2 to Zen3, RDNA1 to RDNA2; all TSMC 7nm).

Yes there’s not that much change, if we look at the neural engine, 17 tflops, as opposed to 15.8 tflops, it was a huge increase from 6 tflops, in the last iPad mini 5, compared to the iPad mini 6, at 15.8 tflops. My M1 Mac mini only runs at 11 tflops, less than my iPad mini 6, running an A15 chip at 15.8 tflops even though both use the 5nm chip density. The M2 runs at 16 tflops, roughly the same as the A15 iPad mini 6, as you say, the memory is important, the M1 uses on the chip die unified memory and I’m guessing PCIe4 ssd flash. My Ryzen 7 Asus mini PC, seems to take a lot longer, to do a factory reset, than my Mac mini M1, since recent software updates, it seems to be able to take advantage of the unified memory and PCIe4 memory, although I have a PCIe4 fan assisted backup stick, which runs in thunderbolt input output. Even so before I begin loading non operating system software, it does seem to take longer, even though I have 32GB of RAM, on the mini computer, at 8 cores, in 7nm, still the Mac mini M1, is 5nm. Further my mini computer, has no neural network transistors, unlike the newer AMD chips, which run DDR5, PCIe4+5, in 5nm, but I can run MS Server 2022, on the mini computer. Even though I have to run it, on a separate flash drive, no dual booting, any more.

I wonder if Apple would have used an ARMv9 design if 3nm was ready on time.

not worth getting the upgrade! great video

I thought it was confirmed there are new cores ?

Well, that being said, A17 will blow absolutely everything out of the water

Apple are victims of their own hype due to global chip problems the A16 is just a mild refresh of the A15 due to the 4nm process, their marketing wank department has to make every new iphone seem like it is a quantum leap forward to get the idiots to ditch their perfectly good phone for a new one every year. The iphone 14 is just a mild makeover of the 13 the 14 max is just a 13 in a bigger case with a bigger battery, the 14 pro is just a 13 pro with a better camera. There is no compelling reason to change from a 13 series to a 14 series so Apple had to try some smoke and mirrors with their "dynamic island" and "photonic engine" to make these phones seem "new"

Lolol, anyone for an Apple Newton revival?

So much speculation... the main reason apple avoided mentioning or comparing to the A15 is because the non pro model is still shipping with the A15.. So it be best to avoid using the A15 for comparison. The transistor count is different although not by a big number.

Very informative. Now, I really want to see how Apple will address "Mac Pro" chip issue. It's quite hard for them to scale current design to "Pro level", both CPU & GPU.

Given we probably don’t need massive performance improvements from an already top line chip, let’s see instead what this does to battery life…

No difference will be noticed in opening and using normal apps or games

It's new. We have new architecture. Everest for the BIG cores and Sawtooth for the small ones. From "leaks" ARMv9 is very likely to be in the chip.

I think the new chip is great. Apple cannot revolutionize the chip world every year and they are already far ahead of everyone else so just buy the new iPhone and enjoy it.

new Leaked geek-bench scores suggest that the A16 is 17% faster than A15 in cpu multi core score

ALWAYS ON )challenging like 5 years) BY SAMSUNG AND SELL FOR CHEAP CELLPHONE _ APPLE NEW NOT HAPPENING 48MP I Have 108 and 200 on Motorola by SAMSUNG

15:21 You are german which means the price of the iPhone 14 series increased for you so I don't know what you're talking about 😂😂 Your analysis is good but fails to take into account that the small upgrade isn't because of N3 delays but rather that Apple's chip division lost a lot of talent and is currently struggling to bring the usual improvements while TSMC is actually saving them by giving them a manufacturing advantage over the competition.

Its new cause its 4nm

6 cores 2+4 i have 10 8GEN1 QUALCOMM - and don't show me graps I can do better and test not on Same IOS worthless- 0.4 Low price not whey sell very High Price

@maxtech waiting on your take