Discussion Apple Silicon SoC thread

[Eug]

M1
5 nm
Unified memory architecture - LP-DDR4
16 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 12 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache
(Apple claims the 4 high-effiency cores alone perform like a dual-core Intel MacBook Air)

8-core iGPU (but there is a 7-core variant, likely with one inactive core)
128 execution units
Up to 24576 concurrent threads
2.6 Teraflops
82 Gigatexels/s
41 gigapixels/s

16-core neural engine
Secure Enclave
USB 4

Products:
$999 ($899 edu) 13" MacBook Air (fanless) - 18 hour video playback battery life
$699 Mac mini (with fan)
$1299 ($1199 edu) 13" MacBook Pro (with fan) - 20 hour video playback battery life

Memory options 8 GB and 16 GB. No 32 GB option (unless you go Intel).

It should be noted that the M1 chip in these three Macs is the same (aside from GPU core number). Basically, Apple is taking the same approach which these chips as they do the iPhones and iPads. Just one SKU (excluding the X variants), which is the same across all iDevices (aside from maybe slight clock speed differences occasionally).

EDIT:



M1 Pro 8-core CPU (6+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 16-core GPU
M1 Max 10-core CPU (8+2), 24-core GPU
M1 Max 10-core CPU (8+2), 32-core GPU

M1 Pro and M1 Max discussion here:

Page 78 - Discussion - Apple Silicon SoC thread

Page 78 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M1 Ultra discussion here:

Page 109 - Discussion - Apple Silicon SoC thread

Page 109 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M2 discussion here:

Page 127 - Discussion - Apple Silicon SoC thread

Page 127 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

Second Generation 5 nm
Unified memory architecture - LPDDR5, up to 24 GB and 100 GB/s
20 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 16 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache

10-core iGPU (but there is an 8-core variant)
3.6 Teraflops

16-core neural engine
Secure Enclave
USB 4

Hardware acceleration for 8K h.264, h.264, ProRes

M3 Family discussion here:

Page 215 - Discussion - Apple Silicon SoC thread

Page 215 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M4 Family discussion here:

Page 263 - Discussion - Apple Silicon SoC thread

Page 263 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[Doug S]

It will lack mmWave and it won’t be as fast and the call reception won’t be as good Qualcomms in the first iteration.

Apple had a hard time with developing modems

None of that really matters though. 5G is already capable of far more speed than anyone needs in a phone. Qualcomm's X80 modem supports theoretical speeds of 10 Gbps down and 3.5 Gbps up. If I could get only 10% of those theoretical numbers it wouldn't make the tiniest bit of difference to me, and I think the same is true most of us. A 90% reduction in theoretical speed would not translate to anything like a 90% reduction in actual speed - those numbers rely on connecting to a bunch of antennas at once, but if you have a congested cell you won't be able to do that and if you don't even one or two antennas will get you hundreds of Mbps.

I'm not gonna care if someone does a test and sees say 400 Mbps on 5G with Qualcomm and 200 Mbps with Apple when testing from the same locations. What matters is how little signal you can have and still get say 5 Mbps, and how little signal you can have and maintain a call. Since as you say Apple has had a long hard road getting here there's no way it will match Qualcomm, the question is how far behind are they and whether it is far enough to really matter. Intel modems were behind Qualcomm's when Apple used them, while there was a lot of loud people in forums and youtube, it didn't make much difference in actual use (I had one and didn't notice any difference compared to my older phone that used a Qualcomm modem)

I'll be curious to learn whether it is LTE/5G only as I've long predicted, and how its power use compares to Qualcomm. That's where you'll notice if, we've all seen how quickly using cellular especially with a weak signal burns down your battery compared to wifi. I expect that's what Apple is focusing on power rather than supporting 6x MIMO like Qualcomm, and while I don't expect them to be as efficient as Qualcomm in version 1.0, I'm curious how far behind they are and how much improvement they make with 2.0.

 

[jdubs03]

Yeah version 2 will be what I’m looking for to see how their progress is going. But as a reminder (according to Gurman) Apple is planning to exceed Qualcomms’ top-line offering in version 3 in 2027.

I’d imagine they won’t put their own version in their Pros until at least version 2. But I’m skeptical of that and think it’d be version 3.

 

[mvprod123]

iPhone 16e

Apple debuts iPhone 16e: A powerful new member of the iPhone 16 family

Apple today announced iPhone 16e, a new addition to the iPhone 16 lineup that offers powerful capabilities at a more affordable price.

www.apple.com

Apple today announced iPhone 16e, a new addition to the iPhone 16 lineup that offers powerful capabilities at a more affordable price. iPhone 16e delivers fast, smooth performance and breakthrough battery life, thanks to the industry-leading efficiency of the A18 chip and the new Apple C1, the first cellular modem designed by Apple.

iPhone 16e is powered by Apple’s latest-generation A18 chip, which enables fast, smooth performance, incredible power efficiency, and Apple Intelligence. The 6-core CPU is up to 80 percent faster than the A13 Bionic chip on iPhone 11, handling both everyday and intensive tasks with ease — from simple workloads, to more demanding actions with Apple Intelligence. The 4-core GPU powers stunning graphics performance and unlocks next-level mobile gaming on the go, enabling graphically demanding AAA titles and hardware-accelerated ray tracing for more realistic lighting and reflections. The 16-core Neural Engine is optimized for large generative models and runs machine learning (ML) models up to 6x faster than A13 Bionic.

Expanding the benefits of Apple silicon, C1 is the first modem designed by Apple and the most power-efficient modem ever on an iPhone, delivering fast and reliable 5G cellular connectivity. Apple silicon — including C1 — the all-new internal design, and the advanced power management of iOS 18 all contribute to extraordinary battery life.

 

[mvprod123]

https://twitter.com/x/status/1892245307120861341

 

[Nothingness]

Why am I not surprised Apple went for power efficiency first? Now let's see what the reality is

 

[mvprod123]

 

[Eug]

https://twitter.com/x/status/1892245307120861341

With 16e, the price went up dramatically compared to the iPhone SE 3, but that's fine. I always thought the SE 3 sucked, as it had a screen that was too small for most of the market, and the battery life was absolutely horrible. I wonder how much of the battery improvement is truly due to C1 though. Cuz now that the phone is much bigger, it will likely have a much bigger battery.

Although the price has increased, the 16e gets Face ID and double the storage, and competes favourably price-wise compared to used iPhone 15 pricing, so I think it will sell well. It will also be very popular with the carriers as the "budget" option for free or cheap iPhones. I do know around here, the kids looking for a budget iPhone have been shunning the iPhone SE 3, and have been buying older used non-Pro iPhones, so this new 16e addresses that market nicely.

It will be interesting to see the performance benchmarks of C1. I personally don't really care about max cellular speeds but I care about connection reliability in areas of borderline signal. However, it seems Apple is likely sticking with Qualcomm for the 17 Pro Max I will be buying this year.

 

[name99]

View attachment 117611

I find the satellite features to be the most interesting technical aspect.
You'd figure that these could be omitted from the lowest-end phone if they cost anything to add. So does that mean that the C1 contains
- cellular modem (of course)
- GPS (presumably, that how the packaging normally works)
- AND satellite connectivity?

The modem seems to be as expected
- no 6GHz and no mm Wave (though the digital part may handle these, just the RF front end is omitted?)
- deep bwd compatiblity, so both TDD and FDD (so works in China!), and HSPA/+ (works back to 3G) and even EDGE/GSM. But apparently no voice CDMA 🤣 No point in picking a fight with QC over something no-one cares about anyway!
- 4x MIMO and Gigabit, so while its not starting at the highest end, it's also not starting at the low end

One remaining question is whether the WiFi and BT are also Apple (either in C1 or as an updated W chip). Presumably teardowns on Feb 28 will tell us.

 

[Eug]

Where is mmWave deployed anyway besides in the US? Here in Canada, it essentially doesn't exist. Our latest iPhones don't have it either.

 

[poke01]

Where is mmWave deployed anyway besides in the US? Here in Canada, it essentially doesn't exist. Our latest iPhones don't have it either.

It’s non existent in Australia as well

 

[mvprod123]

With 16e, the price went up dramatically compared to the iPhone SE 3, but that's fine. I always thought the SE 3 sucked, as it had a screen that was too small for most of the market, and the battery life was absolutely horrible. I wonder how much of the battery improvement is truly due to C1 though. Cuz now that the phone is much bigger, it will likely have a much bigger battery.

Although the price has increased, the 16e gets Face ID and double the storage, and competes favourably price-wise compared to used iPhone 15 pricing, so I think it will sell well. It will also be very popular with the carriers as the "budget" option for free or cheap iPhones. I do know around here, the kids looking for a budget iPhone have been shunning the iPhone SE 3, and have been buying older used non-Pro iPhones, so this new 16e addresses that market nicely.

It will be interesting to see the performance benchmarks of C1. I personally don't really care about max cellular speeds but I care about connection reliability in areas of borderline signal. However, it seems Apple is likely sticking with Qualcomm for the 17 Pro Max I will be buying this year.

Apple promises better battery life than regular iPhone 16

 

[mvprod123]

https://twitter.com/x/status/1892251158884040810

 

[okoroezenwa]

Apple promises better battery life than regular iPhone 16

View attachment 117628

Yeah this is the real comparison. At some point we’ll know the battery capacity of the 16e and can estimate how much is gained from changing modems (and going with a 4-core GPU).

 

[Eug]

Apple promises better battery life than regular iPhone 16

View attachment 117628

Yes but Apple also dropped MagSafe, which takes up some room. I’m wondering if the battery in the 16e is bigger than even the 16 battery.

Anyhow, I don’t doubt C1 is more efficient, but how much more I don’t know.

 

[mvprod123]

Some new interesting info from Reuters:

https://www.reuters.com/technology/apple-reveals-first-custom-modem-chip-shifting-away-qualcomm-2025-02-19/

The C1 subsystem is the most complex technology Apple has ever built, with a baseband modem manufactured with advanced 4-nanometer chipmaking technology and a transceiver made with 7-nanometer technology, Johny Srouji, Apple's senior vice president of hardware technologies, said in an interview at one of Apple's silicon labs in Sunnyvale, California.

The chips had to be tested with 180 carriers in 55 countries to ensure they will work in all the places Apple ships iPhones.
"We build a platform for generations," Srouji said. "C1 is the start, and we're going to keep improving that technology each generation, so that it becomes a platform for us that will be used to truly differentiate this technology for our products."
One of the ways Apple hopes the C1 will set its iPhones apart is by tightly integrating it with its processor chips.
For example, if an iPhone encounters congested data networks, the phone's processor can signal to the modem which traffic is the most time sensitive and put it ahead of other data transfers, making the phone feel more responsive to the user's needs, said Arun Mathias, vice president for wireless software at Apple.

The C1 chips also have custom GPS systems and satellite connectivity for when iPhone users are away from mobile data networks. But they will lack some features, such as the ability to connect to what are known as millimeter wave 5G networks.
Millimeter-wave technology is one of Qualcomm's strongholds. Apple executives declined to say when the company's chips will have the technology or how quickly Apple will phase out Qualcomm's chips.

Apple's Srouji said that the company's goal was not to match the specifications of its chip rivals but instead to design products specific to the needs of Apple products.
"We're not the merchant vendor to go compete with Qualcomm and MediaTek and others. I believe we're building something truly differentiating that our customer will benefit from," he said.

 

[Eug]

The higher end iPhone 17 models are rumoured to have Qualcomm again, because of mmWave. However, given that most of the world doesn't use mmWave anyway (and the iPhones in those locations don't support mmWave either), it is theoretically possible they could use C1 in those outside the US. However, I don't think they will do that.

 

[Doug S]

I find the satellite features to be the most interesting technical aspect.
You'd figure that these could be omitted from the lowest-end phone if they cost anything to add. So does that mean that the C1 contains
- cellular modem (of course)
- GPS (presumably, that how the packaging normally works)
- AND satellite connectivity?

Why is that surprising? Their Globalstar satellite uses L band which is smack in the middle of most of the cellular frequencies, so there is very little additional hardware required for them to handle it.

Once you support midband cellular you're into the 5/6 GHz wifi bands so you could build that into chip, along with BT. I'm not sure if the talk about Apple designing their own wifi/BT chip will actually mean a separate chip. Maybe at first, but it'll either become part of C2 or C3 or be made part of the SoC. I expect they'll integrate cellular onto the iPhone SoCs someday, but I doubt they would for Apple Silicon. Pointless to waste silicon area on something not very many people need or want in a PC.

 

[Doug S]

The higher end iPhone 17 models are rumoured to have Qualcomm again, because of mmWave. However, given that most of the world doesn't use mmWave anyway (and the iPhones in those locations don't support mmWave either), it is theoretically possible they could use C1 in those outside the US. However, I don't think they will do that.

They need time to get some real world feedback on C1 before they feel comfortable putting it into a premium iPhone. If you figure they need to tape out C2 by say December to make 2026 iPhones that will give them several quarters worth of real world data to know what to fix. I would be shocked if C1 doesn't already support midband and mmwave, so that they can get some real world testing using special models. They just don't need it on the 16e (the SE 3 didn't do mmwave either) and since that's the most difficult to handle hardware design wise it makes sense to leave it off at first.

More importantly than the feedback on the C1 hardware they'll be able to get feedback on the baseband software. I expect that a lot of early teething problems we hear about will be addressed with software updates over the year. That's by far the more difficult part of doing this compared to the modem chip design, and that's where most the problems with stuff like carrier compatibility issues, dropped calls or poor handoff between cellular and wifi calling etc will crop up.

 

[Doug S]

Some new interesting info from Reuters:

https://www.reuters.com/technology/apple-reveals-first-custom-modem-chip-shifting-away-qualcomm-2025-02-19/

The C1 subsystem is the most complex technology Apple has ever built, with a baseband modem manufactured with advanced 4-nanometer chipmaking technology and a transceiver made with 7-nanometer technology

That's pretty aggressive, I suppose that's how Apple can beat Qualcomm power wise - by using a better process. I wonder if the reason why Qualcomm embedded modems on Snapdragon SoCs beat their discrete modems power wise is simply due to process advantage? Maybe there isn't any real advantage integrating modems beyond that?

 

[Mopetar]

Having everything on the same die is almost always better from a power perspective even if the node isn't as advanced. Better process node certainly helps, but I don't think the differences are as large as they once were historically in that regard. I suspect they realized that they probably couldn't achieve the same or better performance with their initial release, but recognized that they could abandon that goal and instead aim for higher efficiency.

Being able to better integrate with their own hardware will add some advantages there naturally, and having something to market on for an initial release is important. It's almost better for them to only 90% as good in most areas, but better in one key one than to only be 95% as good across the board.

Where they go from here is going to be more interesting than what they have now. I'm sure they want to have the same undisputed best crown that their SoCs have, but Qualcomm is no slouch.

 

[SpudLobby]

That's pretty aggressive, I suppose that's how Apple can beat Qualcomm power wise - by using a better process. I wonder if the reason why Qualcomm embedded modems on Snapdragon SoCs beat their discrete modems power wise is simply due to process advantage? Maybe there isn't any real advantage integrating modems beyond that?

Yes, that seems exactly right. It’s not even the data movement or anything of being off-chip, just power from a leading process.

 

[SpudLobby]

Why is that surprising? Their Globalstar satellite uses L band which is smack in the middle of most of the cellular frequencies, so there is very little additional hardware required for them to handle it.

Once you support midband cellular you're into the 5/6 GHz wifi bands so you could build that into chip, along with BT. I'm not sure if the talk about Apple designing their own wifi/BT chip will actually mean a separate chip. Maybe at first, but it'll either become part of C2 or C3 or be made part of the SoC. I expect they'll integrate cellular onto the iPhone SoCs someday, but I doubt they would for Apple Silicon. Pointless to waste silicon area on something not very many people need or want in a PC.

Do we know it’s really that simple to integrate WiFi/Bluetooth onto the same cellular chip?

 

[jdubs03]

I would expect they’re aiming for the C3 in fall 2027 to be on the level of the Snapdragon that would release early 2028. That’s what Gurman was mentioning a few months back. Perhaps the non-pro iPhone 18s will have the C2, but then the Pro series gains the C3. I’ve been consistent that Apple wouldn’t want their top of the line product to not be at the cutting edge at the time of release.

They say they don’t want to compete on specifications of their competitors. But I think that’s downplaying this release and/or modesty. You compete to win, not be 2nd place.

 

[mikegg]

They say they don’t want to compete on specifications of their competitors. But I think that’s downplaying this release and/or modesty.

I think it's an admission that it's won't be as good as Qualcomm's modems. They're not setting expectations high.

This is a cost savings measure more than a technical one. The technical benefits are a bonus.

 

[Eug]

They need time to get some real world feedback on C1 before they feel comfortable putting it into a premium iPhone. If you figure they need to tape out C2 by say December to make 2026 iPhones that will give them several quarters worth of real world data to know what to fix. I would be shocked if C1 doesn't already support midband and mmwave, so that they can get some real world testing using special models. They just don't need it on the 16e (the SE 3 didn't do mmwave either) and since that's the most difficult to handle hardware design wise it makes sense to leave it off at first.

More importantly than the feedback on the C1 hardware they'll be able to get feedback on the baseband software. I expect that a lot of early teething problems we hear about will be addressed with software updates over the year. That's by far the more difficult part of doing this compared to the modem chip design, and that's where most the problems with stuff like carrier compatibility issues, dropped calls or poor handoff between cellular and wifi calling etc will crop up.

Reviewing the rumours, it seems that the new rumoured large sized iPhone 17 Air in Sept. will have C1 and thus no mmWave. However the other models will have Qualcomm. So, paraphrasing the pundits, for phones released in 2025:

Apple C1
iPhone 16e
iPhone 17 Air

Qualcomm
iPhone 17
iPhone 17 Pro
iPhone 17 Pro Max

This prediction seems a little off to me though, as this would suggest that iPhone 17 and iPhone 17 Air are different product tiers. It would be weird to have no mmWave on the large sized 17 Air if there is mmWave on the smaller 17.