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

M5 Family discussion here:

Page 431 - Discussion - Apple Silicon SoC thread

Page 431 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[MS_AT]

Very fun

libmalloc/doc/xzone_malloc.md at af3c5dc3a540eeec030930b35b1349f4de400206 · apple-oss-distributions/libmalloc

Contribute to apple-oss-distributions/libmalloc development by creating an account on GitHub.

github.com

Interesting they are using M$ mimalloc as base (though if I am not mistaken, Windows itself is not using it by default).

 

[Joe NYC]

From Bloomberg:

https://twitter.com/x/status/1997352821453447399

 

[mvprod123]

From Bloomberg:

https://twitter.com/x/status/1997352821453447399

Sooner or later, it would have happened.

 

[fastandfurious6]

what's the real reason?

 

[Joe NYC]

what's the real reason?

Bored?

 

[mvprod123]

From Bloomberg:

https://twitter.com/x/status/1997352821453447399

https://twitter.com/x/status/1998064266390478872

 

[Mopetar]

Either the initial report was bogus or Johny found several very large money bags on his desk when he came in to work this week. My employer should try that. I'd love my job too if they did that. For large enough bags I could be convinced to live my team as well.

 

[mvprod123]

Code leak reveals A19 iPad, M4 Air with Apple’s N1 chip likely coming soon

A new leak details Apple’s 2026 iPad strategy, bringing flagship-level chips to more affordable models.

www.macworld.com

Apple Studio Display 2 leak reveals ProMotion, HDR, A19 chip upgrades

Internal code seen by Macworld reveals major upgrades coming to Apple's Studio Display in 2026.

www.macworld.com

 

[Tigerick]

Code leak reveals A19 iPad, M4 Air with Apple’s N1 chip likely coming soon

A new leak details Apple’s 2026 iPad strategy, bringing flagship-level chips to more affordable models.

www.macworld.com

Apple Studio Display 2 leak reveals ProMotion, HDR, A19 chip upgrades

Internal code seen by Macworld reveals major upgrades coming to Apple's Studio Display in 2026.

www.macworld.com

Hmm, interesting. Apple bumps up one speed grade for low-end iPad. One reason I could think of is upcoming Apple TV 4k might get bump from A17 Pro (N3B) to A18 (N3E). Lots of people are expecting new Apple TV 4K but Apple is actually considered Apple TV SoC one step behind standard iPad. That's why Apple might launch iPad 12 and Apple TV 4K at the same time next year.

FYI, the main reason for Apple TV 4K upgrade is to support AV1 decoder which only available from A17 Pro onwards...

 

[Tigerick]

Another interesting tidbit from Macrumor regarding iPad12:

It says that J581 and J588 are the codenames for the upcoming 12th-generation ‌iPad‌, but codenames are typically sequential. Codenames are how Apple references unreleased devices in its software. In prior code leaks, J581 and J582 appeared to reference the low-cost ‌iPad‌ 12.

I still have doubts about low-cost MacBook rumors. Currently, Amazon and Bestbuy are selling M4-MacBook Air with 16+256 @ $750-$799. How much do you think Apple has to set the price for low-cost MacBook with 8+128?

 

[mikegg]

Another interesting tidbit from Macrumor regarding iPad12:


I still have doubts about low-cost MacBook rumors. Currently, Amazon and Bestbuy are selling M4-MacBook Air with 16+256 @ $750-$799. How much do you think Apple has to set the price for low-cost MacBook with 8+128?

$700 permanent price is getting more and more likely.

When it's on sale, I can see $550 - $599. So still cheaper than M4 Air.

 

[Doug S]

Hmm, interesting. Apple bumps up one speed grade for low-end iPad. One reason I could think of is upcoming Apple TV 4k might get bump from A17 Pro (N3B) to A18 (N3E). Lots of people are expecting new Apple TV 4K but Apple is actually considered Apple TV SoC one step behind standard iPad. That's why Apple might launch iPad 12 and Apple TV 4K at the same time next year.

FYI, the main reason for Apple TV 4K upgrade is to support AV1 decoder which only available from A17 Pro onwards...

Having iPad jump all the way from A16 to A19 and the Studio Display also getting A19 makes me think there will be a lot of stuff that ultimately settles on A19 as the last N3 family SoC before the price jumps with N2 (which is more of a problem at base iPad / ATV pricing than it is for iPhones - or low end Macbooks for that matter)

Wouldn't make sense to use A18 for ATV unless they have some stockpile to work through - if so they'd probably not release it until they stop selling iPhone 16 & 16e. Which could be next spring, if iPhone 16 and 16e leave the price list together and are replaced by iPhone 17e. Since there's no set time on the calendar people are expecting 17e they could run through their entire stock of A18s prior to its intro, meaning they may not need a product to soak up non-Pro A18s and can coalesce everything on A19.

Another reason for an Apple TV upgrade would be to replace Broadcom wifi/BT with Apple N1 to reduce BOM.

 

[Doug S]

$700 permanent price is getting more and more likely.

When it's on sale, I can see $550 - $599. So still cheaper than M4 Air.

Yes. For further proof of their margin space, I invite people to look at iPad Air. Comes with M3, not A series, and list price is $599 at 8/128. That SoC costs >50% more than A series (based on area) and the display costs more than a Macbook's (higher resolution, touch layer) and it uses the more expensive Broadcom wifi/BT.

They could easily sell a Macbook using A series and N1 at $599. Though maybe listing it for $699 and leaving plenty of room for promos down as low as $550 like you say and edu pricing perhaps as low as $499 for once a year back to school type deals. College students would eat that up, and after four years using a Mac when they get out in the workforce they'll want one. That's probably the reason more and more corporate IT is supporting Macbook as an option.

 

[eek2121]

Having iPad jump all the way from A16 to A19 and the Studio Display also getting A19 makes me think there will be a lot of stuff that ultimately settles on A19 as the last N3 family SoC before the price jumps with N2 (which is more of a problem at base iPad / ATV pricing than it is for iPhones - or low end Macbooks for that matter)

Wouldn't make sense to use A18 for ATV unless they have some stockpile to work through - if so they'd probably not release it until they stop selling iPhone 16 & 16e. Which could be next spring, if iPhone 16 and 16e leave the price list together and are replaced by iPhone 17e. Since there's no set time on the calendar people are expecting 17e they could run through their entire stock of A18s prior to its intro, meaning they may not need a product to soak up non-Pro A18s and can coalesce everything on A19.

Another reason for an Apple TV upgrade would be to replace Broadcom wifi/BT with Apple N1 to reduce BOM.

There is also a small, growing list of games that either require an something newer than the A15 or more RAM.

Rant: more developers need to take the time to make apps and games compatible with the Apple TV. For some games like Final Fantasy, it blows my mind that there isn’t Apple TV support.

 

[Jan Olšan]

Either the initial report was bogus or Johny found several very large money bags on his desk when he came in to work this week. My employer should try that. I'd love my job too if they did that. For large enough bags I could be convinced to live my team as well.

The initial report said that Srouji informed the CEO himself. That means he was not planning a sudden surprise quit like when you come to consider your employer an enemy. So obviously he wanted a friendly parting, which means he would not go immediately but agree on a transition period. And keep it secret, again to be friendly to Apple, protect the stock price and so on.

That transition period may now be on, for what we know. After all, he said he won't quit "anytime soon", which is a good choice when you know you are preparing your exit but are not allowed to admit it. Of course he would not admit it.

 

[name99]

Having iPad jump all the way from A16 to A19 and the Studio Display also getting A19 makes me think there will be a lot of stuff that ultimately settles on A19 as the last N3 family SoC before the price jumps with N2 (which is more of a problem at base iPad / ATV pricing than it is for iPhones - or low end Macbooks for that matter)

Wouldn't make sense to use A18 for ATV unless they have some stockpile to work through - if so they'd probably not release it until they stop selling iPhone 16 & 16e. Which could be next spring, if iPhone 16 and 16e leave the price list together and are replaced by iPhone 17e. Since there's no set time on the calendar people are expecting 17e they could run through their entire stock of A18s prior to its intro, meaning they may not need a product to soak up non-Pro A18s and can coalesce everything on A19.

Another reason for an Apple TV upgrade would be to replace Broadcom wifi/BT with Apple N1 to reduce BOM.

The reason for A19 might ultimately be related to AI.
If Apple plans to thread AI throughout the UI, then getting as many devices as possible placed able to handle the UI elements might be worth the dollar or two savings possible if they instead used older chips.

I can't say what they have in mind for the UI (and even they might not be absolutely sure); they just might be certain that, like having a powerful GPU was an essential part of what they planned for every UI after the iPhone launch, so having powerful AI potential is an essential part of what they plan for every UI going forward.

For the same reason I expect new S-tier chips for the Watches this year. Of course watches can delegate tasks to the phone, but they seem to want the watch to be somewhat functional standalone (or for power savings).

 

[name99]

Yes. For further proof of their margin space, I invite people to look at iPad Air. Comes with M3, not A series, and list price is $599 at 8/128. That SoC costs >50% more than A series (based on area) and the display costs more than a Macbook's (higher resolution, touch layer) and it uses the more expensive Broadcom wifi/BT.

They could easily sell a Macbook using A series and N1 at $599. Though maybe listing it for $699 and leaving plenty of room for promos down as low as $550 like you say and edu pricing perhaps as low as $499 for once a year back to school type deals. College students would eat that up, and after four years using a Mac when they get out in the workforce they'll want one. That's probably the reason more and more corporate IT is supporting Macbook as an option.

I'd agree with this.

There's so much scope for Apple to eat into the "Don't care about Windows, but want a cheap laptop" market. Windows seems to have very few people left who actually like it; the best I see on the internet is a kind of tolerance of the "Well you need it for games" variety.

Windows on ARM also seems to be flailing. Superficially Qualcomm's new Elite Extreme Ultra Pro Max Titanium Founder's Edition SoC looks good (the high end is superficially between a Pro and a Max; the low end between an M and a Pro) but look at the details and you see that it's a product *designed* by marketing, albeit put together by some very good engineers. The CPUs are fine, but everything that doesn't appear in standard ad copy has been ditched. So it has inferior memory bandwidth compared to Apple, and it probably has a substantially inferior NoC (inferior QoS, much less limited network support for handling generic networking details [compression and crypto, packet construction/validation, that sort of thing]), and what looks like a problematic new security architecture [based on a supposed always-present cellular connection because, yes, Qualcomm...]. But more important than all that, it appears, based on multiple years now of experience, that QC and MS have a, uh, tense, relationship, probably because MS has contempt for QC's drivers and QC has contempt for MS' driver architecture. So the two STILL seem incapable of working together productively to ship a fully reliable Windows box that does everything a Windows user expects.

Which is all a way of saying that low end consumer Windows is vulnerable (on x86 or ARM) in a way that it has never been before. Now LOWEST end Windows -- you can still get terrible machines at Best Buy for $250, and lousy machines for $400, but I expect that in two or three years even that will dry up as people get wise that they should buy a 3-yr old MacBook Air on eBay over a new Windows box.

 

[Jan Olšan]

"Apple Mac True Power Consumption: power meter vs powermetrics"

The youtuber says the "telemetric" data that most people trust to show Apple Silicon power consumption, can be understating the actual power consumption significantly. Or at least in his results...

 

[Doug S]

There is also a small, growing list of games that either require an something newer than the A15 or more RAM.

Rant: more developers need to take the time to make apps and games compatible with the Apple TV. For some games like Final Fantasy, it blows my mind that there isn’t Apple TV support.

Anyone ever seen any figures or even hints what Apple TV's installed base is? I love mine but I've never once thought "man I would like to game on this" because it isn't designed for that and would need some sort of controller since the remote isn't exactly useful beyond a narrow category of games.

If the market isn't very large, and it would require the purchase of a 3rd party gaming controller which pushes you into an even smaller niche, I can see why developers ignore it. If Apple wanted gaming to be a thing they'd promote it for that, incentivize developers, offer a controller for it etc. It would still be a drop in ocean compared to the revenue they make from mobile gaming - and would suffer in comparisons with real game consoles.

 

[Doug S]

The initial report said that Srouji informed the CEO himself. That means he was not planning a sudden surprise quit like when you come to consider your employer an enemy. So obviously he wanted a friendly parting, which means he would not go immediately but agree on a transition period. And keep it secret, again to be friendly to Apple, protect the stock price and so on.

That transition period may now be on, for what we know. After all, he said he won't quit "anytime soon", which is a good choice when you know you are preparing your exit but are not allowed to admit it. Of course he would not admit it.

He's in his early 60s. He may not be thinking in terms of going to another company or starting his own, but planning retirement.

 

[jdubs03]

"Apple Mac True Power Consumption: power meter vs powermetrics"

The youtuber says the "telemetric" data that most people trust to show Apple Silicon power consumption, can be understating the actual power consumption significantly. Or at least in his results...

I’m pretty sure Notebookcheck, and ComputerBase run their power draw test via a meter. If I’m wrong someone correct me.

 

[Doug S]

If Apple was fudging on power draw by 3x that would be reflected in poor battery life on Macbooks.

I've also seen others do power tests measuring at the wall and it still does well. Either there's something borked in the equipment he's measuring with, or the Mac Studio he's testing is burning a lot of excess power for unknown reasons.

 

[coercitiv]

If Apple was fudging on power draw by 3x that would be reflected in poor battery life on Macbooks.

It wouldn't be 3x though, more like 2x under specific conditions that we can consider as outliers for normal usage.

Apple themselves quote 145W for the 2025 Mac Studio with M4 Max and 270W for the M3 Ultra version of the same 2025 device (wall measured). From that we can see the unit is definitely built to handle ~300W power usage, the question that remains is whether such a value can be obtained on the M4 Max.

Spoiler: Mac Studio 2025 power consumption and thermal rating

IMHO it comes down to engineering choices that Apple made for this form factor: they may have chosen to rely on thermals and max clocks to arrive at the maximum power draw allowed on a device. Max clocks would likely limit peaks, thermals would likely limit the average value. Given robust power delivery and cooling, this would be an elegant way of maximizing performance. They probably have some current limits enforced too.

We can see in the video how the max power draw is obtained during a test that puts the device near thermal throttling level with the fans running at max speed. This aligns with the speculation above.

Spoiler: Max load thermals and fan speed

All this being said, even if the findings in the video are true (which I'm not arguing they are), they would still not change what we know about overall Apple chip performance and efficiency. They would complicate direct comparisons for devices with active cooling though.

 

[Cardyak]

For the same reason I expect new S-tier chips for the Watches this year. Of course watches can delegate tasks to the phone, but they seem to want the watch to be somewhat functional standalone (or for power savings).

Seems likely regardless. Apple has been upgrading the S chips in the Watch every 3 years now for a while. They used the little core present in the A13 (Thunder) for 3 years in the S6, S7, S8. Then they moved to the little core in the A16 (Sawtooth) for 3 years in the S9 and S10 (S10 was used for 2 years running).

Given that the Apple Watch usually features last years small core, I expect the 2026 release will use the little core found in the A19.

 

[eek2121]

Anyone ever seen any figures or even hints what Apple TV's installed base is? I love mine but I've never once thought "man I would like to game on this" because it isn't designed for that and would need some sort of controller since the remote isn't exactly useful beyond a narrow category of games.

If the market isn't very large, and it would require the purchase of a 3rd party gaming controller which pushes you into an even smaller niche, I can see why developers ignore it. If Apple wanted gaming to be a thing they'd promote it for that, incentivize developers, offer a controller for it etc. It would still be a drop in ocean compared to the revenue they make from mobile gaming - and would suffer in comparisons with real game consoles.

Large enough that they periodically do a refresh.

We own 3, and my kids play games on one of them. Quite a few people I know own one.

"Apple Mac True Power Consumption: power meter vs powermetrics"

The youtuber says the "telemetric" data that most people trust to show Apple Silicon power consumption, can be understating the actual power consumption significantly. Or at least in his results...

If Apple was fudging on power draw by 3x that would be reflected in poor battery life on Macbooks.

I've also seen others do power tests measuring at the wall and it still does well. Either there's something borked in the equipment he's measuring with, or the Mac Studio he's testing is burning a lot of excess power for unknown reasons.

Looks accurate to me. Another issue is a lot of reviewers don’t measure power under a worst case scenario. Instead they will do things like simulate web browsing, watch a video, etc. These are areas where the M-series excels. However, when the chip is fully loaded, efficiency tends to fall.

When I was working, I had a Macbook with an M2 in it. I was never able to get more than 3-6 hours of battery life out of it because of all the stuff running used a ton of CPU, RAM, and IO. Sometimes I’d work on my Zen 3 based laptop, and I noticed I could work longer on the PC., usually close to 6 hours.

While these can’t be compared directly due to possible battery size differences, OS differences, compiler differences, etc. The main takeaway I am trying to get at is that the efficiency is overhyped. They excel at light workloads, but not so much at heavy workloads.

Try running a Blender render to measure battery or some other torture test and measure how long it takes to kill the battery, and you will see what I mean.

Don’t get me wrong, the chips are great.