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

 

[Eug]

TSMC: 3nm Chips for Smartphones and HPCs Coming This Year

www.anandtech.com

PDF transcript of TSMC earnings call



I'm assuming M3 is N3E, meaning M3 chips trickle out 2023 Q4 but Macs won't be out until 2024.

 

[Tigerick]

TSMC: 3nm Chips for Smartphones and HPCs Coming This Year

www.anandtech.com

PDF transcript of TSMC earnings call

View attachment 83394

I'm assuming M3 is N3E, meaning M3 chips trickle out 2023 Q4 but Macs won't be out until 2024.

We should be expecting M3-based iMac and MBP13 in October based on Mark Gurman. At least the first iteration of M3 will be based on TSMC's N3 process

 

[Eug]

We should be expecting M3-based iMac and MBP13 in October based on Mark Gurman. At least the first iteration of M3 will be based on TSMC's N3 process

I’m not confident in Mark Gurman’s prediction this time around. I could very well be wrong, but I’m not convinced they will be using N3/N3B for M3.

 

[Tigerick]

I’m not confident in Mark Gurman’s prediction this time around. I could very well be wrong, but I’m not convinced they will be using N3/N3B for M3.

Yeah, release timings and yield might be issues, that's why I omit M3-based MacBook Air 13 & 15. If M3 is based on N3E, we might be expecting first release around one year later (might be earlier depends on yield), ie end of 2024.

That is the reason I think Apple might be focusing on iMac and MBP13 which are not high-volume Macs. I have created a thread @ Macrumor Forum to discuss potential higher price point on upcoming MacBook Pro 13 here. You are welcomed to pitch in

 

[Mopetar]

What makes you think otherwise? Gurman has an exceedingly good track record regarding his predictions, to the extent that he obviously has some good inside sources.

 

[Glo.]

If the MacBook Pro 13 is first to receive M3 chip, then that M3 series leak with 6P/6E, 36 GB of RAM and 18 core GPU appears to be for that laptop.

So the final configuration of M3 appears to be 6P/6E, 192 bit bus and 20 GPU cores.

 

[Doug S]

I'm assuming M3 is N3E, meaning M3 chips trickle out 2023 Q4 but Macs won't be out until 2024.

I'm not sure it matters whether they release M3 Macs with N3 at first or wait until N3E is available. They could get a bit more performance from N3E if they wait, but either way it is a bump over M2 and M3 goes in base level hardware for customers who are less performance sensitive. The Pro/Max/etc that always comes later would be N3E from day one and get that performance bump, and that's where buyers care more about performance.

The decision of whether to release M3 Macs this fall or wait until N3E is ready may come down to TSMC's capacity - with those yield issues they don't want to risk any shortage of A17s because of making M3s. Plus since N3E chips will cost less despite being slightly larger thanks to fewer EUV production steps - that may be enough incentive to wait a few months to launch as you suggest.

 

[Eug]

What makes you think otherwise? Gurman has an exceedingly good track record regarding his predictions, to the extent that he obviously has some good inside sources.

He's good, but I think "exceedingly good track record" might be overstating it a bit. However, even when he's right, sometimes he's quite a bit off in terms of the timing or other details like the specs.

ie. I'm absolutely sure new Macs are coming, just like he says. It's just a matter of when.

For example:

2022: New Mac Pro and 7K Display coming later this year

The Mac Pro didn't come until 2023, and the 7K display hasn't happened (yet).

I'm not sure it matters whether they release M3 Macs with N3 at first or wait until N3E is available. They could get a bit more performance from N3E if they wait, but either way it is a bump over M2 and M3 goes in base level hardware for customers who are less performance sensitive. The Pro/Max/etc that always comes later would be N3E from day one and get that performance bump, and that's where buyers care more about performance.

The decision of whether to release M3 Macs this fall or wait until N3E is ready may come down to TSMC's capacity - with those yield issues they don't want to risk any shortage of A17s because of making M3s. Plus since N3E chips will cost less despite being slightly larger thanks to fewer EUV production steps - that may be enough incentive to wait a few months to launch as you suggest.

Rumour has it N3B yields are an issue and cost is high. Plus everyone here seems to consider it essentially a temporary node, with some people claiming A17 will be BOTH N3B this year and then N3E next year. Preparing for this scenario, Apple may have thought it would just make more sense to design M3 for N3E, reserving all the N3B capacity for the much, much higher volume A17.

And it's not as if there's any rush for M3. M1 came out in late 2020, and M2 came out in mid 2022, so it makes sense if M3 comes out in early 2024.

 

[smalM]

"N3 is already in production with good yield."
The yield is not as good as for N5 at same time of ramp up – otherwise C.C. Wei would have mentioned it.

"Rumour has it N3B yields are an issue and cost is high."
A rumour. We don't really know how good or bad the yield is and we do not know the price difference between N3 and N3E.

"Plus everyone here seems to consider it essentially a temporary node, with some people claiming A17 will be BOTH N3B this year and then N3E next year."
Both 20SOC and N10 were temporary nodes, but A8 never made it to 16FF and A11 never made it to N7.

"And it's not as if there's any rush for M3. M1 came out in late 2020, and M2 came out in mid 2022, so it makes sense if M3 comes out in early 2024."
I wouldn't consider the time from 2020 to 2022 to be exemplary for what Apple will do with ASi in the future.

 

[Eug]

"N3 is already in production with good yield."
The yield is not as good as for N5 at same time of ramp up – otherwise C.C. Wei would have mentioned it.

"Rumour has it N3B yields are an issue and cost is high."
A rumour. We don't really know how good or bad the yield is and we do not know the price difference between N3 and N3E.

True but who else will use N3B? (Not Intel.)

"Plus everyone here seems to consider it essentially a temporary node, with some people claiming A17 will be BOTH N3B this year and then N3E next year."
Both 20SOC and N10 were temporary nodes, but A8 never made it to 16FF and A11 never made it to N7.

Fair point, but Macs don't use A17 either.

Also, A9 was on both TSMC 16FF and Samsung 14FF at the same time, and A5 migrated from 45 nm to 32 nm.

"And it's not as if there's any rush for M3. M1 came out in late 2020, and M2 came out in mid 2022, so it makes sense if M3 comes out in early 2024."
I wouldn't consider the time from 2020 to 2022 to be exemplary for what Apple will do with ASi in the future.

Perhaps not but there's no rule the MacBook Air has to come out yearly either. These aren't iPhones.

 

[smalM]

True but who else will use N3B? (Not Intel.)

I think N3 is the process TSMC could make available in time for the A17.
I presume M3 is still an A17X design and will be on N3 too. I also think both will stay on N3.

Also, A9 was on both TSMC 16FF and Samsung 14FF at the same time

TSMC's 16FF failed and Apple wasn't sure TSMC would really be able to pull 16FF+ half a year in. A9 on 14LPE was their plan B.

and A5 migrated from 45 nm to 32 nm.

Eug, please....

Perhaps not but there's no rule the MacBook Air has to come out yearly either.

There is no rule.

 

[Doug S]

Perhaps a more interesting question than when M3 Macs will arrive and whether Apple will switch from N3B to N3E in the middle of the A17 / iPhone 15 cycle is what will A18 be made with? TSMC has said N3P will start HVM in "H2 2024" which will be too late for A18. If so, A18 must use N3E.

I suppose that would give them a small performance boost, given that N3E is a boost over N3B. Even if they switch to making A17s from N3E next year they would set its frequency based on N3B's characteristics (i.e. even though N3E A17s should be capable of clocking higher they wouldn't be allowed to, they'll just use less power than N3B A17s)

So A18 might be a bit underwhelming as IPC increases will have to provide most of the magic. Perhaps waiting until N3E for M3 will make sense - if they want to wait for A19/N2 for M4 to make it a more worthwhile jump. So maybe the "18 month gap thing" for Apple Silicon will be true through M4, though solely due to TSMC's process schedule snafus and not because that's how Apple wanted it.

The one unknown is Finflex. AFAIK N3B supports this as well as N3E, but the tools and the architectural knowledge on how best to use it would be a less mature state when Apple was designing A17. So even if Apple doesn't get much of a boost from use of N3E for A18 they might get it thanks to a smarter / more nuanced application of Finflex.

 

[eek2121]

Hoping for improved GPU performance TBH. CPU performance is “good enough” though improvements are always welcome. I have been on the fence when it comes to buying a Mac Mini. I have a Macbook Pro, but it is technically company owned. GPU performance trails my 3090 and 4090 by a pretty significant margin. I am not expecting *90 class performance, but at least 4070 class would be nice. Oh and Apple still hasn’t gotten good at drivers. Hopefully they will improve things now that Unreal Engine is supported.

Fun fact: Apple TV has a considerable amount of marketshare as a “gaming console” type device (did you know that the Apple TV games? My kids play games on it nonstop). I could see Apple releasing an “Apple TV Arcade” type device with better specs and a controller. They do appear to be paying more attention to gaming as of late after all, and it wouldn’t take much effort to take a step forward. An Apple TV with a base level M2 or M3 at $299 would be a compelling offer…

 

[Eug]

Fun fact: Apple TV has a considerable amount of marketshare as a “gaming console” type device (did you know that the Apple TV games? My kids play games on it nonstop). I could see Apple releasing an “Apple TV Arcade” type device with better specs and a controller. They do appear to be paying more attention to gaming as of late after all, and it wouldn’t take much effort to take a step forward. An Apple TV with a base level M2 or M3 at $299 would be a compelling offer…

Apple TV 4K 128 GB is my kids' gaming console. It's hamstrung a bit by the fact that Apple mandates all games be controllable by the Apple remote control, but otherwise it is fine for a 9 year-old as a gaming machine. Apple A15 is quite a capable SoC (Geekbench 6.1 SC 2250 MC5500), and the 4 GB RAM is sufficient for those games. I bought him an Xbox controller and he's happy for now.

What he really wanted was a Nintendo Switch, but there was no way I was going to buy one at this point considering it's 7 years-old. I'm waiting for the next Nintendo console before jumping in, so probably 2024.

However, I don't see Apple creating a separate Apple TV Arcade device anytime soon. They'll just continue upgrading their Apple TV across the board with A series upgrades. The latest A-series SoCs are already as fast as most people's main computers these days, if not faster. The hardcore gamers aren't going to be buying Apple TV for gaming, so there isn't really a need to cater to those people.

 

[Muadib]

I've had the Apple tv 4k since 2020, and just learned something new. Then again I use mine mostly for Zwift, which is sort of a game.

 

[eek2121]

Apple TV 4K 128 GB is my kids' gaming console. It's hamstrung a bit by the fact that Apple mandates all games be controllable by the Apple remote control, but otherwise it is fine for a 9 year-old as a gaming machine. Apple A15 is quite a capable SoC (Geekbench 6.1 SC 2250 MC5500), and the 4 GB RAM is sufficient for those games. I bought him an Xbox controller and he's happy for now.

What he really wanted was a Nintendo Switch, but there was no way I was going to buy one at this point considering it's 7 years-old. I'm waiting for the next Nintendo console before jumping in, so probably 2024.

However, I don't see Apple creating a separate Apple TV Arcade device anytime soon. They'll just continue upgrading their Apple TV across the board with A series upgrades. The latest A-series SoCs are already as fast as most people's main computers these days, if not faster. The hardcore gamers aren't going to be buying Apple TV for gaming, so there isn't really a need to cater to those people.

Hence why it makes sense to release a version with a controller. Those buttons on the Apple TV remote can map neatly to a controller layout. Note that you can pair a regular controller (such as an Xbox controller) with the Apple TV today.

Don’t discount future Apple gaming efforts. Apple would love very much to go for a large piece of the gaming pie. There is a reason (besides AI/ML) why they are putting such beefy GPUs in all of their chips. UE and a stable port of No Man’s Sky is likely the first of a series of announcements we will see. I am not saying for sure that Apple will release a gaming oriented Apple TV, but I wouldn’t bet against the idea.

Gaming, after all, is part of their subscription revenue strategy via Apple Arcade. Currently, I can play a game like Stardew Valley on everything from a phone to a TV to a Mac without purchasing a new license. Save games sync to iCloud and there is even an achievements frwmework.

The Vision Pro will also have games available for it at launch. Apple also is rumored to have a less expensive “non-pro” variant in house, which may come to market later.

I am a PC gamer and it would take a lot for me to consider using an iOS (or macOS) device for gaming, but Apple is definitely moving in that direction. Only time will tell how serious they decide to get.

 

[MadRat]

Honestly, its difficult to believe Apple hasn't absorbed Nintendo by now. Its the one vendor that could do it well considering Apple's existing (stable) ecosystem.

 

[Tigerick]

Hoping for improved GPU performance TBH. CPU performance is “good enough” though improvements are always welcome. I have been on the fence when it comes to buying a Mac Mini. I have a Macbook Pro, but it is technically company owned. GPU performance trails my 3090 and 4090 by a pretty significant margin. I am not expecting *90 class performance, but at least 4070 class would be nice. Oh and Apple still hasn’t gotten good at drivers. Hopefully they will improve things now that Unreal Engine is supported.

Fun fact: Apple TV has a considerable amount of marketshare as a “gaming console” type device (did you know that the Apple TV games? My kids play games on it nonstop). I could see Apple releasing an “Apple TV Arcade” type device with better specs and a controller. They do appear to be paying more attention to gaming as of late after all, and it wouldn’t take much effort to take a step forward. An Apple TV with a base level M2 or M3 at $299 would be a compelling offer…

Yep, that's what I believe what Apple will come out in the future. In fact, I have created a thread in MR forum dedicated to Apple TV Arcade edition...

One thing lots of people don't know is A15 is last A series SoC to use LPDDR4x memory. Guess what, M1 also is the only M series SoC to use LPDDR4x. So the next version is naturally M1 from BOM and technical points..we shall see

 

[Doug S]

Don’t discount future Apple gaming efforts. Apple would love very much to go for a large piece of the gaming pie.

They already have a huge piece of the gaming pie. The mobile gaming market is nearly DOUBLE the combined size of console and PC gaming, and the iOS app store has double the revenue of the Play Store. So if Apple is grabbing 2/3 of the mobile gaming market, and the mobile gaming market is double the size of the combined PC+console gaming market, then Apple alone is responsible for more gaming revenue than the entire PC and console gaming market!

 

[eek2121]

They already have a huge piece of the gaming pie. The mobile gaming market is nearly DOUBLE the combined size of console and PC gaming, and the iOS app store has double the revenue of the Play Store. So if Apple is grabbing 2/3 of the mobile gaming market, and the mobile gaming market is double the size of the combined PC+console gaming market, then Apple alone is responsible for more gaming revenue than the entire PC and console gaming market!

While I don’t doubt people are spending money on mobile games, i do doubt the accuracy of this chart. Most games on iOS are free with IAP, or a few dollars up front.

Less than 7 billion people worldwide have cell phones. That chart is basically claiming that on average each user spends more than $20 on gaming. That seems to me to be unrealistic.

In some parts of the world, people bring home and support themselves/families with less then that per month…

EDIT: Looking at Apple and Google earnings report now: this chart makes even less sense.

 

[Doug S]

While I don’t doubt people are spending money on mobile games, i do doubt the accuracy of this chart. Most games on iOS are free with IAP, or a few dollars up front.

Less than 7 billion people worldwide have cell phones. That chart is basically claiming that on average each user spends more than $20 on gaming. That seems to me to be unrealistic.

In some parts of the world, people bring home and support themselves/families with less then that per month…

EDIT: Looking at Apple and Google earnings report now: this chart makes even less sense.

You are forgetting in app purchases. The "freemium" model has taken over the app world, especially games. Make them free, and make the money charging people for weapons, better armor, skipping levels and so forth. Obviously most of those 7 billion are spending zero on games. But if 1/10th of them spend $200 on average...

Apple recorded $78 billion in services revenue last year, part of that is the 30% or 15% they are collecting on those in app purchases. That headline $136 billion is presumably total amount of in app gaming spend on both iOS and Android, including the bulk of the revenue going to the game vendors.

If you think that data is wrong, show someone other than IDC with numbers you believe in. You can't just say you don't believe IDC without offering some sort of alternative data, or some reason why you think the data they have collected is so far off what you believe. Do you doubt just their mobile revenue figures or do you think they are wrong on PC & console? Is it just that you don't believe mobile gaming dominates over PC & console? That's been true for years now, no one sane doubts it.

 

[darkswordsman17]

Honestly, its difficult to believe Apple hasn't absorbed Nintendo by now. Its the one vendor that could do it well considering Apple's existing (stable) ecosystem.

Too late for that, it'd be too costly and too risky, namely because Nintendo is notorious for wanting absolute control, which Apple almost certainly would not give them. I'm baffled Apple didn't buy them up back when the Wii U and 3DS had Nintendo freaking out (like 2013-2014 I think?). It would've been pretty cheap, but I think Nintendo has increased in value exponentially since then so it'd just be too high of a price and it'd be really difficult to integrate Nintendo in corporate structure sense. I think it was a pairing that made a ton of sense as far as what each could offer, as Apple could have managed the hardware, development tools, and the online system (3 things Nintendo notoriously struggles with). It instantly would've given Apple gaming cred. And it would have freed Nintendo up to focus on things they're good at and actually are wanting to do (games and weird quirky accessories for those games).

I'm actually surprised Apple hasn't basically ripped off some of Nintendo's ideas, like Joy-Con like controllers that attach to iPhones/iPads, and a dock for iPads that would turn it into a smart speaker/AppleTV/charge cradle. Basically make a Homepod that an iPad (or iPhone) magnetically connects to as a display (basically digital picture frame, clock, notification tab/FaceTime hub).

While I don’t doubt people are spending money on mobile games, i do doubt the accuracy of this chart. Most games on iOS are free with IAP, or a few dollars up front.

Less than 7 billion people worldwide have cell phones. That chart is basically claiming that on average each user spends more than $20 on gaming. That seems to me to be unrealistic.

In some parts of the world, people bring home and support themselves/families with less then that per month…

EDIT: Looking at Apple and Google earnings report now: this chart makes even less sense.

I wouldn't be surprised if the video game market makes for a somewhat accurate representation of relative wealth equality/inequality in the world. Supposedly on mobile, most games have like literally a handful of people spending bonkers amounts of money (thousands if not tens of thousands of dollars each) whilst most average users spend very little to almost nothing. And I have a hunch people in the West spend a lot more on games than those other people do, such that comparatively they'd be whales. Most people I know play games regularly, and I'd guess most spend hundreds if not a thousand dollars on it per year. One of the interesting changes to modern gaming is that people seem to focus on specific franchises if not specific games. The stuff that has come out due to the Microsoft acquisition of Activision has been eye opening. Sony makes like a billion dollars a year on Call of Duty (which costs them basically nothing to develop since they're not the developer or publisher) whereas their big name titles they pay to get made cost like a couple of hundreds of millions of dollars and while they do well enough it seems, not having to spend that plus making a billion per year is obviously preferable to Sony (which is why I think we've seen them open up to porting their games to other platforms as that's relatively cheap but opens up other avenues of revenue vs being limited to Playstation). Also in the report was that most of the players spending that money basically only play Call of Duty, so its no wonder Sony was worried about Microsoft potentially turning off that source.

 

[A///]

I want my damn m3 mac book pro already.

 

[Mopetar]

He's good, but I think "exceedingly good track record" might be overstating it a bit. However, even when he's right, sometimes he's quite a bit off in terms of the timing or other details like the specs.

I'd say anyone over 50% in the rumor business is quite good. Most are worse than a fair coin. One site that actually tracks correctness has Gurman at ~86.5% for last year. That's exceedingly good.

ie. I'm absolutely sure new Macs are coming, just like he says. It's just a matter of when.

For example:

One counter example proves he's not perfect. It doesn't do much to argue that he's not correct more than most.

Rumour has it N3B yields are an issue and cost is high. Plus everyone here seems to consider it essentially a temporary node, with some people claiming A17 will be BOTH N3B this year and then N3E next year.

Who's rumor and what's the record of everyone here for their beliefs. I suspect if anyone cared to actually calculate it, a coin flip would be more accurate. We're probably not as bad as MLID, but that's hardly saying much.

 

[Mopetar]

They already have a huge piece of the gaming pie. The mobile gaming market is nearly DOUBLE the combined size of console and PC gaming, and the iOS app store has double the revenue of the Play Store. So if Apple is grabbing 2/3 of the mobile gaming market, and the mobile gaming market is double the size of the combined PC+console gaming market, then Apple alone is responsible for more gaming revenue than the entire PC and console gaming market!

Microtransactions are clearly out of control. I doubt Apple gets as much of that as you might think though.