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]

Dude literally caved to the market and added USB to the ipod in a year, where as our current apple is more hardlining people into buying usb hubs and extra power bricks. He never pushed for firewire or thunderbolt because they were more expensive, it was because they were much faster and better. And IMO thunderbolt failed and became USB because without steves pressure on his suppliers, nobody wanted to keep innovating.

Side note: macOS 26 Tahoe finally kills off FireWire. I wasn't even aware it was still supported, considering the last Mac with it built-in was in 2011 or 2012.

Ironically though, I still have FireWire drives full of clean installs of old OS X versions. Makes repurposing ancient equipment much easier.

Right. I know that Apple always had an internal Intel build of Mac OS X from the start of development of that OS - starting a decade before they switched, and long before they knew they would switch. It was insurance, and it was a way to gauge where they were.

I'm sure Apple did have Mac OS X running on A series for long before the M series design, but don't overlook my statement here: "shifting MacOS not only to ARM, not only to preferentially ARM, but to preferentially Apple Silicon"

Yeah, and the SoC in the pre-M series Mac dev box was even called an A12Z too.

 

[johnsonwax]

So you're saying we can all blame Apple for the immense delays and mobile failure of the G5?

No but you're revealing the point that others are trying to make here. Apple could design the ISA, but they couldn't manufacture the product, and that was the problem. They were dependent on Moto and IBM to do that, as well as dependent on those two parties finding a market to provide the necessary scale for PowerPC, and they largely failed on that front, leaving Apple to carry most of that load, particularly after the G3. I don't think anyone but Apple used G5, and almost nobody else used G4. Apple didn't have enough control over the product to make it work. Now, that doesn't mean they would have made it work, but Motorola's financial struggles at that time and IBMs shifting focus meant that their partners were increasingly not partnering.

No it isn't. Its a continuation of Apples motives to ship good products but the idea that they need to make their own chips isn't really the full picture. In fact the only reason they make their own chips stems out of the iphone, and their angle for maximum battery life. Intel was literally working with apple to make a chip for the iphone but ultimately they couldn't do it.

Source for that? Everything I've ever read here and been told by insiders isn't that Intel couldn't make the iPhone chip, it's that they decided that it was going to take too much focus off of their x86 desktop business and were skeptical Apple could turn it into a viable product that they abandoned it as a business decision.

And you saw a lot of this at that time. Apple believed in the product, but Intel didn't and wouldn't commit to it. Apple had to get their 5 year exclusive deal with AT&T for similar reasons - Apple has to sweeten that pot to a huge degree to get a major carrier on board. This was the preceding problem they had with their retail channel - CompUSA didn't believe it would become a large enough revenue driver to train their staff and provide a good retail experience. Only Apple was willing to invest in that and had to create their own stores to do it. Apple was selling their product in Sears prior to the Apple Store launch. The former was the largest retailer in the world, and now doesn't exist, the latter is now the 14th largest retailer in the US. Apple was proven right in their belief in the product and their commitment to that paid off.

I really like Apple as a company but I have to completely disagree with your thinking here. Apple silicon isn't a direct result of steve job's ideology, but rather a result of the companies he has/had to work with at the time, and the people he would hire in the company. Steve's legacy is really knowing timing. Knowing when to make the ipod and working with industry vendors for the tiny HDDs, knowing to delay the tablet and give it a custom OS, knowing not to jump on the phone too early and wait for touch screen tech. He knew what made a good product and he did his best to make it happen.

But that doesn't explain why Apple has gone so strongly vertical. All of those things are true, but Apple could simply have operated as an integrator, being a step ahead of the rest of the industry. Instead they pulled every scrap of IP in-house, from sensors, to languages, to retail, to silicon, to services and in some cases to assembly. When Apple decided to move from stamped metal frames to aluminum unibody, they didn't wait for the industry to catch up to that - they designed mills specifically for their need and hired DMG Mori and Fanuc to manufacture them and for a period of a couple of years there Apple was buying roughly half of all the CNC mills being manufactured on the planet - and they were largely built to spec. Apple didn't time the market, they created it. DMG Mori built a secret factory that ONLY made mills for Apple. When HP execs asked their engineers for a similar product, they came back and said it would take 2-3 years in order to secure enough mills to do it - Apple had locked up the entire industry.

Apple had the only reliable fingerprint tech in the consumer space, in part because they bought the company that made the sensors, but also because they designed silicon to be able to process the sensor fast enough to work. Apple understood that in time that would become standard and commonplace, but they forced the timetable forward by several years by engineering it into existence. And in the process, Apple held ALL of the IP. They owned the sensors that nobody else could buy, they designed the silicon that nobody else could buy, they designed the software that nobody else could buy. That wasn't just a 'we're going to time the market' it was them controlling the market entirely and that happened because they recognized that they needed to own those elements.

Apple silicon is a very logical iteration from its origins (iphone) but ultimately I think the apple we have today isn't being run at all on the vision steve jobs had. Like, if he were alive, I think he would abhor the idea of apple trying to sell peripherals and adapters. Dude literally caved to the market and added USB to the ipod in a year, where as our current apple is more hardlining people into buying usb hubs and extra power bricks. He never pushed for firewire or thunderbolt because they were more expensive, it was because they were much faster and better. And IMO thunderbolt failed and became USB because without steves pressure on his suppliers, nobody wanted to keep innovating.

What the f even are you talking about? Steve didn't 'cave' on USB. One year later is when Apple shipped both the iTunes Music Store and iTunes for Windows. Apple had Firewire as a generally standard feature, but PCs didn't. In order for iPod to be a viable product for Windows, it had to be USB. That's why they switched.

Apple has explained why they don't ship a power adapter with every iPhone - everyone already has one, if not 10. I've got a drawer full of lightning and USB to AC adapters. Most people I know do.

But part of the reason why iPod prices fell under Jobs is that Apple didn't lose overall revenue on those sales. For every dollar they dropped the price they made up a dollar in the sale of peripherals and adapters. Watch this video and tell us again with a straight face that Jobs abhorred selling peripherals. This was a real product that Apple sold, and which Jobs pitched.

Thunderbolt failed because PC users would rather have cheap stuff than move their platform forward. It's why the iPhone got PCIe before most PCs did and why most PCs had to label their USB ports because they were too cheap to just make all of them good ones.

 

[Eug]

No but you're revealing the point that others are trying to make here. Apple could design the ISA, but they couldn't manufacture the product, and that was the problem. They were dependent on Moto and IBM to do that, as well as dependent on those two parties finding a market to provide the necessary scale for PowerPC, and they largely failed on that front, leaving Apple to carry most of that load, particularly after the G3. I don't think anyone but Apple used G5, and almost nobody else used G4. Apple didn't have enough control over the product to make it work. Now, that doesn't mean they would have made it work, but Motorola's financial struggles at that time and IBMs shifting focus meant that their partners were increasingly not partnering.

I thought the G4 was popular in embedded, but just not in the speeds and performance that Apple wanted. Embedded wanted low power low to mid performance G4s for not too much money. Apple wanted mid to high performance G4s.

Anyhow, my Synology NAS from 2013 runs on the G4.

What the fuck even are you talking about? Steve didn't 'cave' on USB. One year later is when Apple shipped both the iTunes Music Store and iTunes for Windows. Apple had Firewire as a generally standard feature, but PCs didn't. In order for iPod to be a viable product for Windows, it had to be USB. That's why they switched.

My understanding from that time was that Jobs did not want the iPod on Windows at all. It was designed as a halo product for Apple. If you wanted an iPod, you had to get a Mac. However, they then quickly realized the money that could made if it was Windows compatible so then they built iTunes for Windows and iPods for Windows.

BTW, iTunes for Windows had the same interface as iTunes for Mac, complete with the same Aqua buttons, etc.

 

[Doug S]

So yeah, Apple could do that, but there would need to be significant benefits. Essentially, Apple is already doing that but maintaining ARM compatibility, probably because it benefits them to be able to run most open source software, and almost everything out there has an ARM branch being maintained. That's not something Apple can just replace, and there's no reason to believe those maintainers would add an Apple ISA branch quickly, which would almost certainly require buying Apple hardware/software to compile, at least for a while.

I don't think that matters. Most software doesn't have an "x86" and "arm" branch, it has a Windows branch, a Linux branch, and a macOS branch. If there might be a handful of ifdefs for x86 or ARM even that would be pretty rare unless it doing something pretty low level (or the programmer isn't that smart) The CPU ISA rarely makes a difference for garden variety user mode C or C++ source code, let alone higher level languages.

Plus most Mac users aren't downloading and compiling stuff from source anyway. They're getting binaries that were specifically compiled for macOS, and if Apple still used x86 or had invented their own ISA from scratch that would be the compiler target used for those binaries.

I just don't think Apple depends all that much on the software infrastructure of the rest of the world. Certainly not an extent that it would influence their decision whether stick with ARM or invent their own ISA. There just isn't any gain for Apple to invent their own ISA. They already have the ability to add custom instructions under their license, as AMX showed. Unless they decided some type of profoundly new ISA (not just another RISC, but something that's not RISC or CISC at all) was the future. And who knows maybe there were guys deep inside the spaceship campus putting the first concepts of that on a whiteboard this week and it'll be shipping in 2037 lol

Don't underestimate how much Apple uses a bunch of tiny hidden ARM cores to manage stuff behind the scenes. If they counted all of them the iPhone might have over a dozen cores in total. They benefit from using ARM designed M0 etc. cores in places where performance isn't a factor and it isn't worth their designers time to roll their own. If they went their own way I suppose they could continue to use little ARM cores but there probably are some benefits having everything from top to bottom standardized around one ISA (or family of ISAs since some of those little cores are still 32 bit)

 

[Doug S]

What the f even are you talking about? Steve didn't 'cave' on USB. One year later is when Apple shipped both the iTunes Music Store and iTunes for Windows. Apple had Firewire as a generally standard feature, but PCs didn't. In order for iPod to be a viable product for Windows, it had to be USB. That's why they switched.

Its even simpler than that. When the iPod was designed USB 1.1 was the fastest there was. 480 Mbit USB 2 wasn't officially standardized until 2000, too late to make the first iPod design. So while they had USB available to them on those early Macs it just wasn't fast enough for a device with a 10 GB drive in it. Syncing would have been painful at 12 Mbps!

Even if the iPod had shipped with USB from day one it wouldn't have worked that well for Windows PCs since they were only just getting USB ports. Any PC older than a year or two wouldn't have USB ports at all and it wasn't until a couple years after USB 2.0's standardization that any of them had USB 2.0 ports.

 

[mvprod123]

Wider cores?

 

[Eug]

iPhone 17 Pro: First Look at the New Vapor Chamber System



Unlike conventional graphite pads, vapor chambers use a phase-change mechanism: liquid evaporates under heat, moves across the plate, then condenses back into liquid—redistributing heat more efficiently. This system is designed to maintain stable performance under high workloads, including gaming, video rendering, and AI inference tasks.

The need for this system becomes especially clear when considering the new A19 Pro chip, which is expected to generate substantial heat not only from gaming or GPU-intensive tasks, but also from on-device execution of generative AI models. These workloads are highly demanding and require sustained thermal performance over time.

To support this, Apple may also be shifting toward lighter chassis materials, such as aluminum, which allow for better integration of the vapor chamber into the device frame. This enhances the system’s overall thermal efficiency, ensuring that excess heat is transferred effectively away from critical components and toward the external shell for dissipation.

---

On another note: I've been playing with my kid's iPhone 16e some more, and despite the widespread criticism that this is a crippled phone that isn't particularly cheap and that people should just buy the iPhone 15, I would definitely disagree with that.

Aside from the missing wide angle lens, I think the 16e is overall a superior iPhone to the 15, despite being cheaper than the 15. The other missing features on the 16e that the 15 has are Dynamic Island (meh) and MagSafe, but for MagSafe that is remedied by buying a $10 MagSafe case.

OTOH, 16e includes the AI-powered Photos Clean Up tool which works really well, and the 16e includes the Audio Mix feature for Spatial Audio. This works great in mimicking the clean sound from a separate third-party microphone, and none of the 15 series includes this feature.

I suspect these are probably just Apple software restrictions but I do wonder if some of could be related to the 8 GB RAM and A18 in the iPhone 16e. The iPhone 15 has 6 GB RAM and the A16 SoC which is two generations behind.

Also, the iPhone 16e also supports Visual Intelligence like other iPhone 16 models, but I wasn't able to test it yet since Apple blocks anything associated with ChatGPT on iCloud accounts for kids under 13.

I Tried the iPhone 16's New Visual Intelligence, and It Feels Like the Future

Apple's Visual Intelligence is only in beta, but it could say a lot about where phones are going.

www.cnet.com

 

[johnsonwax]

I thought the G4 was popular in embedded, but just not in the speeds and performance that Apple wanted. Embedded wanted low power low to mid performance G4s for not too much money. Apple wanted mid to high performance G4s.

G3 was the real star in embedded having a pretty good performance/power ratio and a number of variants, including a radiation hardened one for space applications.

Some of the companies that had already committed to PowerPC tended to go on and use the G4 if they needed general purpose compute, but the G4 didn't get as much traction outside of Apple. Gamecube was a G3 variant. There were no G4 consoles. My understanding is that the embedded market just didn't need the additional performance and cost which Apple was chasing upmarket to compete with Intel. Devices like the TiVo that started with a 4 series PPC CPU went to other architectures before the G4 arrived because they really just needed it to drive the interface, with ASICs handling the computational intensive codec work. Ultimately they mostly ran to ARM. Synology was in the market of selling NAS that you could treat as compact linux boxes (I have one as well - but runs a Marvell Kirkwood).

 

[Eug]

Before Apple released the M1 MacBook Air, I predicted Apple’s MacBook Air line could go with Ax chips. No need for the AxX/AxZ chips, which would later be rebranded as the Mx chips.

I was of course completely wrong back then but maybe it might finally happen in 2026 or 2027.

https://twitter.com/x/status/1939558413459177897

Ming-chi Kuo is predicting an A18 Pro MacBook around 13”.

 

[Eug]

Confirmed!!!111bbq

New MacBook With A18 Pro Chip Spotted in Apple Code

Apple is developing a MacBook with the A18 Pro chip, according to findings in backend code uncovered by MacRumors. Earlier today, Apple analyst...

www.macrumors.com

 

[Doug S]

Interesting, I guess that will give them a lower entry price on the laptop side - sort of the equivalent of the difference between the base iPad and the iPad Air. Based on that I guess this thing will be called just "Macbook".

 

[okoroezenwa]

If they can match the M1 Air’s price of 649 that’d be major. I’m doubtful though.

 

[Eug]

I wonder if an A18 Pro MacBook would get 16 GB RAM base. FWIW, all the iPhones with A18 (iPhone 16e, 16, 16 Plus) and A18 Pro (iPhone 16 Pro, 16 Pro Max) have 8 GB RAM.

A18 Pro MacBooks would probably work well for my kids when they go off to high school. If we bought those, that means my household would have macOS running on A series and iPadOS running on M series. However, I would only buy if it was 16 GB at a decent price.

 

[LightningZ71]

Is this maybe a new, "connected to the cellular network all the time" line of MacBooks? The A18 should be plenty potent enough to drive a MacBook, but should have excellent energy management, especially when using the cellular data link.

 

[Doug S]

If they can match the M1 Air’s price of 649 that’d be major. I’m doubtful though.

Apple has generous margins, so they have flexibility when it comes to their entry level products if they decide to use it. Look at the base iPad, it uses an iPhone SoC (A16 in this case but I'm sure it will be upgraded to A18 by the time this new Macbook comes out) and retails for $349. No way they are getting the same margins on that that they get on the iPad Air, let alone iPad Pro. They've chosen to sacrifice margins to price it like that.

Now think about the BOM of an iPad versus a Macbook. The display in the iPad is a little smaller (11") but still very high resolution, and it includes touch so a 13" display for a Mac probably costs less with the touch being left out. The iPad has a rear camera, you can drop that on the Mac. The iPad has to pack everything in a smaller space, that requires additional engineering effort that isn't free. The Mac on the other hand will require more RAM, more NAND, a few ports, and a keyboard. I think Apple could EASILY price this at $499 if they wanted to - because it would have a higher margin than the iPad.

Now whether they'd go that low depends on their goal. Do they just figure "we need a lower entry price since Mini doesn't interest the ever growing number of people who only buy laptops not desktops" or does their research indicate they could make big inroads in Mac market share by if they could beat the psychologically important $500 point.

But maybe it'll be $799 and be basically the same form factor as MBA but with a lower price reflected in the lower specs. That wouldn't be shocking but it would seem like mostly wasted effort - that it was done as a way to soak up excess "Pro" iPhone SoCs rather than trying to create a new market segment for the Mac.

 

[Doug S]

Is this maybe a new, "connected to the cellular network all the time" line of MacBooks? The A18 should be plenty potent enough to drive a MacBook, but should have excellent energy management, especially when using the cellular data link.

Cellular isn't free, even now that Apple has their modem. It isn't just the cost of the modem, but the patent licensing you have to pay. The future upcharge for cellular might not be as big as it is on the iPad since they'll lose the Qualcomm tax, but there will still be an upcharge. And I will still say it is stupid and pointless due to tethering, and every year that brings us more accessible and better/faster tethering makes building in cellular in a laptop more pointless.

 

[igor_kavinski]

It would make more sense to develop a really thin iPhone that you plug on the underside of a thin Macbook body and voila! A dual use device. Receive a call? Just use the iPods!

Apple, don't you dare steal this idea without giving me royalties!

 

[smalM]

Apple, don't you dare steal this idea without giving me royalties!

Isn't it cute he thinks he's the first one to come up with that concept?

 

[Eug]

Apple explored iPhones and iPads that dock inside larger devices like MacBooks | AppleInsider

A newly published patent reveals that Apple has explored iPhones and iPads that could be docked inside larger devices similar to a MacBook Pro in order to provide extra functionality when needed.

appleinsider.com

 

[poke01]

Apple explored iPhones and iPads that dock inside larger devices like MacBooks | AppleInsider

A newly published patent reveals that Apple has explored iPhones and iPads that could be docked inside larger devices similar to a MacBook Pro in order to provide extra functionality when needed.

appleinsider.com

View attachment 126453

Apple was 5 years early lol

 

[LightningZ71]

Cellular isn't free, even now that Apple has their modem. It isn't just the cost of the modem, but the patent licensing you have to pay. The future upcharge for cellular might not be as big as it is on the iPad since they'll lose the Qualcomm tax, but there will still be an upcharge. And I will still say it is stupid and pointless due to tethering, and every year that brings us more accessible and better/faster tethering makes building in cellular in a laptop more pointless.

I don't expect the service to be free. I also don't expect such a laptop to be cheap either. My point is that it could be a sort of "hypermobility" line that, just like the iPads that they sell that have an integrated cellular modem and service from the likes of AT&T and Verizon, can be always connected and ready to go, so long as you have coverage. No, this isn't some innovative or new segment, it's just new in the ARM Mac era. While the M series are exceptionally power efficient, they don't have integrated cellular modems to my knowledge. This would fill that niche.

 

[Eug]

Apple already built a cellular MacBook Pro a long time ago.



However, this was in the 3G era, and just a prototype. It never made it into production. (Some guy got a hold of a junked broken one, managed to repair it, and then tried to sell it on eBay, but then not surprisingly the listing got pulled shortly afterwards.)

 

[Mopetar]

If they go down that route it's either a cheaper MacBook that will have paltry specs for one or maybe two hundred dollars less than the M-series laptop or they'll make something new.

An 11" MacBook aimed at the college market for $500 would certainly shake things up a bit. Of course by the time you get more RAM and storage it will cost $700 or $800, but that's besides the point!

 

[Eug]

If they go down that route it's either a cheaper MacBook that will have paltry specs for one or maybe two hundred dollars less than the M-series laptop or they'll make something new.

An 11" MacBook aimed at the college market for $500 would certainly shake things up a bit. Of course by the time you get more RAM and storage it will cost $700 or $800, but that's besides the point!

The 12” MacBook fans have been wanting a new one since my 2017 model was discontinued back in 2019. With the current chassis and the reduced bezel size of the M4 MacBook Air, they could fit a 12.5” screen in there if it gets the notch. That would meld well with an A18 Pro plus 16/256 GB that Apple could sell at a pricey US$899, but which would be frequently discounted at third party retailers to $799.

 

[igor_kavinski]

Apple was 5 years early lol

Yes. RATS!