M3 Chip Generation - Discussion Megathread

[senttoschool]

The fact that you think ray tracing exists as "pure unnecessary eye candy for triple-A games" says volumes about how little you know of where Apple is going.

Ray tracing is an essential element of a variety of graphics algorithms that are relevant to AR, for example casting realistic shadows... I suspect that is vastly more important in Apple's calculations than game eye-candy, which is just a minor side benefit.

Makes me think if Apple will do a quick written press release saying the M2 has been switched to the M3 for Vision Pro before it gets released.

Obviously, it didn’t make sense to announce M3 at WWDC but at the same time, it doesn’t make much sense to ship Vision Pro with M2 if M3 is so near.

 

[jeanlain]

The fact that you think ray tracing exists as "pure unnecessary eye candy for triple-A games" says volumes about how little you know of where Apple is going.

IMO, Apple is going in the same direction that they've been going for years WRT iPhone GPUs. It's marketing and hype. Note, I'm only talking about iPhone GPUs, not Mac GPUs.

Who currently needs RT hardware on a phone? Maybe 10 years from now, the RT unit will be used in many games as cumbersome rasterization techniques (which are much more power-efficient and give almost identical results) will be abandoned. But now it's only a marketing gimmick.

Apple doesn't care that almost no game and no iOS app uses the GPU of an iPhone at its fullest. I contend that all the "console quality" games they showcase at iPhone events (not just the latest one) essentially serve to demonstrate the technical superiority of the iPhone vs an android phone.
It's like Geekbench results really. Those who spend hundreds in the latest smartphone just want the most powerful one. And what better demonstration than the ability to run a current console game? For me, that's all there is. Whether users will actually purchase these games doesn't matter much to Apple.

 

[clam zero]

IMO, Apple is going in the same direction that they've been going for years WRT iPhone GPUs. It's marketing and hype. Note, I'm only talking about iPhone GPUs, not Mac GPUs.

Who currently needs RT hardware on a phone? Maybe 10 years from now, the RT unit will be used in many games as cumbersome rasterization techniques (which are much more power-efficient and give almost identical results) will be abandoned. But now it's only a marketing gimmick.

Personally I wasn't dissatisfied with the way they presented it. They're rolling out a new GPU a year late. The raytracing capabilities haven't been tested in any real applications, and may not ultimately be good enough or practical enough for any games that are likely to emerge in the near future; they exist primarily to provide commonality with upcoming A-series GPUs (on which these coprossessors can at least be used for non-real-time image rendering), serve hypothetical XR-related needs, and provide a platform which Apple can iterate on. So how should they talk about it? It would be weird to just mention it in passing. Better for them to show a single proof-of-concept, lingering on the topic for just long enough to dissuade suspicion, before moving on to the next bit. Best-case scenario, developers like the new hardware and it enables new capabilities for games and apps for many years to come.

So ultimately, there will probably come a day when Apple markets its devices as full-fledged gaming powerhouses, suitable for gamers who want the best visual fidelity of any platform. I hope that day isn't too far away. But for the time being they seem to be moving ahead with a healthy amount of caution.

 

[clam zero]

Makes me think if Apple will do a quick written press release saying the M2 has been switched to the M3 for Vision Pro before it gets released.

Obviously, it didn’t make sense to announce M3 at WWDC but at the same time, it doesn’t make much sense to ship Vision Pro with M2 if M3 is so near.

Personally I doubt the first generation of Apple RT hardware will be capable enough to give the Vision Pro any crazy new capabilities (especially considering its battery life is already only 2 hours). Better to get the thing out the door ASAP and see how people like it. Establishing consumer interest and refining the UX are more important than offering theoretical support for a class of applications which doesn't exist yet.

 

[T'hain Esh Kelch]

Makes me think if Apple will do a quick written press release saying the M2 has been switched to the M3 for Vision Pro before it gets released.

Obviously, it didn’t make sense to announce M3 at WWDC but at the same time, it doesn’t make much sense to ship Vision Pro with M2 if M3 is so near.

It would make no sense to ship it with an M3. Developers have been working with the M2, and Apple certainly wouldn't want to ship a model that developers haven't been working on. Software crashing left and right? No chance of happening.

Also, Apple just needs this out so people can experience it. The M3 would make absolutely no difference. And they are famous for making rev. B products are large leap over the rev. A models anyway, so the rev B. will likely ship with an M4, with improved ray tracing.

 

[Confused-User]

It would make no sense to ship it with an M3. Developers have been working with the M2, and Apple certainly wouldn't want to ship a model that developers haven't been working on. Software crashing left and right? No chance of happening.

That's a joke, right? I mean I think it is, except it's not actually funny.

If they can't ship an M3 that runs *every single app* built for the M2 without crashing, they might as well pick up their marbles and go home. But there is zero chance of that happening. Full backwards compatibility is just table stakes for being in the SoC game.

You know what can cause developers major headaches? And certainly will, likely multiple times? OS changes and library changes. That already happens yearly for MacOS and iOS, though fortunately mostly only for devs that haven't been paying attention. It's a very safe bet that Apple will have to redo at least a few things here and there before the 1.0 release, and equally safe that they'll redo a bunch more over the next few years. That's life in the fast lane on a new platform, and no developer with the faintest idea what they're doing would expect otherwise.

Also, Apple just needs this out so people can experience it. The M3 would make absolutely no difference. And they are famous for making rev. B products are large leap over the rev. A models anyway, so the rev B. will likely ship with an M4, with improved ray tracing.

The M3 would make quite a large difference! The Vision is pushing the envelope in a lot of ways. Having a faster (and more efficient) SoC would be a significant benefit.

As I wrote right after it was announced, I expect that the Vision Pro will ship with the M3. However I won't be surprised if I'm wrong - it depends on whether the M3 ships early enough. And Apple can occasionally be perverse.

Edit to add: Your contention also shows ignorance of recent history. Apple shipped the dev kit for Apple Silicon Macs with an A12Z chip, but shipped the first AS Macs with the M1.

 

[PgR7]

It would make no sense to ship it with an M3. Developers have been working with the M2, and Apple certainly wouldn't want to ship a model that developers haven't been working on. Software crashing left and right? No chance of happening.

Also, Apple just needs this out so people can experience it. The M3 would make absolutely no difference. And they are famous for making rev. B products are large leap over the rev. A models anyway, so the rev B. will likely ship with an M4, with improved ray tracing.

Im curious, are you a software developer?

 

[dmccloud]

It would make no sense to ship it with an M3. Developers have been working with the M2, and Apple certainly wouldn't want to ship a model that developers haven't been working on. Software crashing left and right? No chance of happening.

Also, Apple just needs this out so people can experience it. The M3 would make absolutely no difference. And they are famous for making rev. B products are large leap over the rev. A models anyway, so the rev B. will likely ship with an M4, with improved ray tracing.

The Apple Developer Transition Kit that was available prior to the M1 going on sale was a modified Mac Mini running an A12z SoC. Because the code base was the same as the M1 and later M2 used, you didn't suddenly see a lot of crashes when M1 Macs hit the market - in fact, the opposite occurred because the M1 Macs were ready for primetime while the DTK was a cobbled-together hodgepodge of parts. On the PC side, you don't see software crashing left and right when AMD and Intel release new processor generations, because despite any changes to the socket, architecture, RAM support, etc. the code base remains unchanged.

The M3 would make quite a large difference! The Vision is pushing the envelope in a lot of ways. Having a faster (and more efficient) SoC would be a significant benefit.

As I wrote right after it was announced, I expect that the Vision Pro will ship with the M3. However I won't be surprised if I'm wrong - it depends on whether the M3 ships early enough. And Apple can occasionally be perverse.

Edit to add: Your contention also shows ignorance of recent history. Apple shipped the dev kit for Apple Silicon Macs with an A12Z chip, but shipped the first AS Macs with the M1.

Most likely any adoption of the M3 into Vision Pro will come later, when 3nm production has sufficient yields to provide enough silicon for A17 and M3 (along with its variants) to alleviate channel shortages like we're already seeing with the 15 Pro and Pro Max. I can't see Vision Pro getting M3 or some other 3nm based silicon before the Mac lineup does. With the release of the iPhone 15, there's quite a bit of 5nm production capacity that has freed up, which can be used for Vision Pro.

 

[name99]

The Apple Developer Transition Kit that was available prior to the M1 going on sale was a modified Mac Mini running an A12z SoC. Because the code base was the same as the M1 and later M2 used, you didn't suddenly see a lot of crashes when M1 Macs hit the market - in fact, the opposite occurred because the M1 Macs were ready for primetime while the DTK was a cobbled-together hodgepodge of parts. On the PC side, you don't see software crashing left and right when AMD and Intel release new processor generations, because despite any changes to the socket, architecture, RAM support, etc. the code base remains unchanged.




Most likely any adoption of the M3 into Vision Pro will come later, when 3nm production has sufficient yields to provide enough silicon for A17 and M3 (along with its variants) to alleviate channel shortages like we're already seeing with the 15 Pro and Pro Max. I can't see Vision Pro getting M3 or some other 3nm based silicon before the Mac lineup does. With the release of the iPhone 15, there's quite a bit of 5nm production capacity that has freed up, which can be used for Vision Pro.

Apple will happy to sell as many Vision Pro's in a year as iPhones in a week (ie about a million)!
Meaning this is not a serious argument for not using an M3.

There may be good arguments in terms of risk, but IMHO the "mac developer kit" argument is the correct analogy – the Vision Pro's being tested and demo'd today are NOT the shipping version, but something close enough to at least let developers get a feel, while also learning of all the most serious problems that need to be fixed ASAP.

 

[Confused-User]

Apple will happy to sell as many Vision Pro's in a year as iPhones in a week (ie about a million)!
Meaning this is not a serious argument for not using an M3.

Your phone number is off by a factor of maybe 4.5 (they sold over 225 million last year)! But that just makes your argument more correct.

On the other hand I keep hearing their AVP target is less than half that (400k). No idea if that number's true. Of course that would also strengthen your argument.

I think he was comparing against Macs, though. Still, while their volume is much lower (11.5% of phones?), relative silicon area difference is somewhat less lower (Mx chips are all variously larger than Ax), and the difference compared to the AVP is still huge.

I think the proper comparison is against all N3 usage, though, which makes the argument completely ridiculous. Even if you think there's a shortage of N3 chips, which seems not to be the case (and which certainly won't be by the start of next year).

 

[senttoschool]

It would make no sense to ship it with an M3. Developers have been working with the M2, and Apple certainly wouldn't want to ship a model that developers haven't been working on. Software crashing left and right? No chance of happening.

Also, Apple just needs this out so people can experience it. The M3 would make absolutely no difference. And they are famous for making rev. B products are large leap over the rev. A models anyway, so the rev B. will likely ship with an M4, with improved ray tracing.

Is it like when all the iOS apps broke when A17 Pro was released? A16? A15? A14? And so on?

You realize Apple uses standard APIs right? It’s not the job of the app to know the underneath hardware unless the app is specifically targeting a brand new hardware feature.

My speculation comes from the fact that the Vision Pro is a $3500 device so Apple can afford to use the latest and greatest. Apple would like to put its best foot forward for the VP. Furthermore, the VP seems like one device that could use as much horsepower and efficiency as possible.

 

[dmccloud]

Is it like when all the iOS apps broke when A17 Pro was released? A16? A15? A14? And so on?

You realize Apple uses standard APIs right? It’s not the job of the app to know the underneath hardware unless the app is specifically targeting a brand new hardware feature.

My speculation comes from the fact that the Vision Pro is a $3500 device so Apple can afford to use the latest and greatest. Apple would like to put its best foot forward for the VP. Furthermore, the VP seems like one device that could use as much horsepower and efficiency as possible.

Two big issues at play there. First and foremost is the limited production of 3nm silicon in comparison to the 5nm silicon used in every current product NOT named iPhone 15 Pro/Pro Max. Apple is not going to cannibalize an already limited supply of 3nm parts in order to bump the specs for Vision Pro, especially when it's likely to sell considerably fewer units given the ~$3000 price point. The second issue is the fact that it's not the M2 which handles all the input from the cameras, sensors, microphones, etc. That is the job of the R1 chip:

A unique dual‑chip design enables the spatial experiences on Apple Vision Pro. The powerful M2 chip simultaneously runs visionOS, executes advanced computer vision algorithms, and delivers stunning graphics, all with incredible efficiency. And the brand-new R1 chip is specifically dedicated to process input from the cameras, sensors, and microphones, streaming images to the displays within 12 milliseconds — for a virtually lag-free, real-time view of the world.

Focusing on M2 vs. M3 actually overlooks what is doing the bulk of the work inside Vision Pro, and I'd bet the farm that Apple isn't about to replace R1 just a few months before VP actually launches.

 

[Retskrad]

The A17 draws 14W (!!!!) every time you open an app, browse the web or do other tasks that need burst of power. I am sorry but this is inexcusable. The Apple Silicon team, before losing most of their best chip designers after the release of the M1, would never release a chip like this. Apple's chip team is now undisciplined and is no longer special compared to other teams at competing teams in the chip industry. They are following the AMD and Intel playbook by drastically increasing wattage to gain small performance improvements and no IPC gains.

When there's smoke, there's fire. I don't see how anyone can deny at this point that Apple has lost most of their best chip designers.

 

[senttoschool]

Two big issues at play there. First and foremost is the limited production of 3nm silicon in comparison to the 5nm silicon used in every current product NOT named iPhone 15 Pro/Pro Max. Apple is not going to cannibalize an already limited supply of 3nm parts in order to bump the specs for Vision Pro, especially when it's likely to sell considerably fewer units given the ~$3000 price point. The second issue is the fact that it's not the M2 which handles all the input from the cameras, sensors, microphones, etc. That is the job of the R1 chip:

Apple will use N3E for M3. N3E is the primary N3 node and reportedly has much better yield than N3B. N3B seems like a special node setup for Apple to meet A17 deadline.

The Vision Pro has ~2 hours of battery life. It can use the efficiency everywhere.

Doesn't make much sense for a $1,100 Macbook Air to get M3 supply over Vision Pro. If Apple is only make 400k - 1 million VP in 2024, the supply won't make any difference. Apple sells 4-8 million Macs per quarter.

 

[Xiao_Xi]

A17 draws 14W (!!!!) every time you open an app, browse the web or do other tasks that need burst of power.

Do you have a link for it? I find it very suspicious because Geekerwan has reported that the A17 consumes 14W doing Geekbench multi-core.

 

[leman]

The A17 draws 14W (!!!!) every time you open an app, browse the web or do other tasks that need burst of power.

This is absolutely untrue. Where did you get this ridiculous notion?

Frankly, I start suspecting that you are not arguing in good faith. You come here every few pages, drop a post along the lines "Apple lost all designers, they are doomed", ignore the rest of the discourse and disappear again until you re back to repeat the same statement but in a different wrapper.

 

[APCX]

The A17 draws 14W (!!!!) every time you open an app, browse the web or do other tasks

What? Where is the proof of this?

I am sorry but this is inexcusable.

Would it be more excusable if it was untrue?

The Apple Silicon team, before losing most of their best chip designers

How do you know who their best chip designers are? What insight do you have?

after the release of the M1, would never release a chip like this.

Their best designers wouldn’t release a chip that is significantly faster while retaining efficiency? Really? They don’t sound like good designers then.
.

Apple's chip team is now undisciplined and is no longer special compared to other teams at competing teams in the chip industry. They are following the AMD and Intel playbook by drastically increasing wattage to gain small performance improvements and no IPC gains.

Verifiably untrue.

When there's smoke, there's fire. I don't see how anyone can deny at this point that Apple has lost most of their best chip designers.

Where there is misinformation and baseless speculation there is hot air.

 

[PgR7]

The A17 draws 14W (!!!!) every time you open an app, browse the web or do other tasks that need burst of power. I am sorry but this is inexcusable. The Apple Silicon team, before losing most of their best chip designers after the release of the M1, would never release a chip like this. Apple's chip team is now undisciplined and is no longer special compared to other teams at competing teams in the chip industry. They are following the AMD and Intel playbook by drastically increasing wattage to gain small performance improvements and no IPC gains.

When there's smoke, there's fire. I don't see how anyone can deny at this point that Apple has lost most of their best chip designers.

Lots of people will tell you that are wrong, but without telling you why you are wrong. I dont see it that bad, its a good chip, but they just let it boost all it can, if they cap it at 8W you would say its a good chip. Its like the older chips had a ball under the throttle so it wont go full throttle and now they got rid of the ball, going halfway throttle is the most efficient chip.

 

[APCX]

Lots of people will tell you that are wrong, but without telling you why you are wrong. I dont see it that bad, its a good chip, but they just let it boost all it can, if they cap it at 8W you would say its a good chip. Its like the older chips had a ball under the throttle so it wont go full throttle and now they got rid of the ball, going halfway throttle is the most efficient chip.

People told this person multiple times why they are wrong, with evidence. Yet they keep repeating the same claims. Some people don’t like being told the truth.

 

[Retskrad]

From outside looking in, Apple critics, like Me, and people who like to defend Apple, are both guessing. Who knows what’s really going on inside Apple. I’m just putting the pieces together. When Apple does a good job, I praise them. When they stumble and are in a slump, I keep the same energy and say they are doing a poor job. No one has yet argued a compelling reason why the A17 peaks like a laptop/desktop chip. Under no circumstance should a phone chip, with a much smaller battery and thermal envelope, do that. If Qualcomm or Samsung did that, people would heavily criticize them for it.

 

[MRMSFC]

From outside looking in, Apple critics, like Me, and people who like to defend Apple, are both guessing. Who knows what’s really going on inside Apple. I’m just putting the pieces together. When Apple does a good job, I praise them. When they stumble and are in a slump, I keep the same energy and say they are doing a poor job. No one has yet argued a compelling reason why the A17 peaks like a laptop/desktop chip. Under no circumstance should a phone chip, with a much smaller battery and thermal envelope, do that. If Qualcomm or Samsung did that, people would heavily criticize them for it.

Translation:

“I don’t have a single source for what I’m saying”

 

[name99]

Your phone number is off by a factor of maybe 4.5 (they sold over 225 million last year)! But that just makes your argument more correct.

On the other hand I keep hearing their AVP target is less than half that (400k). No idea if that number's true. Of course that would also strengthen your argument.

I think he was comparing against Macs, though. Still, while their volume is much lower (11.5% of phones?), relative silicon area difference is somewhat less lower (Mx chips are all variously larger than Ax), and the difference compared to the AVP is still huge.

I think the proper comparison is against all N3 usage, though, which makes the argument completely ridiculous. Even if you think there's a shortage of N3 chips, which seems not to be the case (and which certainly won't be by the start of next year).

Oh, right stupid me. Watches are per week now. Phones are per business day.

 

[APCX]

From outside looking in, Apple critics, like Me, and people who like to defend Apple, are both guessing.

No. Many here are investigating and using data to come to some reasonable conclusions. You are repeating unfounded claims ad nauseum.

Who knows what’s really going on inside Apple.

Certainly not you, which is deadly for your argument because you are making claims about what’s going on inside Apple. Due to that, you need to be providing evidence. Most of us aren’t interested in Apple’s hiring process. We just want good chips and computers. Fortunately Apple is currently delivering that.

I’m just putting the pieces together.

Not necessarily in the right order though.

When Apple does a good job, I praise them. When they stumble and are in a slump, I keep the same energy and say they are doing a poor job. No one has yet argued a compelling reason why the A17 peaks like a laptop/desktop chip.

They haven’t because there isn’t much proof that it does.

Under no circumstance should a phone chip, with a much smaller battery and thermal envelope, do that.

Fortunately it doesn’t appear to.

If Qualcomm or Samsung did that, people would heavily criticize them for it.

 

[MayaUser]

From outside looking in, Apple critics, like Me, and people who like to defend Apple, are both guessing. Who knows what’s really going on inside Apple. I’m just putting the pieces together. When Apple does a good job, I praise them. When they stumble and are in a slump, I keep the same energy and say they are doing a poor job. No one has yet argued a compelling reason why the A17 peaks like a laptop/desktop chip. Under no circumstance should a phone chip, with a much smaller battery and thermal envelope, do that. If Qualcomm or Samsung did that, people would heavily criticize them for it.

So are you telling us that you dont know how A17 pro is working? Ok, till now, neither did i but i havent make silly assumptions and keep my heads down until the M3 family arrives and i have real world usage on my work in 3d modelling

 

[Pressure]

The A17 draws 14W (!!!!) every time you open an app, browse the web or do other tasks that need burst of power. I am sorry but this is inexcusable. The Apple Silicon team, before losing most of their best chip designers after the release of the M1, would never release a chip like this. Apple's chip team is now undisciplined and is no longer special compared to other teams at competing teams in the chip industry. They are following the AMD and Intel playbook by drastically increasing wattage to gain small performance improvements and no IPC gains.

When there's smoke, there's fire. I don't see how anyone can deny at this point that Apple has lost most of their best chip designers.

While technically true that Geekerwan managed to get the A17 Pro to draw 14 Watt under load it was also under certain conditions.

1. They increased the thermal envelope by using external cooling
2. The chip used this increased thermal headroom to boost clock speeds

Under normal circumstances the thermal headroom of the design wouldn't allow it and no one else has been able to replicate that under ordinary conditions.