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

 

[igor_kavinski]

x86 will become nieche use case compared to ARM-based products.

Hearing this for how many decades now?

 

[Doug S]

At any rate, Apple’s mobile chips aren’t really hurting for more CPU power. I think their focus on the GPU is the right move for now. It isn’t like you can buy a smartphone that has something faster in it. None of the released smartphones come anywhere close to the iPhone in terms of performance.

For mobile I agree, while I noticed the performance increase when I upgraded from 11 Pro Max to 14 Pro Max last year it was harder to see than past upgrades - and that was a three year jump. Where Apple cares is for the Mac and lesser extent iPad Pro, and since they share the same cores they still have reason to keep pushing in a way they might not if those cores were only ever going to be used in a phone.

Whether it is better to expend more resources on the GPU (or NPU) than the CPU is another matter, and would depend on Apple's longer term goals. For instance, Vision Pro and its needs, and whatever other products there might be in the pipeline.

 

[Glo.]

Hearing this for how many decades now?

Thats the first time I have read.

"cough" In this thread "cough"

 

[Glo.]

For mobile I agree, while I noticed the performance increase when I upgraded from 11 Pro Max to 14 Pro Max last year it was harder to see than past upgrades - and that was a three year jump. Where Apple cares is for the Mac and lesser extent iPad Pro, and since they share the same cores they still have reason to keep pushing in a way they might not if those cores were only ever going to be used in a phone.

Whether it is better to expend more resources on the GPU (or NPU) than the CPU is another matter, and would depend on Apple's longer term goals. For instance, Vision Pro and its needs, and whatever other products there might be in the pipeline.

Next generation of mobile computing?

AR/VR glasses/headsets and Tablets, and obviously - smartphones.

 

[igor_kavinski]

Thats the first time I have read.

"cough" In this thread "cough"

He suggests two years more before ARM laptops are buy worthy: https://me.pcmag.com/en/migrated-18...rt-card-11-years-in-and-still-not-a-bulls-eye

But a lot can happen in that timeframe. ARM could stagnate while x86 CPUs catch up or surpass them in power efficiency.

 

[Glo.]

He suggests two years more before ARM laptops are buy worthy: https://me.pcmag.com/en/migrated-18...rt-card-11-years-in-and-still-not-a-bulls-eye

But a lot can happen in that timeframe. ARM could stagnate while x86 CPUs catch up or surpass them in power efficiency.

Look at post higher, below the one you responded.

Thats where the industry will go in big picture of tech devices.

 

[smalM]

Hearing this for how many decades now?

Installed base of devices in billion units in 2022:

4.45 Smartphones
0.93 Laptops
0.57 Tablets
0.47 Desktop PC

x86 is already a niche use case compared to ARM-based products.
However, the niche is still quite large...

 

[poke01]

Installed base of devices in billion units in 2022:

4.45 Smartphones
0.93 Laptops
0.57 Tablets
0.47 Desktop PC

x86 is already a niche use case compared to ARM-based products.
However, the niche is still quite large...

It's also in Nintendos Switch and also the upcoming console from Nintendo

 

[eek2121]

Installed base of devices in billion units in 2022:

4.45 Smartphones
0.93 Laptops
0.57 Tablets
0.47 Desktop PC

x86 is already a niche use case compared to ARM-based products.
However, the niche is still quite large...

Yes, now look at ASPs.

For the cost of an RTX 4090 you can buy several smartphones.

New major AAA Games on PC/console sell for $60-$100/pop. Most games on mobile are well under $10.

Apples and orange. AMD, Intel, and NVIDIA would not be the giant multibillion dollar companies they are today if x86 did not exist.

If everyone stopped buying x86 PCs/PC parts/servers of any kind tomorrow, the loss of jobs would plunge the world into a recession or possibly worse. NVIDIA may make big money on AI and GPGPU workloads, but there is good reason they still cater to gamers as well.

 

[Doug S]

Yes, now look at ASPs.

For the cost of an RTX 4090 you can buy several smartphones.

New major AAA Games on PC/console sell for $60-$100/pop. Most games on mobile are well under $10.

Apples and orange. AMD, Intel, and NVIDIA would not be the giant multibillion dollar companies they are today if x86 did not exist.

If everyone stopped buying x86 PCs/PC parts/servers of any kind tomorrow, the loss of jobs would plunge the world into a recession or possibly worse. NVIDIA may make big money on AI and GPGPU workloads, but there is good reason they still cater to gamers as well.

They still cater to gamers because they've already depreciated their investment in developing new GPUs from AI sales so selling GPUs is even more profitable than it used to be. But if they were forced to choose between AI and gamers, they'd drop the gaming market with less than a microsecond's thought.

 

[poke01]

Yes, now look at ASPs.

For the cost of an RTX 4090 you can buy several smartphones.

New major AAA Games on PC/console sell for $60-$100/pop. Most games on mobile are well under $10.

Apples and orange. AMD, Intel, and NVIDIA would not be the giant multibillion dollar companies they are today if x86 did not exist.

If everyone stopped buying x86 PCs/PC parts/servers of any kind tomorrow, the loss of jobs would plunge the world into a recession or possibly worse. NVIDIA may make big money on AI and GPGPU workloads, but there is good reason they still cater to gamers as well.

Gaming is a very niche maket compared to smartphones. Hundreds of millions of smartphones get sold every year.

Consoles take at least 6 years to go past 100 million. Smartphones go past 100 million every year

 

[Mopetar]

That desktop figure seems low. I know that not everyone has one, but the number of desktop PCs in the various offices around the world adds a lot to the count, even if not all families have one. I think many of them tend to stick around for much longer than phones or laptops as well. There's the tech enthusiasts who have have multiple desktops for various purposes. Add school and library computer labs, net cafes in some countries, and other stuff like that and the number seems low.

 

[Hitman928]

Installed base of devices in billion units in 2022:

4.45 Smartphones
0.93 Laptops
0.57 Tablets
0.47 Desktop PC

x86 is already a niche use case compared to ARM-based products.
However, the niche is still quite large...

Where is this data coming from?

 

[Nothingness]

Where is this data coming from?

I bet this: https://www.statista.com/statistics/1228163/global-device-installed-base-by-type-worldwide/

 

[igor_kavinski]

But if they were forced to choose between AI and gamers, they'd drop the gaming market with less than a microsecond's thought.

I would love for them to drop out of the x86 GPU market. We don't need them anymore. Intel and AMD will be enough to take gaming forward. We don't need you anymore, Jensen, you awful bully with leather fetish. Heck, I want Elon Musk to do a hostile takeover of nGreedia and then burn it to the ground.

 

[Tigerick]

Twitter

This leaker mentioned about Qualcomm sharing performance number for M3 Pro and M4, the numbers are from GeekBench5. Let's take average numbers shown below:-

GeekBench5

4P+4E:
M2 ST 1898, MT 8911
A17 Pro (2P+ 4E) ST 2200, MT 6200
M3 ST 2400, MT ?
M4 ST 2600, MT 13500

8P+4E:
M2 Pro ST 1940, MT 14965
8P+6E:
M3 Pro ST 2400, MT 19000

 

[Doug S]

Twitter

This leaker mentioned about Qualcomm sharing performance number for M3 Pro and M4, the numbers are from GeekBench5. Let's take average numbers shown below:-

GeekBench5

4P+4E:
M2 ST 1898, MT 8911
A17 Pro (2P+ 4E) ST 2200, MT 6200
M3 ST 2400, MT ?
M4 ST 2600, MT 13500

8P+4E:
M2 Pro ST 1940, MT 14965
8P+6E:
M3 Pro ST 2400, MT 19000

He said those were Qualcomm "predictions", i.e. worth no more than the predictions you or I might make.

And can I just say against how much I wish that Geekbench's author had time limited Geekbench 5 so it would stop running not long after Geekbench 6 was out? I'm sick to death of people still quoting GB5 numbers when GB6 has been out for ages!

 

[Glo.]

He said those were Qualcomm "predictions", i.e. worth no more than the predictions you or I might make.

And can I just say against how much I wish that Geekbench's author had time limited Geekbench 5 so it would stop running not long after Geekbench 6 was out? I'm sick to death of people still quoting GB5 numbers when GB6 has been out for ages!

Predictions done by engineers are far more accurate, than predictions made by random forum users, or youtube leakers.

 

[Doug S]

Predictions done by engineers are far more accurate, than predictions made by random forum users, or youtube leakers.

Yes but Qualcomm engineers have reason to sandbag Apple's potential gains to leave them an target to hit, or at least not to want to believe Apple will have the gains they themselves will believe they can make on their own designs.

If they closely studied Apple's cores and determined what they are likely to do next and how much benefit that had they might come up with better answers than most, but they will absolutely NOT be allowed to do that by company IP lawyers. They will need to avoid even looking at something like Geekerwan's review that details the A17 changes, because lawyers are so paranoid about engineers opening up the company to lawsuits for patent violations. Probably doubly so given where Qualcomm's Nuvia engineers came from.

Those "predictions" basically read as a sort of placeholder "Apple will have gains similar to the last couple iterations". While that's likely not an unreasonable assumption they don't have any more basis for it than you or I do. Maybe Apple has a massive ground up redesign coming. If so, none of us knows. They (and we) don't even know when M4 will appear, what process it will be made on, and what Axx generation cores it will contain.

 

[smalM]

Yes, now look at ASPs.

For the cost of an RTX 4090 you can buy several smartphones.

Yeah, let's talk money (last year):
Nvidia $27B revenues
Smartphones $400B revenues
Gaming revenues (2021):
Mobile/casual $148.0B
PC games 33.7B
Console 28.3B

Maybe reality looks different from what you think it is...

 

[Doug S]

Yeah, let's talk money (last year):
Nvidia $27B revenues
Smartphones $400B revenues
Gaming revenues (2021):
Mobile/casual $148.0B
PC games 33.7B
Console 28.3B

Maybe reality looks different from what you think it is...

You can keep arguing with him all you want, but some people just don't want to accept that the PC gaming market is just not that important anymore. It could disappear tomorrow and it wouldn't put anyone out of business, and barely affect the fortunes of most.

 

[igor_kavinski]

It could disappear tomorrow and it wouldn't put anyone out of business, and barely affect the fortunes of most.

The Linux desktop market is tiny compared to the PC gaming market yet it still doesn't cease to exist because of passionate users. It's the same with PC gamers. Even if Intel/AMD/nGreedia were to stop all x86 development today, it would take decades for the PC gaming scene to die out (newer generations would shift to the latest and greatest but the old timers are gonna stick with what they love till death do they part).

 

[Glo.]

Yes, now look at ASPs.

For the cost of an RTX 4090 you can buy several smartphones.

New major AAA Games on PC/console sell for $60-$100/pop. Most games on mobile are well under $10.

Apples and orange. AMD, Intel, and NVIDIA would not be the giant multibillion dollar companies they are today if x86 did not exist.

If everyone stopped buying x86 PCs/PC parts/servers of any kind tomorrow, the loss of jobs would plunge the world into a recession or possibly worse. NVIDIA may make big money on AI and GPGPU workloads, but there is good reason they still cater to gamers as well.

TAM is much more important, than individual ASPs.

VAST majority of markets currently are on ARM. We better deal with it and understand that desktop is dying platform.

The Linux desktop market is tiny compared to the PC gaming market yet it still doesn't cease to exist because of passionate users. It's the same with PC gamers. Even if Intel/AMD/nGreedia were to stop all x86 development today, it would take decades for the PC gaming scene to die out (newer generations would shift to the latest and greatest but the old timers are gonna stick with what they love till death do they part).

Again. TAM is everything. PC gamign generates 30 bln USD in revenue. Its 1/5th of mobile gaming, but its still 30 bln USD in revenue.

Mobile gaming however will consume both: console and PC gaming. Its just natural progression of capabilities, and when smartphone SOCs will become robust, in capabilities, just like Apple's SOCs, but not only their - thats the moment when we will see more and more people turning to mobile gaming, because there will be more and more content, more and more games, and more and more people will play them.

Not to mention - its WAY easier to develop for mobile, and then scale it up to desktop and console, than develop for PC/console and scale it down to mobile.

I mean, 40 people team can develop a game in 2 years that can work on: mobile, console and PC increasing TAM incredibly, while cutting costs. Try to develop PC/Console-only game in 2 years with 40 people team. Good luck with that.

 

[Doug S]

The Linux desktop market is tiny compared to the PC gaming market yet it still doesn't cease to exist because of passionate users. It's the same with PC gamers. Even if Intel/AMD/nGreedia were to stop all x86 development today, it would take decades for the PC gaming scene to die out (newer generations would shift to the latest and greatest but the old timers are gonna stick with what they love till death do they part).

Oh I'm not saying it will disappear, I think the PC gaming market will outlive all of us. I was just saying if someone had a magic wand and caused it to disappear, it would leave barely a ripple in its wake.

 

[Tigerick]

iPhone 16 Series Chip Technology Moves to Volume Production

Apple has already given order commitments to chip supplier TSMC for its second-generation 3nm chip fabrication process known as N3E, which is...

www.macrumors.com

As expected, Apple has started the transition to N3E with A17/A18Pro, that's mean incoming M3/M3 Pro/Max/Ultra PCs are still manufactured using N3B process.

@Eug What do you think???