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

 

[The Hardcard]

no real use cases other than ultra-niche, such as:

…- lidar, LLM, data science etc are also niche easily <1% user market share

What does it matter if real use cases are ultra-niche or not? First, 1% market share is still millions of people. Second, I’m not convinced that is only 1%. I think it is probably closer to 10% since there are thousands of niches and certain ones are above 1 million people each. There are, for instance, 3 million commercial truck drivers in the United States.

sounds like software issue 10000%

It is a software issue, however, not just in the way you are implying. While there are no doubt a lot of poor decisions, bugs, and lack of optimization, fixing all of that would not eliminate the issue.

The issue is that the software has to be complex, has to be extremely branchy, and on one thread. It is software that enforces real-time compliance with hundreds of government regulations and dozens of individual company rules that hit differently, depending on the time of day, the location, the type of work being done in relation to the types of work done and other conditions going back 8 days at any given moment.

It is simply more work than even current processors can do on a thread and still smoothly keep up with the user. This is not an uncommon scenario and why advances in CPU, GPU, and NPU cores will continue to provide real world benefits to users for years to come. Double the performance of the A19 Pro P core will still leave millions of “ultra niche”users looking for more.

- ios/android AAA games are very few
- they all run near smooth 60fps on iphone 12

I don’t know where you’re coming from here. There are numerous videos on YouTube of even the iPad M4 not getting 30 frames per second on certain current games. Plus, games are one of many moving compute targets. The demand for gaming resources continues to increase.

finally: need real AAA games or professional highend computing?

just get a Halo handheld

Phones are the only devices with this type of computing power that does not entail compromises. Halo handhelds are still too bulky to be seamlessly carried around and used at will.

 

[Eug]

^^^ I just saw a review of the 17 Pro Max vs the 16 Pro Max and even Minecraft was noticeably smoother on the 17 Pro Max. The GPU is an enormous upgrade even just over this one generation.

 

[smalM]

recently I read the average age of an iPhone in use was 5.7 years IIRC.

Average life span.
As an iPhone.
My oldest iPhone is a 7 Plus used as an iPod in my car.

 

[johnsonwax]

So back in early 2021 Tim Cook reported Apple had an installed base of 1.65 billion active Apple products, including "over 1 billion" iPhones. The most recent number for their installed base was 2.35 billion earlier this year, but the number of iPhones was not broken out. But if the overall installed base grew by 40% surely iPhones have grown a lot. Maybe not by the whole 40%, I'm not sure what else is included in that aside from iPhone, Mac, iPad and Watch. Do Airpods count?

Apple's yearly iPhone sales fluctuate, but have averaged around 225 million a year for the past decade. So given that Apple sold their billionth iPhone in 2015 that's well over 3 billion sold in total. Most of the ones older than a decade have likely been broken or retired by now, but everything sold in the past decade is still getting security updates (the 6S released in 2015 received its most recent security update a few weeks ago) so aside from breakage those are still gonna be in use somewhere.

So it is easy to believe the installed base has an average age of 5.7 years. The "upgrade cycle" of 3.7 years is talking about first owner. Obviously the people able to spend $500 to $1000 or more on a new smartphone aren't going to hold onto it until it is ready for retirement ~10+ years on.

There was a similar statistic I saw a few years ago that the average car in the US was 12.5 years old. The people are buying brand new cars aren't holding on them for 20+ years account for that 12.5 year average. They sell or trade them after 3-5 years and buy another brand new one, and it goes through multiple owners until it finally ends up in the crusher when it is ~25-30 years old (or is retired sooner if its in an accident or has a mechanical issue serious enough it isn't worth repairing)

A lot of this disconnect comes from the fact that the iPhone resale market is relatively strong and the Android one mostly nonexistent outside the small share of flagship phones. The 2nd hand iPhone market is around $65B/yr. That's about 40% the size of the global PC market - and that's just iPhones.

Apple laid out this plan back in 2017 when they said they were going for more durable phones with longer replacement cycles, longer software support cycles, a greater investment in the resale market, and a shift toward services - because while Apple doesn't take a lot of the direct revenue off of the 2nd hand sales, those 2nd hand customers are equally valid for services, and in some cases more valid service customers when it comes to repairs. Services scale not off of annual sales, but off of installed base. That's a big part of why their service revenue exploded so much.

But you don't see that in the Android market. Android is competing much more for first party sales. The OEMs don't make meaningful money off of services - not even Samsung so that secondhand market isn't really worth nurturing. ApplePay has almost 800 million users and processed about $6T in transactions last year. Apple gets as much as 15 basis points from those transactions. Those secondary iPhone sales are as good as the primary in that regard.