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

 

[Nothingness]

It's a 16" MacBook Pro with the M1 Max. 8P2E CPU core configuration and 32 GPU cores.

Ha it's significantly beefier than my 13" MBP. I doubt BG3 will run smoothly on it (it seems act 3 is heavy).

And this shows my question about what MBP M1 you had made sense

 

[Eug]

When they originally announced the deal it had a two year option available. I think most of us expected Apple could get their own modem working without that extension, but I guess the Intel modem group was more dysfunctional than anyone guessed.

I’m not sure if “most of us” is accurate. I certainly wasn’t optimistic. Just looking back at the press release showed you just how bad the situation Apple was in at the time. It was arguably the most un-Apple press release that Apple had ever released.

 

[poke01]

why wouldn’t Apple use LPDRR5X? They used LPDRR4X before.

 

[poke01]

The base M chip may get more P cores because future Vision Pro generations may need more as the OS gets more demanding.

 

[FlameTail]

why wouldn’t Apple use LPDRR5X? They used LPDRR4X before.

They are using their influence and $$$ to create a custom LPDDR5A!

/joke

 

[FlameTail]

The base M chip may get more P cores because future Vision Pro generations may need more as the OS gets more demanding.

Current 4P+4E setup in M3 is roughly equal to a 6P Ryzen or 2P+8E Intel Core.

For Macbook Air's price you can get laptops with 8P Ryzens and Cores with more multi-thread performance.

 

[eek2121]

You guys are forgetting that Apple may have simply extended the deal because of 💰💰💰

If Apple builds a modem, they will have to license the IP from Qualcomm. Based on Qualcomm’s history, I could see Qualcomm charging more for licensing than it would cost to simply buy the modem. That is how they snuff out competitors after all.

I do not buy the argument that Apple can’t build a good modem. It isn’t an incredibly challenging thing to do in today’s world of software defined radios.

Speaking of modems, how is it that we don’t have any Macs with them?

 

[Nothingness]

It isn’t an incredibly challenging thing to do in today’s world of software defined radios.

Sorry, but you don't know what you are talking about.

 

[Tigerick]

why wouldn’t Apple use LPDRR5X? They used LPDRR4X before.

Cost savings and clever engineering. Here is the Metal benchmark between M2 Pro and M3 Pro, even though M3 Pro's memory BW have been cut from 200GB/s (256-bit LPDDR5) to 150GB/s (192-bit LPDDR5); due to dynamic caching, the performance is only marginal slower. Also M3 Pro has one lesser GPU core, that's mean Apple is able to squeeze performance out of slower LPDDR5.

Therefore, there is headroom for Apple to continue offering future A-series with 64-bit 8GB LPDDR5 without upgrading to more expensive LPDDR5x. So far rumors said A18 series going to come 8GB as standard, that's mean standard iPhone 16 will offer 8GB as standard, it is very Apple way to increase memory size with LPDDR5 but not with LPDDR5x...

As for future iPhone, I am expecting Apple will offer LPDDR6 with A20 Pro in year 2026, so big changes are two years away...you may guess how big memory size going to be?

 

[Eug]

Speaking of modems, how is it that we don’t have any Macs with them?

 

[FlameTail]

Apple's IPC gains YoY have slowed to a crawl over the past few generations.

We thought the A17/M3 generation will bring a major IPC gain, with 3nm and all, but that did not materialise.

They have mainly relied on clock speeds to increase performance over the past few years.

The issue with this is that it is not a sustainable way to increase increase performance, as power efficiency takes a hit in the gut.

Especially with next gen AMD/Intel rumoured to have 20-30% performance improvement, Apple is at risk of being surpassed and left behind.

 

[FlameTail]

First leak of Apple A18 Pro!

ST performance looks solid. >15% improvement. MT performance might lag behind next gen Android SoCs (said to exceed 10 000).

 

[poke01]

MT performance might lag behind next gen Android SoCs (said to exceed 10 000).

MT performance is so easy to achieve. Just add more cores and click them lower. It’s not a mind blowing thing.

 

[Doug S]

why wouldn’t Apple use LPDRR5X? They used LPDRR4X before.

I expect they will use LPDDR5X, it reduces power and increases performance. But Apple needs WAY more than Android OEMs who put it their flagship device, plus there is some degree of "our competition is doing it so we must do it too for when buyers compare specs" that Apple doesn't have to worry about.

When the availability and price work for them, they'll switch. When they do, no doubt people will whine "when is Apple going to support LPDDR5T!"

 

[Doug S]

Speaking of modems, how is it that we don’t have any Macs with them?

Because Qualcomm charges based on the price of the device it is going into, so they'd be paying a fortune to put a modem in a $2000 Macbook. Most end users don't care, they tether their phones so what's the point of a modem in a laptop?

 

[Eug]

Because Qualcomm charges based on the price of the device it is going into, so they'd be paying a fortune to put a modem in a $2000 Macbook. Most end users don't care, they tether their phones so what's the point of a modem in a laptop?

Same reason they put modems in iPads I guess.

That said, we stopped subscribing to cellular service for my wife's iPad. It's now WiFi only. She has a 60 GB phone plan and I have 150 GB, both 5G+ with tethering support, we don't travel much, and most importantly, we can charge on the train, and Apple's iOS auto-tethering support works OK now, so having cellular support on the iPad is no longer as useful.

 

[Doug S]

Same reason they put modems in iPads I guess.

That said, we stopped subscribing to cellular service for my wife's iPad. It's now WiFi only. She has a 60 GB phone plan and I have 150 GB, both 5G+ with tethering support, we don't travel much, and most importantly, we can charge on the train, and Apple's iOS auto-tethering support works OK now, so having cellular support on the iPad is no longer as useful.

The iPad was introduced in 2010, long before tethering was an option for most people because cellular companies discouraged or even actively prevented tethering by their customers.

Another reason you don't see it built into laptops is that you can get a little USB cellular dongle and have cellular capability in any laptop you want, rather than selecting from the small number that have it as a BTO option.

 

[Eug]

I think the Vision Pro is gonna need a lot more computing power going forward, just to get FaceTime looking less creepy:

https://twitter.com/x/status/1752401339752263840

The iPad was introduced in 2010, long before tethering was an option for most people because cellular companies discouraged or even actively prevented tethering by their customers.

? I guess it was different where you're from, but I got a CA$30 6 GB tethering monthly plan at the iPhone launch in 2008. At the time, that was about US$28 AFAIK. (The iPhone launched in Canada with the iPhone 3G in 2008. the original 2G iPhone was not released in Canada.)

Actually, I had tethering long before that, but the data was much more expensive and it was 2G EDGE which was too slow to be of any use. 3G was OK for the time though.

In fact, in the last 15+ years, I've never NOT had tethering support. My main issues with tethering in the past were:

1) Cost of data plans
2) Battery drain on the iPhone
3) Necessity for repeated manual connection on the client device

All of these issues have been addressed for me now:

1) I pay ~CA$47 / US$35 per month for 150 GB 5G+ including Canada and USA, tethering included (no roaming charges in USA).
2) I have a decent USB charging pack, and trains have USB ports and/or electrical plugs
3) Auto tethering reconnection usually works now

Another reason you don't see it built into laptops is that you can get a little USB cellular dongle and have cellular capability in any laptop you want, rather than selecting from the small number that have it as a BTO option.

Agreed.

 

[FlameTail]

Judging by the Apple Vision Pro reviews, future Mx chios definitely need more GPU grunt and display engines.

 

[SpudLobby]

You guys are forgetting that Apple may have simply extended the deal because of 💰💰💰

If Apple builds a modem, they will have to license the IP from Qualcomm. Based on Qualcomm’s history, I could see Qualcomm charging more for licensing than it would cost to simply buy the modem. That is how they snuff out competitors after all.

I do not buy the argument that Apple can’t build a good modem. It isn’t an incredibly challenging thing to do in today’s world of software defined radios.

Speaking of modems, how is it that we don’t have any Macs with them?

Lmao. The cost for Apple to do their own per unit, provided it’s good as you seem to imagine is the case currently, is going to be much less than Qualcomm’s pricing for a modem. That’s the whole reason they bought Intel’s division dude. They want to build something that’s almost as good or not noticeably different but lower cost. Qualcomm’s modems are unbelievably good from peak performance, idle + dynamic power, compatibility and standards perspectives and reception.

They didn’t buy Intel’s division because QC modems weren’t cutting it.

 

[SpudLobby]

The reason Macs don’t have them is because Qualcomm charges Apple by the SKU entry price or whatever. It’d be insanely expensive and it’s too niche.

 

[MarkizSchnitzel]

it is very hard to tell the 10% who do negative work from the 80% who do competent but not outstanding work.

Isn't that what team leaders and middle management is supposed to do?

 

[Doug S]

Isn't that what team leaders and middle management is supposed to do?

In theory, sure.

In practice, it is almost impossible to tell the difference between someone completing an "average" amount of work and someone who does work that makes more work for everyone else. The latter does not become apparent until later parts of the project plan he puts together have missing pieces that when addressed throw the entire schedule off. Code or other "output" (depending on the job) he wrote but didn't test well appears to work but contributes some random errors that take a long time to be tracked down to what he wrote. He mostly fixed a problem so it is closed as an open issue, and when it returns is wrongly seen as a new issue and wastes time until someone realizes it is actually the old problem that wasn't properly fixed.

If someone does something that immediately blows up the next day or following Monday sure that's obvious and if it happens a few times then you know you don't want him on the team. But that's not how negative work manifests, because people creating obvious immediate problems are either booted off the team or if you can't fire them given unimportant tasks that won't negatively affect everyone else. Negative work takes weeks if not months to be seen for what it was, even longer for a consistent pattern of a responsible party to emerge. That's why it causes such big problems, because no one is going to immediately think "oh this has to do with what Bob did back in October" - real negative work results in multiple man days if not multiple man weeks being wasted until it is found/corrected. You have to REALLY be paying attention to establish that pattern. Screwing up and causing more work for others is something everyone does - sometimes high performers do it more often than average performers simply because they often have their hand in so many things at once.

Seeing for who they are the person who at his best is one of the low performers, but creates enough negative work that overall he's a drag on the team is not easy. The only way a team lead sees it is if he takes the "lead" part seriously (rather than just being the guy who knows the most / has been their longest but doesn't necessarily pay close attention to what everyone else is doing, because he figures that's management's job) Depending on how many people a manager has under him he SHOULD be noticing this, but in my experience it takes a pretty exceptional manager to actually do so. In reality it is more likely that multiple team members notice and bring it to the attention of the manager, but then he's got a problem - are they correct, or are they just trying to get rid of a guy they don't like?

 

[Ajay]

In practice, it is almost impossible to tell the difference between someone completing an "average" amount of work and someone who does work that makes more work for everyone else. The latter does not become apparent until later parts of the project plan he puts together have missing pieces that when addressed throw the entire schedule off. Code or other "output" (depending on the job) he wrote but didn't test well appears to work but contributes some random errors that take a long time to be tracked down to what he wrote. He mostly fixed a problem so it is closed as an open issue, and when it returns is wrongly seen as a new issue and wastes time until someone realizes it is actually the old problem that wasn't properly fixed.

No. It's easy. When I worked as a developer - we all knew who performed well and who didn't within our group (everyone working on the same project). Also, any software/firmware company worth it's salt does continuous testing. Maybe remote changes this a bit - IDK.

 

[SteinFG]

Why do you assume that? If Apple was going to be an early adopter of faster DRAM speeds on the Mac, where are the Macs with LPDDR5X?

My guess is they wanted to get to 8533 MT/s ram on M3 Pro/Max, but something went wrong. M2 Pro has 256-bit 6400 MT/s memory, so 204 GB/s theoretical memory speed. M3 Pro chip got smaller 192-bit. To keep the same max memory speed, they'd need ram that at least 8533 MT/s