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

 

[IEC]

NPU - 16 core design - 38 TOPs

 

[IEC]

28B transistors,
120GB/s memory bandwidth

 

[Mopetar]

28B transistors,
120GB/s memory bandwidth

That gives it about 3 billion more transistors than the M3. The also did mention it being made on 2nd generation 3nm technology.

 

[Ghostsonplanets]

M4!

4x Performance cores, 6x Efficiency cores, 10 core GPU. New display engine. Uses half the power for same performance as M2.

50% faster (ST?MT?) than M2 but also adding 2 more E cores and higher clocks. No mention of IPC improvements. Seems like Apple really has stalled.

Still, M4 is a good upgrade over M1/M3/M3. 2 more cores, GPU is now 10 core, new display engine. Also 1/2 power to match M2 perf. Really good for the end user.

 

[Mopetar]

50% faster (ST?MT?) than M2 but also adding 2 more E cores and higher clocks. No mention of IPC improvements. Seems like Apple really has stalled.

Given how quickly they released it after the M3, I wonder if it's same cores, just using the N3E node. Maybe a few tweaks or bug fixes, but otherwise the same cores.

 

[Eug]

50% faster (ST?MT?) than M2 but also adding 2 more E cores and higher clocks. No mention of IPC improvements. Seems like Apple really has stalled.

Given how quickly they released it after the M3, I wonder if it's same cores, just using the N3E node. Maybe a few tweaks or bug fixes, but otherwise the same cores.

Note that it's up to 10-core CPU.

 

[IEC]

iPad Pro starting at 11" $999, 13" $1299, base model has 256GB

 

[uzzi38]

50% faster (ST?MT?) than M2 but also adding 2 more E cores and higher clocks. No mention of IPC improvements. Seems like Apple really has stalled.

Still, M4 is a good upgrade over M1/M3/M3. 2 more cores, GPU is now 10 core, new display engine. Also 1/2 power to match M2 perf. Really good for the end user.

Likely MT performance, they didn't talk about a new CPU core. And given the comparison to M2, the comparison device is almost certainly the last M2 iPad.

So you're looking at up to 50% MT performance in a limited power budget with 2 more E cores, 2 generations of CPU improvement and a node shrink. Big improvement for the iPad, but idk, seems kinda like a very mid refresh on the CPU side.

NPU and GPU upgrades both look quite nice though, big upgrades on those sides from the looks of it.

 

[StinkyPinky]

I wonder if the M4 fixes all those vulnerabilities.

 

[Eug]

Note that it's up to 10-core CPU.

256/512 TB models get 9-core - 3P/6E
1/2 TB models get 10-core - 4P/6E

So, that's how they kept the price down. I predicted they'd increase the price US$100 and double the base storage to 256 GB.

However, I predicted M3 with no core binning.
Instead, they did M4 with core binning.

 

[Mopetar]

I'm going to get one of the new 13". I'll have to try out editing in it with Final Cut as it should be pretty beefy.

Anyone have recommendations for a good case manufacturer? I just got a cheap one off Amazon from some no-name Chinese company and the case is falling apart. I'd like something that I can expect to last 5+ years.

 

[SarahKerrigan]

Likely MT performance, they didn't talk about a new CPU core. And given the comparison to M2, the comparison device is almost certainly the last M2 iPad.

So you're looking at up to 50% MT performance in a limited power budget with 2 more E cores, 2 generations of CPU improvement and a node shrink. Big improvement for the iPad, but idk, seems kinda like a very mid refresh on the CPU side.

NPU and GPU upgrades both look quite nice though, big upgrades on those sides from the looks of it.

The benchmark is Affinity Photo's internal bench, which AFAIK is MT.

I'm actually kind of impressed with +50% if it's demonstrated across a reasonably wide set of multithreaded workloads. I don't think that's a bad bump across two gens.

 

[Eug]

I'm going to get one of the new 13". I'll have to try out editing in it with Final Cut as it should be pretty beefy.

I'm going to get the 11" 256 GB 9-core. I'm waiting until next month for the back-to-school promotion though.

I'm gonna get the new Magic Keyboard too.

Anyone have recommendations for a good case manufacturer? I just got a cheap one off Amazon from some no-name Chinese company and the case is falling apart. I'd like something that I can expect to last 5+ years.

Get their high priced keyboard stand?

BTW, on my M1 Mac mini, I had the keynote animated splash screen open before the keynote. I left to do something and when I came back my screensaver was on, and all choppy. I wonder if it was because the keynote animated splash screen was in the background. It shouldn't cause that, but that's the first time the screensaver has ever slowed down like that.

 

[poke01]

Regarding the Gurman article on using M4, I wonder if his source was the iPadOS 14.5 betas containing indications of iPad 16,x and the T8132. That information was apparently around for a while before Gurman's article, so maybe he doesn't have any actual sources for this beyond that. Others might have put the same 2+2 together, but he's more widely published so he'll get the "credit" if we see M4 iPads unveiled tomorrow.

Are they going to ding his 86.5% "prediction accuracy" when it turns out his 18 month cadence thing was bs? Or will they give him credit, since he was kinda/sorta right based on M1->M2 and M2->M3, if you squint a bit and count the months in the way most favorable to that conclusion?

He the code name right for the M4 and other chips in the M4 family. He was the first publication to report that.

Likely MT performance, they didn't talk about a new CPU core. And given the comparison to M2, the comparison device is almost certainly the last M2 iPad.

So you're looking at up to 50% MT performance in a limited power budget with 2 more E cores, 2 generations of CPU improvement and a node shrink. Big improvement for the iPad, but idk, seems kinda like a very mid refresh on the CPU side.

NPU and GPU upgrades both look quite nice though, big upgrades on those sides from the looks of it.

with Apple just like AMD, it’s best to wait for benchmarks. The M3 also seemed like a meh upgrade CPU but it did get improved.

We won’t know anything till they get reviewed by proper reviewers.

 

[Ghostsonplanets]

Note that it's up to 10-core CPU.

Just saw your post about 9 Core with 3P+6E. Interesting. Before it was a binned GPU only.

Likely MT performance, they didn't talk about a new CPU core. And given the comparison to M2, the comparison device is almost certainly the last M2 iPad.

So you're looking at up to 50% MT performance in a limited power budget with 2 more E cores, 2 generations of CPU improvement and a node shrink. Big improvement for the iPad, but idk, seems kinda like a very mid refresh on the CPU side.

NPU and GPU upgrades both look quite nice though, big upgrades on those sides from the looks of it.

M2 on iPad Pro was also a bit power limited, right? So efficiency gains + 2 cores on M4 can help a bit into the 50% gains.

 

[Eug]

He the code name right for the M4 and other chips in the M4 family. He was the first publication to report that.

The T8132 code name was posted in early March at various places, after being tweeted out by a known leaker. Gurman didn't post it until late April.

However, many at the time thought T8132 would be M3 on iPad Pro.

 

[uzzi38]

M2 on iPad Pro was also a bit power limited, right? So efficiency gains + 2 cores on M4 can help a bit into the 50% gains.

Of course yeah, it's a passively cooled tablet. It's going to be capped at like 7-8W sustained or something at best, both the CPU and GPU in prior Mx series products could exceed that individually in the right workloads, much less individually. So this isn't going to hit the top end of the power curve per core. That node shrink is going to play a nice role in those efficiency gains here, as is the 2 extra e-cores, which excel in sub-1w per core situations.

 

[Eug]

Of course yeah, it's a passively cooled tablet. It's going to be capped at like 7-8W sustained or something at best, both the CPU and GPU in prior Mx series products could exceed that individually in the right workloads, much less individually. So this isn't going to hit the top end of the power curve per core. That node shrink is going to play a nice role in those efficiency gains here, as is the 2 extra e-cores, which excel in sub-1w per core situations.

Technically, it wasn't a node shrink. It was a node expansion.

 

[uzzi38]

Technically, it wasn't a node shrink. It was a node expansion.

M2 is on N5P iirc. M2 is the comparison point to M4 here, so in this comparison there is a node shrink taking place.

 

[Hitman928]

Technically, it wasn't a node shrink. It was a node expansion.

He's comparing to the prior iPad which had the M2.

 

[Eug]

M2 is on N5P iirc. M2 is the comparison point to M4 here, so in this comparison there is a node shrink taking place.

He's comparing to the prior iPad which had the M2.

Ah, gotcha.

 

[Nothingness]

NPU is fastest in any AI PC - 38 trillion ops/s

Isn't Qualcomm supposed to reach 45 TOPS?

Edit: That's sarcasm. These figures are not meaningful outside of real applications.

 

[SpudLobby]

Well there it is. M4 indeed.

But remember when everyone said “omg, you think they’ll have a + 20% ST uplift? It’s possible!”

Sure, it’s possible but didn’t seem likely. And guys if they had that they’d be talking about it.

 

[Eug]

1.5X CPU (vs. M2)

Testing conducted by Apple in March and April 2024 using preproduction iPad Pro 13‑inch (M4) units with 10‑core CPU and 16GB of RAM, as well as production iPad Pro 12.9‑inch (6th generation) units with 8‑core CPU and 16GB of RAM. Tested with Affinity Photo 2 v2.4.0.2301 using the built‑in benchmark version 21000. Performance tests are conducted using specific iPad units and reflect the approximate performance of iPad Pro.

4X 3D render (vs M2)

Testing conducted by Apple in March and April 2024 using preproduction iPad Pro 13‑inch (M4) units with 10‑core CPU and 16GB of RAM, as well as production iPad Pro 12.9‑inch (6th generation) units with 8‑core CPU and 16GB of RAM. Prerelease Octane X 2024.1 (a4) 4-09-2024 for iPad tested using a scene with 780,000 meshes and 27 million unique primitives, utilizing hardware-accelerated ray tracing on M4‑based systems and software-based ray tracing on all other units. Performance tests are conducted using specific iPad units and reflect the approximate performance of iPad Pro.

Isn't Qualcomm supposed to reach 45 TOPS?

Edit: That's sarcasm. These figures are not meaningful outside of real applications.

Heh. They weren't as forthcoming with NPU info as they were with CPU and GPU.

I wonder if the M4 fixes all those vulnerabilities.

That could be another reason (besides the cost of N3B) to jump to M4 and cut M3 series short.

Apple Chip Flaw Lets Hackers Steal Encryption Keys

A group of researchers has found a serious security vulnerability in Apple’s M-series of chips that would allow attackers to steal secret keys from Mac and iPad devices. The problem affects Apple M1, M2 and M3 chips — which have been used in Apple desktops, notebooks and tablets since late

www.zetter-zeroday.com

 

[poke01]

Well there it is. M4 indeed.

But remember when everyone said “omg, you think they’ll have a + 20% ST uplift? It’s possible!”

Sure, it’s possible but didn’t seem likely. And guys if they had that they’d be talking about it.

True, it looks a clock speed increase again. That’s not bad cause they are using a different node.

At this point I will stop expecting IPC improvements from Apple and when they come they will come.