Discussion Apple Silicon SoC thread

Page 427 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Eug

Lifer
Mar 11, 2000
24,114
1,760
126
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:

Screen-Shot-2021-10-18-at-1.20.47-PM.jpg

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:


M1 Ultra discussion here:


M2 discussion here:


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:


M4 Family discussion here:

 
Last edited:

Doug S

Diamond Member
Feb 8, 2020
3,567
6,301
136
If that frequency is correct, then that is a straight up 13% improvement in IPC. I’m surprised we only see a 6% with the A19 series. Perhaps they’ve bifurcated their micro architecture design.

(Maybe my 4500 ST guess is a tad high, even for a Max version. 4300 is a safer bet).

Measuring with the Clang subtest (to eliminate the possibility of a wildcard like a bigger SME unit in the M series cluster affecting results) I see an 11.1% gain with a 0.3% reported clock gain.

That doesn't make sense unless M5 has a different P core than what's in A19, or GB6 didn't accurately report the clock rate. The latter seems more likely, though waiting for a newer P core would be a possible reason why M5 might be delayed until January.

This is also in a passively cooled device that (unless it was added for M5) doesn't have a vapor chamber like the iPhone 17 Pro. Apple doesn't downclock to account for stuff like that, they just let throttling take care of it. Throttling may have affected the reported clock rate, and probably affected the results versus what you'd see in a Macbook Pro with active cooling.
 

mvprod123

Senior member
Jun 22, 2024
398
452
96
Measuring with the Clang subtest (to eliminate the possibility of a wildcard like a bigger SME unit in the M series cluster affecting results) I see an 11.1% gain with a 0.3% reported clock gain.

That doesn't make sense unless M5 has a different P core than what's in A19, or GB6 didn't accurately report the clock rate. The latter seems more likely, though waiting for a newer P core would be a possible reason why M5 might be delayed until January.

This is also in a passively cooled device that (unless it was added for M5) doesn't have a vapor chamber like the iPhone 17 Pro. Apple doesn't downclock to account for stuff like that, they just let throttling take care of it. Throttling may have affected the reported clock rate, and probably affected the results versus what you'd see in a Macbook Pro with active cooling.
Memory bandwidth also matters.
 

jdubs03

Golden Member
Oct 1, 2013
1,280
902
136
Measuring with the Clang subtest (to eliminate the possibility of a wildcard like a bigger SME unit in the M series cluster affecting results) I see an 11.1% gain with a 0.3% reported clock gain.

That doesn't make sense unless M5 has a different P core than what's in A19, or GB6 didn't accurately report the clock rate. The latter seems more likely, though waiting for a newer P core would be a possible reason why M5 might be delayed until January.

This is also in a passively cooled device that (unless it was added for M5) doesn't have a vapor chamber like the iPhone 17 Pro. Apple doesn't downclock to account for stuff like that, they just let throttling take care of it. Throttling may have affected the reported clock rate, and probably affected the results versus what you'd see in a Macbook Pro with active cooling.
Fair points. It basically has to be either or.
A new P-core is kind of due now, the delay could portend that. Definitely possible.

Or yeah it has higher max frequency on the same arch as the A19s, but that’s not accounted for due to throttling. A max frequency of 4.7 GHz (mvprod rumor) would make sense with that ST score and the corresponding IPC increase of 6% like from A18P to A19P.

I don’t think we’ll have much longer to wait. My guess is by the end of October.
 

Eug

Lifer
Mar 11, 2000
24,114
1,760
126
I find it interesting that the M5 release is rumoured to be mirroring the M4 release, with iPad Pro first, and then many moons later, the M5 Macs.
 
  • Like
Reactions: Mopetar

poke01

Diamond Member
Mar 8, 2022
4,195
5,542
106
So far the IPC increase is about ~10% for the M5. Looks like the speed demon theory about all chip designer using mad clocks is dead.
 

Tigerick

Senior member
Apr 1, 2022
844
797
106
FP16 gets 85% boost in TFLOPs partly due to 10% boost in clock speed of GPU....

The boost also requires bigger memory bandwidth; that's why Apple is using LPDDR5x-9600 for 6 GPU cores; bump up from LPDDR5x-8533 (5 GPU).

I haven't watched the Geekerwan video; but based on GB6 Metal performance figures, each GPU core get around 43% boost per core. That's mean even with 10 GPU cores, M5's metal performance will surpass M3 Pro's 18 cores. :cool:
iPad M5-Metal.png

Yep, M5's Metal performance has reached M3 Pro which has about 78000+ scores. If this M5 iPad Pro comes with full GPU dies, then M5 should have 10 cores with impressive graphics power especially RT performance.
 

name99

Senior member
Sep 11, 2010
652
545
136
If that frequency is correct, then that is a straight up 13% improvement in IPC. I’m surprised we only see a 6% with the A19 series. Perhaps they’ve bifurcated their micro architecture design.

(Maybe my 4500 ST guess is a tad high, even for a Max version. 4300 is a safer bet).
It's *possible* that some big functionality (like the new trace cache) couldn't quite ship fully working in the A19 so was chicken-bitted off, by three months or so later they had fixed the issue for the M5.

Though, to be honest, it's more likely that this is all statistical variation and by the time we see enough numbers things will converge to something like an overall 10% for both A19 and M5.
 

jdubs03

Golden Member
Oct 1, 2013
1,280
902
136
It's *possible* that some big functionality (like the new trace cache) couldn't quite ship fully working in the A19 so was chicken-bitted off, by three months or so later they had fixed the issue for the M5.

Though, to be honest, it's more likely that this is all statistical variation and by the time we see enough numbers things will converge to something like an overall 10% for both A19 and M5.
You know more than me. But I’m leaning towards they’ve figured something out to get that extra IPC. Right now ST range is 4079-4175 (3 samples), and the frequency is reported as the same on all three (rounded to 4.42 GHz). But even the lowest number there implies a 10% increase in IPC.
 
  • Like
Reactions: Mopetar

mvprod123

Senior member
Jun 22, 2024
398
452
96
You know more than me. But I’m leaning towards they’ve figured something out to get that extra IPC. Right now ST range is 4079-4175 (3 samples), and the frequency is reported as the same on all three (rounded to 4.42 GHz). But even the lowest number there implies a 10% increase in IPC.
What if Apple decided not to increase the clock speed in the M5 due to the faster LPDDR5X-10700 RAM to avoid increasing power consumption?
 

jdubs03

Golden Member
Oct 1, 2013
1,280
902
136
What if Apple decided not to increase the clock speed in the M5 due to the faster LPDDR5X-10700 RAM to avoid increasing power consumption?
To give it an additional 4-6% in performance? Is GB 6.5 that sensitive to memory bandwidth?

Clang is at 12% so that goes against it being memory specific.
 
  • Like
Reactions: Mopetar

Eug

Lifer
Mar 11, 2000
24,114
1,760
126
Ah no worries there. I was getting lower upload speeds than my 16 Pro Max so I’m thinking I got a dud. Hopefully my carrier will let me replace it and maybe I’ll get luckier.

That’s a nice score right there though. Basically what I would be expecting.
This test is probably meaningless, but here ya go. I did not have anything to compare it against though. BTW, Telus' 5G+ just means it isn't throttled to 250 Mbps, and as mentioned, there is no mmWave support here in Canada. iPhone was set to 5G Auto, not 5G On.

Speedtest-iPhone17PM.jpg
 

Eug

Lifer
Mar 11, 2000
24,114
1,760
126
I've seen a few different people now say the iPhone Air sometimes overheats when recording video outdoors in warm weather. The same people are saying the 17 Pro does not in the same conditions.

I can understand this, as the heat dissipation on this 17 Pro Max really is great. The soft aluminum being prone to denting and scratches is a downer, but so far this phone has stayed noticeably cooler than my prior 12 Pro Max. It'd be good if they could somehow wedge a vapour chamber into the iPhone Air in a year or two.

Also, I'd hazard to guess the real world battery life on the 17 Pro Max is nearly twice as long as the 12 Pro Max. (And it's not due to battery degradation. The 12 Pro Max has 100% battery health, because Apple replaced its battery earlier this year.) In that context, going from a 12 Pro Max to an iPhone Air would actually be fine in terms of comparative battery life, but going from a 12 Pro Max to a 17 Pro Max is a battery life revelation. I wish I had this battery life back in August when I was on vacation. I was continually fearing my battery dying on my 12 Pro Max as we were often away from a charging port when traveling. I don't think I'd have anywhere near the battery anxiety on the 17 Pro Max, because its battery life is truly stellar.

BTW, on my 12 Pro Max, I’ve deactivated cellular, signed out of iCloud and Messages, and turned off background checking for a bunch of apps, but kept mail and what not on it. I still would occasionally surf on it for short periods and check email. With that type of limited usage, I’ve seen that the battery on even that 12 Pro Max can actually last for a couple of days.
 
Last edited:
  • Like
Reactions: Mopetar

Doug S

Diamond Member
Feb 8, 2020
3,567
6,301
136
I've seen a few different people now say the iPhone Air sometimes overheats when recording video outdoors in warm weather. The same people are saying the 17 Pro does not in the same conditions.

Is that one of the reasons why they went to aluminum and vapor chamber in the Pro line, because people on pre 17 iPhones recording a lot of video outdoors were running into this issue, especially when they were recording RAW and had it hooked up to an SSD?
 
  • Like
Reactions: Mopetar

Eug

Lifer
Mar 11, 2000
24,114
1,760
126
Is that one of the reasons why they went to aluminum and vapor chamber in the Pro line, because people on pre 17 iPhones recording a lot of video outdoors were running into this issue, especially when they were recording RAW and had it hooked up to an SSD?
I’m guessing that is one of the big reasons although I don’t know for sure of course. The iPhone 15 series and 16 series both received a lot complaints about overheating under various conditions but especially with video recording. This became so bad that Apple actually made a point to tell the press that it was not due the titanium chassis and was a software issue.


However, it was not limited to the 15 and 16 series. Even with my 12 Pro Max, if I ran CarPlay with Maps navigation and Spotify at the same time as wireless charging it, it would occasionally overheat in the summer if the car was very warm. As for video recoding, my usage is not a good gauge since my recordings are typically below 1 minute and almost always below 4 minutes, and thus I never use external storage.
 

Mopetar

Diamond Member
Jan 31, 2011
8,486
7,723
136
Because Hollywood focuses more and more on megabudget blockbusters, there's a whole new film industry that's sprung up around (relatively) cheap production using gear like iPhones instead of professional cameras, Macs for effects and editing, and distribution on Youtube and other streaming outlets.

Outside of a film shooting special action scenes, unusual environments needed special cameras, or using specific custom lenses to achieve a certain effect there's no reason that a film couldn't be shot on an iPhone.

The hardware is more than good enough for 95% of projects and far less expensive. For example, 28 Years Later was shot on an iPhone and no one would be able to tell unless they were a camera expert.
 

The Hardcard

Senior member
Oct 19, 2021
328
411
106
Is that one of the reasons why they went to aluminum and vapor chamber in the Pro line, because people on pre 17 iPhones recording a lot of video outdoors were running into this issue, especially when they were recording RAW and had it hooked up to an SSD?
My understanding of the titanium was that it was a thin but strong coating over an otherwise aluminum frame, then sandwiched between poorly conductive glass.

I don’t know why they couldn’t do the aluminum unibody with the cutout and just wrapped titanium around the sides. It seems like that give 90 percent of the feel and cut most of the dings, while keeping most of the heat dissipation. The sides wouldn’t match the body of the phone, but they already lost that with the cutout.
 

Jan Olšan

Senior member
Jan 12, 2017
572
1,129
136
Outside of a film shooting special action scenes, unusual environments needed special cameras, or using specific custom lenses to achieve a certain effect there's no reason that a film couldn't be shot on an iPhone.

The hardware is more than good enough for 95% of projects and far less expensive. For example, 28 Years Later was shot on an iPhone and no one would be able to tell unless they were a camera expert.
Have mobile sensors and optics made such huge strides lately?

(Or should I take it in the "Homepod speakers are Hi-Fi audio" sort of sense?)
 
Last edited: