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

 

[StinkyPinky]

It may not be possible to get any better reviews. How much "real world" software is there in common between MacOS and Windows or Linux? It's almost not worth complaining about it anymore, though. Some people just want their SPEC scores.

Perhaps but there are games with native M1 support for example, rather than relying on rosetta. Would just be interesting to see the CPU/GPU unleashed without being hindered by overhead.

 

[Hitman928]

Denial against Apple's cores seems to be strong - haven't you red Ian's review at all? Cinebench is absolutely low efficiency side of M1 as it is 10-thread 128 bit fpu against 16-thread 256 bit fpu cpus where SMT-scaling is great and increases efficiency a lot - and still M1 can be both better performing and more efficient than Zen3. That still is insane achievement - or putting it other way, x86 cpu manufacturers should be ashamed that they are beat so strongly by cpu from some lifestyle company which main target isn't selling cpus.....

M1 is extremely impressive, no doubt. Clear efficiency leaders and absolute performance (outside of memory bound stuff) isn't even far off core for core with x86's best. Apple does have some advantages though with being a vertically integrated company that AMD and Intel just don't have. If Apple had to rely on actually selling these chips to others to make their money, these chips would be very expensive and would be a hard sell for the vast majority of laptops actually sold in the marketplace. They'd also not have the software support they can provide to actually enable much of the functionality used by the chips. They'd be left with a niche product with limited support on the software side. Usually these types of products don't last long. Apple can do it because they control the design from top to bottom now but even if AMD made the same thing tomorrow, it would have a hard time finding a market to sell in any kind of real volume.

AMD's most similar products are their console APUs where they can do things they can't in their general market products as they are designed in cooperation with the console maker who will request certain features/specs and can make sure the console design from top to bottom works. The difference is game consoles are very low margin products that make up for it in software sales. Apple products are seen as premium products and they can charge accordingly. Apple also has the money to hire the best engineers in the world, which they have done to a large degree, so it shouldn't be a surprise that on top of all of their other advantages, they also just have very good engineers to make it all come together in an optimum design.

 

[Hitman928]

Perhaps but there are games with native M1 support for example, rather than relying on rosetta. Would just be interesting to see the CPU/GPU unleashed without being hindered by overhead.

I'd like to see more Rosetta benchmarks come out with the reviews. It seems like we saw a lot more of this with the M1. Obviously this wouldn't be ideal performance wise for the M1max, but at the same time it would actually tell us what kind of performance can be expected in those scenarios where native isn't supported and everyone should understand that it's not representative of architecture performance in those scenarios without accounting for the Rosetta penalty. It was actually pretty surprising with M1 how well it performed in most cases even with only a Rosetta path available.

 

[BorisTheBlade82]

Ok, then show me where M1max or pro has a anywhere near a 500% advantage over Zen 3 in Cinebench. That was his claim. Or really any compute bottlenecked application that isn't using the custom accelerators.

(Edited)

Do I really have to quote from the AT article for you?

In other ST workloads, the M1 Max is more ahead in performance, or at least in a similar range. The performance/W difference here is around 2.5x to 3x in favour of Apple’s silicon.

In multi-threaded tests, the 11980HK is clearly allowed to go to much higher power levels than the M1 Max, reaching package power levels of 80W, for 105-110W active wall power, significantly more than what the MacBook Pro here is drawing. The performance levels of the M1 Max are significantly higher than the Intel chip here, due to the much better scalability of the cores. The perf/W differences here are 4-6x in favour of the M1 Max, all whilst posting significantly better performance, meaning the perf/W at ISO-perf would be even higher than thi s.

That you do not even realize that when it is plainly written down in front of you by Andrei is very much telling.
I have to admit that it is quite satisfying that basically everything I got burnt for here right after presentation has been confirmed by AT.

 

[Eug]

Regardless of what everybody states, hypes up or play down regarding the M1 family's performance, I think we can all agree that if Intel and AMD care about efficiency Apple clearly is the competition to beat right now.

As for reviews I'm still missing one that explains to me why according to Apple itself M1 Pro/Max Macbook Pro's efficiency while browsing is only half that of M1 Macbook Air. Has anybody seen one such yet?

Are you saying that M1 Pro/Max is less efficient than M1? If so, that would make sense since they only have half the efficiency cores of M1.

 

[Hail The Brain Slug]

Do I really have to quote from the AT article for you?


That you do not even realize that when it is plainly written down in front of you by Andrei is very much telling.
I have to admit that it is quite satisfying that basically everything I got burnt for here right after presentation has been confirmed by AT.

Advantage over Intel's chips isn't advantage over Zen 3.

 

[Doug S]

When I saw the review up I knew the comments would be comedy gold, and I wasn't disappointed. Hard to believe there are morons on this site dumb enough to believe Apple is taking a loss on M1 Max. I guess some people will believe anything if it lets them ignore the results.

The one thing I don't get is all the people whining about the "lack of benchmarks", which as far as I can tell means "where are the game benchmarks". If Anandtech writes an article about IBM's new POWER10 CPU will they say the same thing? Sorry but getting the best gaming performance would require game developers 1) porting to ARM64, 2) porting to Metal, 3) porting to TBDR rendering pipeline. But hey, the poor quality of what ports do exist will allow the haters to claim the game benchmarks reflect Apple's "real" performance and all the other benchmarks have been rigged. Some people will believe some incredible bull**** to avoid upsetting their preconceived order of the world.

The only real hope IMHO for those who want to see Mac gaming become a thing is not that PC game developers jump through all those hoops to do a proper port, but that iPhone game developers decide to scale up one or two of their titles to test the water in the Mac game market. Then you'll have to buy them to demonstrate to them it was worth their trouble, or they won't bother with it again.

 

[Doug S]

CPUs cannot max out memory bandwidth on their own. CPU memory scaling maxes out at 243 GB/s on M1 Max. M1 Pro can max out the 204 GB/s though.

Andrei can get the GPU to use up to 90 GB/s.

This allows extra memory bandwidth overhead for the rest of the SoC.

That makes no sense. Why would they design the M1 Pro to allow the SoC to take all the memory bandwidth, then turn around and "assign" certain chunks of bandwidth to different units in the M1 Max?

The latter is obviously hitting some internal limits, and I'll bet with the right GPGPU load you could take a lot more than 90 GB/sec from the GPU. Having trouble finding a real world GPU load that uses more than 90 GB/sec doesn't mean it is limited to 90 GB/sec, it means you need to explore GPGPU loads if you want to see how much bandwidth the GPU is able to consume.

 

[scannall]

When I saw the review up I knew the comments would be comedy gold, and I wasn't disappointed. Hard to believe there are morons on this site dumb enough to believe Apple is taking a loss on M1 Max. I guess some people will believe anything if it lets them ignore the results.

The one thing I don't get is all the people whining about the "lack of benchmarks", which as far as I can tell means "where are the game benchmarks". If Anandtech writes an article about IBM's new POWER10 CPU will they say the same thing? Sorry but getting the best gaming performance would require game developers 1) porting to ARM64, 2) porting to Metal, 3) porting to TBDR rendering pipeline. But hey, the poor quality of what ports do exist will allow the haters to claim the game benchmarks reflect Apple's "real" performance and all the other benchmarks have been rigged. Some people will believe some incredible bull**** to avoid upsetting their preconceived order of the world.

The only real hope IMHO for those who want to see Mac gaming become a thing is not that PC game developers jump through all those hoops to do a proper port, but that iPhone game developers decide to scale up one or two of their titles to test the water in the Mac game market. Then you'll have to buy them to demonstrate to them it was worth their trouble, or they won't bother with it again.

World of Warcraft runs natively on the M1. It might be a place to see a few results anyway.

 

[Doug S]

World of Warcraft runs natively on the M1. It might be a place to see a few results anyway.

It is still a port of unknown quality. How much time did they spend on making it perform well, versus "OK it runs without crashing, that's good enough for the Mac market - time to get the developers back to working on the PC platform where the real money is!"

 

[moinmoin]

Are you saying that M1 Pro/Max is less efficient than M1? If so, that would make sense since they only have half the efficiency cores of M1.

I'm referring to this post of you yourself. You may be posting too much if you forget your own posts.

This is just based on Apple’s specs but here ya go:

https://twitter.com/x/status/1450286627427917825

View attachment 51607

 

[Eug]

That makes no sense. Why would they design the M1 Pro to allow the SoC to take all the memory bandwidth, then turn around and "assign" certain chunks of bandwidth to different units in the M1 Max?

The latter is obviously hitting some internal limits, and I'll bet with the right GPGPU load you could take a lot more than 90 GB/sec from the GPU. Having trouble finding a real world GPU load that uses more than 90 GB/sec doesn't mean it is limited to 90 GB/sec, it means you need to explore GPGPU loads if you want to see how much bandwidth the GPU is able to consume.

I never said that. I just meant that the CPU - in Andrei’s review - could not saturate that bandwidth.

Same thing with GPU.

 

[Eug]

I'm referring to this post of you yourself. You may be posting too much if you forget your own posts.

Oh yeah. However, as you know, that was not a benchmark. That was just based on Apple’s specs.

I’m just giving you a theoretical reason why - fewer efficiency cores.

 

[moinmoin]

However, as you know, that was not a benchmark. That was just based on Apple’s specs.

I said as much, that it's according to Apple itself. And I am wondering why I'm not seeing any review trying to prove or disprove it, and in the former case, get down to what exactly causes that rather big difference. We already guessed before, that's not what I'm interested in now that reviewers and others have the actual hardware (AT apparently doesn't have the M1 hardware they tested before anymore so their review was useless in that regard).

 

[scannall]

It is still a port of unknown quality. How much time did they spend on making it perform well, versus "OK it runs without crashing, that's good enough for the Mac market - time to get the developers back to working on the PC platform where the real money is!"

Blizzard has been making games for the Mac since they started ages ago. Including World of Warcraft. So it should be a solid build.

 

[Hitman928]

Do I really have to quote from the AT article for you?


That you do not even realize that when it is plainly written down in front of you by Andrei is very much telling.
I have to admit that it is quite satisfying that basically everything I got burnt for here right after presentation has been confirmed by AT.

A 5800u at 15W limited TDP scores ~7500 points in MT for a perf/w score of 500. An M1max scores ~12400 at 34 watts for a perf/w score of 365. Clearly Zen 3 is more efficient than M1max. Additionally, for ST, an M1 chip uses just 3.8W to get the same score as M1max which uses 11W. That gives M1 a 2.9x performance efficiency advantage over M1max. They must have really screwed up the M1max design to do so bad in perf/w compared to M1. (/s)

It's easy to cherry pick data points with zero nuance/context and make whatever point you want to make.

 

[Roland00Address]

Same number of displays for the M1 Pro 8-core and 10-core. The 8-core only has 14 GPU cores though, vs 16.

The 8-core comes with a lower wattage power brick.
You have to pay a bit extra to get the bigger power brick.

I do not get it then. Unless it is just complaining paying more and you do not want too, and thus you create false complaints why there should not be an 8 core model in existence.

The 8-Core is fast enough, and it has the additional features over the 4+4 core M1. For example a better screen, more ports, 12 additional watt hour battery*, etc. Let apple salvage some dies and offer a lower price sku.

*Apple is quoting less battery life though. Maybe this is due to less efficiency cores, the screen, or something else that means it may get 3 hours less max battery.

 

[Gideon]

Many people on twitter (and I think some even here) are missing a very crucial part about the power efficiency of these laptops.

People seem to focus on M1 Pro losing some benches by 5% to a 11th gen Tige Lake H (pulling 100W from the wall for that btw) without having seen an M1 mac, let alone actually run any real world workloads themselves on it vs such a H series beast.

One thing that is as certain as taxes to all premium x86 laptops with decent performance, is that if there is MT load, the fans will kick in. It doesn't matter if it's a Mac or PC, AMD, Intel , 45W or 15W series, when you start a heavier multi-threaded workload (say like Cinebench) on an x86 laptop, the fans will kick on in a matter of seconds. It might take 20 seconds on some and 5 on others but they will kick in aggressively far before the bench is finished.

This has always been a thing but from recent times I vividly remember this happening to all Intel MacBooks at work, my mother's R7 Lenovo Yoga (4800U) wife's x360 Spectre (1065G7), and any gaming laptops I've tried to run it on elsewhere.

The thing is that for both the M1 and the M1 Pro/Max (as confirmed by this guy and another YT review I forgot), the fan doesn't even turn on for that Cinebench MT score you see. Yeah, I mean it eventually does, if you run the bench in loops for 5+ minutes straight. But even then it's nothing like 99% of other laptops. It's almost inaudible (at sub 30db and ~30°C skin-temperatures), while most similar-sized laptops by that point are loud or uncomfortably warm . This power-efficency in extension also means that you get the exact same performance on battery that you do while plugged in.

That is the real gamechanger about these laptops for me.

I do software work, that isn't all that taxing overall, but just enough is going on constantly (unit/integration tests running IDEs indexing, app docker containers running in the background) that no matter what x86 laptop I use , in about half an hour the fans are roaring and my palms sweaty ( I like to call it Eminem programming™ for that).

The only laptop that offered about the same performance as best x86 machines and had no issues staying cold and quiet was the M1 Macbook I tried. If it weren't for it's 16GB memory and 1-external monitor limit it would already fit my needs (but that's why I limp by on the temporary M1 mini until I get the 14" with M1 Pro)

The difference really is that huge and the benchmark numbers alone absolutely fail to tell the story.

The last time I remember being as impressed with a laptop was when I benchmarked my father's brand new Pentium M in 2003 vs my higher end Athlon XP desktop

 

[Hitman928]

Many people on twitter (and I think some even here) are missing a very crucial part about the power efficiency of these laptops.

People seem to focus on M1 Pro losing some benches by 5% to a 11th gen Tige Lake H (pulling 100W from the wall for that btw) without having seen an M1 mac, let alone actually run any real world workloads themselves on it vs such a H series beast.

One thing that is as certain as taxes to all premium x86 laptops with decent performance, is that if there is MT load, the fans will kick in. It doesn't matter if it's a Mac or PC, AMD, Intel , 45W or 15W series, when you start a heavier multi-threaded workload (say like Cinebench) on an x86 laptop, the fans will kick on in a matter of seconds. It might take 20 seconds on some and 5 on others but they will kick in aggressively far before the bench is finished.

This has always been a thing but from recent times I vividly remember this happening to all Intel MacBooks at work, my mother's R7 Lenovo Yoga (4800U) wife's x360 Spectre (1065G7), and any gaming laptops I've tried to run it on elsewhere.

The thing is that for both the M1 and the M1 Pro/Max (as confirmed by this guy and another YT review I forgot), the fan doesn't even turn on for that Cinebench MT score you see. Yeah, I mean it eventually does, if you run the bench in loops for 5+ minutes straight. But even then it's nothing like 99% of other laptops. It's almost inaudible (at sub 30db and ~30°C skin-temperatures), while most similar-sized laptops by that point are loud or uncomfortably warm . This power-efficency in extension also means that you get the exact same performance on battery that you do while plugged in.

That is the real gamechanger about these laptops for me.

I do software work, that isn't all that taxing overall, but just enough is going on constantly (unit/integration tests running IDEs indexing, app docker containers running in the background) that no matter what x86 laptop I use , in about half an hour the fans are roaring and my palms sweaty ( I like to call it Eminem programming™ for that).

The only laptop that offered about the same performance as best x86 machines and had no issues staying cold and quiet was the M1 Macbook I tried. If it weren't for it's 16GB memory and 1-external monitor limit it would already fit my needs (but that's why I limp by on the temporary M1 mini until I get the 14" with M1 Pro)

The difference really is that huge and the benchmark numbers alone absolutely fail to tell the story.

The last time I remember being as impressed with a laptop was when I benchmarked my father's brand new Pentium M in 2003 vs my higher end Athlon XP desktop

I don't know if it was having kids or just getting older in general, but my appreciation of a quiet running machine has greatly increased over the years. Having the fans not kick in would be a big selling point for me. With that said, isn't the trade off Apple is making increased weight for the cooling system or is the laptop a similar weight as other machines in its class?

 

[Eug]

I do not get it then. Unless it is just complaining paying more and you do not want too, and thus you create false complaints why there should not be an 8 core model in existence.

The 8-Core is fast enough, and it has the additional features over the 4+4 core M1. For example a better screen, more ports, 12 additional watt hour battery*, etc. Let apple salvage some dies and offer a lower price sku.

*Apple is quoting less battery life though. Maybe this is due to less efficiency cores, the screen, or something else that means it may get 3 hours less max battery.

Battery life for 8-core and 10-core is listed as being the same. The difference here though is you can only get the 8-core in the 14" model. There is no 8-core available for the 16" model. The 16" model has more battery life, but that increased battery life is true compared to both of the 14" models.

I think it's because people saw the benchmarks which showed GB5 at 12500 for the 10-core, and then saw the benchmark of 9950 for the 8-core and then concluded it's foolish to go with the 8-core since the 10-core is only US$200 more, regardless of what their workload and needs are.

Perhaps the way some should be looking at it IMO is that the 8-core M1 Pro gets 30% higher CPU benchmark scores than M1, massively improved memory bandwidth over M1, a hugely improved GPU, and additional hardware encode/decode support absent on the M1.

 

[insertcarehere]

This is for everyone who is claiming M1 Max is leagues ahead of It's x86 competition in energy efficiency.

Techspot review of Ryzen 7 5800U

M1 Max(Package power: 34W) vs Ryzen 9 5980HS(Package power: 35W)
R23 MT: 12375 vs 11024
Difference in performance is only 12% in favour of Apple at similar power draw.

What would happen If I compared It against 15W Ryzen 7 5800U?
M1 Max(Package power: 34W) vs Ryzen 9 5800U(Package power: 15W)
R23 MT: 12375 vs 7394
Apple is 67% faster, but consumes 127% more power than AMD, which makes this AMD CPU 36% more power efficient. BTW It was a 10min long run, just to be clear.

And here I thought people frequenting this website would know better than to equate package power with TDP. Unless the reviewer explicitly monitored package power consumption, it is not possible to compare the two given how different laptop OEMs will calibrate profiles for the same chip.

Case in point, the 5900HS has a 35w TDP, which Asus (the only OEM w/ a 5980HS laptop) happily ignored in the implementation with power profiles that draw 65w & 45w from the CPU, respectively. https://www.ultrabookreview.com/46631-asus-zephyrus-g14-2021-review/

 

[Roland00Address]

@Eug I was comparing 13 inch M1 4+4 core vs the 14 inch M1Pro 6+2 battery life. But yeah we see eye to eye where there is a place for the 6+2 Pro and it should be seen as a step up from the 13 inch M1.

1699, for a 13” M1 4+4 with 16gb ram and 512 storage.
1999, for a 14” M1 6+2 Pro with same ram and storage.
2199, for a 14” M1 8+2 Pro with same ram and storage.

Likewise for gpu there is a 1400 delta from 1699 to 3099 depending on if you want an 08, 14, 16, 24, or 32 core GPU. Lots of step up opportunities, I can make the case one should stop at 2k plus taxes, or spend 2.2k.

 

[Hitman928]

And here I thought people frequenting this website would know better than to equate package power with TDP. Unless the reviewer explicitly monitored package power consumption, it is not possible to compare the two given the laptop OEMs calibration.

Case in point, the 5900HS has a 35w TDP, which Asus (the only OEM w/ a 5980HS laptop) happily ignored in the implementation with power profiles that draw 65w & 45w from the CPU, respectively. https://www.ultrabookreview.com/46631-asus-zephyrus-g14-2021-review/

Your point is valid, but from what I've seen, the 35W number in the Techspot review is with the CPU really drawing 35W. At 65W, Cezanne should score pretty close to 13,000 points in Cinebench r23 as shown in your link.

 

[Leeea]

Anyone have any news on how well these things mine Ethereum?

Thinking with all that bandwidth it would be fantastic.

 

[Markfw]

Anyone have any news on how well these things mine Ethereum?

Thinking with all that bandwidth it would be fantastic.

quoting your own post, 408gb/s

Page 36 - The anti-crypto thread

Page 36 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com