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

 

[Staples]

It is pretty scary how performant this thing is. We all know no other manufacturer has been able to make chips as fast as those in iPhones. I wonder five years down the line if Intel or AMD can make CPUs that even give the Mx series any competition. I've always used PCs because I've been concerned about performance but if Apple is now the most performant, lots of computer nerds may make the switch.

 

[mikegg]

It is pretty scary how performant this thing is. We all know no other manufacturer has been able to make chips as fast as those in iPhones. I wonder five years down the line if Intel or AMD can make CPUs that even give the Mx series any competition. I've always used PCs because I've been concerned about performance but if Apple is now the most performant, lots of computer nerds may make the switch.

A lot of PC nerds claim that you get more performance for the money with Android/Windows. This hasn't been true for a long time. A $400 iPhone SE is significantly faster and has higher performance than any Android phone, period.

And now it's most definitely not true for laptops. The $1,000 Macbook Air's CPU is faster than any Windows laptop, period. And has way longer battery life.

 

[itsmydamnation]

And now it's most definitely not true for laptops. The $1,000 Macbook Air's CPU is faster than any Windows laptop, period.

my highlight

I love statements like this that within 3 seconds of visiting the anandtech front page can be seen as false.

 

[Staples]

A lot of PC nerds claim that you get more performance for the money with Android/Windows. This hasn't been true for a long time. A $400 iPhone SE is significantly faster and has higher performance than any Android phone, period.

And now it's most definitely not true for laptops. The $1,000 Macbook Air's CPU is faster than any Windows laptop, period. And has way longer battery life.

True. I have an iPhone SE. Actually not because it is the fastest but because it is the ONLY half way decent small phone out there. I hate big phones.

As far as laptops, the CPU may beat most things out there but if you use descrete GPUs, the M1 does not win. Maybe in the future but since GPUs are specialized as it is, the competition is more stiff than a powerful ARM vs x86.

 

[mikegg]

my highlight

I love statements like this that within 3 seconds of visiting the anandtech front page can be seen as false.

Which Windows laptop has a faster CPU than M1? And we haven't even factored in additional things like enconding/decoding engine, image engine, AI acceleration, storage controller, and the fastest integrated GPU in the world.

 

[mikegg]

True. I have an iPhone SE. Actually not because it is the fastest but because it is the ONLY half way decent small phone out there. I hate big phones.

As far as laptops, the CPU may beat most things out there but if you use descrete GPUs, the M1 does not win. Maybe in the future but since GPUs are specialized as it is, the competition is more stiff than a powerful ARM vs x86.

Yes, hence, I said fastest CPU.

I expect the future Macbook Pro 16" M1X/M2X integrated GPU to really give discrete GPUs a run for their money though.

 

[Eug]

Which Windows laptop has a faster CPU than M1? And we haven't even factored in additional things like enconding/decoding engine, image engine, AI acceleration, storage controller, and the fastest integrated GPU in the world.

There are tons of Windows laptops faster than M1. Not in that power envelope, but faster nonetheless.

 

[mikegg]

There are tons of Windows laptops faster than M1. Not in that power envelope, but faster nonetheless.

Wait what? Example? Didn't benchmarks prove that the M1 has the fastest overall CPU performance out of any laptop chip?

I mean look at stuff like this:

https://twitter.com/x/status/1328700910546132993

 

[Eug]

Wait what? Example? Didn't benchmarks prove that the M1 has the fastest overall CPU performance out of any laptop chip?

I mean look at this:

https://twitter.com/x/status/1328700910546132993

No they did not. And Apple didn't claim M1 was the fastest overall either.

I think you're referring to the fastest single core claim.

M1 does win some multi-core tests though.

I mean it's an awesome chip, but let's not get ahead of ourselves here.

 

[wlee15]

Yes, hence, I said fastest CPU.

I expect the future Macbook Pro 16" M1X/M2X integrated GPU to really give discrete GPUs a run for their money though.

Here's a laptop with 3950x which is a full size desktop processor.

XMG Apex 15 review: Ryzen 9 3950X in a laptop - and we're not kidding

The AMD Ryzen 3950X processor has 16 cores and 32 threads, clocking up to a maximum of 4.7GHz. It's the flagship of AMD…

www.eurogamer.net

 

[mikegg]

No they did not. And Apple didn't claim M1 was the fastest overall either.

I think you're referring to the fastest single core claim.

M1 does win some multi-core benchmarks though.

I mean it's an awesome chip, but let's not get ahead of ourselves here.

Ok, yea I'm aware that it doesn't win every multi-core benchmark. That doesn't mean it isn't the fastest laptop chip overall. It clearly *is* the fastest laptop chip overall in the world for the vast majority of people. It only loses in a few multi-core benchmarks.

The M1 clearly outclasses AMD and Intel CPUs in single core, encoding/decoding, code compilation, video editing, A.I. inference, and web browsing.

For me, I do a lot of code compilation, web browsing. The M1 is clearly the fastest laptop chip, and probably the fastest laptop/desktop chip for what I do.

 

[mikegg]

Here's a laptop with 3950x which is a full size desktop processor.

XMG Apex 15 review: Ryzen 9 3950X in a laptop - and we're not kidding

The AMD Ryzen 3950X processor has 16 cores and 32 threads, clocking up to a maximum of 4.7GHz. It's the flagship of AMD…

www.eurogamer.net

Yes, I'm aware some companies have crammed a 3950x into a laptop and run it in max 65w mode.

It probably still loses to the M1 chip overall since a 105w 3950x often loses in benchmarks to the M1. Now run it in 65W with limited cooling capacity in a laptop.

The M1 trades blows with desktop Zen3 processors. A Zen2 3950x running in 65W with limited cooling has no chance.

 

[Eug]

BTW, some people are saying that trying to run Premiere Pro on it through Rosetta 2 is a disaster.

 

[beginner99]

lots of computer nerds may make the switch.

If you switch to a closed platform, you loose your "nerd" designation.

 

[mikegg]

BTW, some people are saying that trying to run Premiere Pro on it through Rosetta 2 is a disaster.

Way to cherry-pick a benchmark. 99.99% of people don't ever use Cinema4D.

And here's a result of the M1 beating the 4900HS by 5% in multi-core, and 54% in single core.



Regardless, I think it's safe to say that the M1 is the fastest overall CPU for laptops. And we haven't factored in things like video encoding/decoding, image optimization, AI inference, which the M1 has dedicated accelerators for.

And when we look at the overall package, it looks even worse for AMD 4900HS. M1 has a significantly faster GPU and offers things like instant-on and much longer battery life.

 

[igor_kavinski]

M1 using an SLC cache? It's gonna wear out eventually. What then? Does the device go into a low performance mode? Or does it stop working altogether?

Memory bandwidth issues may get solved with four channels of DDR5 in the future though that's likely a year or two into the future.

If Parallels can deliver the 70-80% performance of native x86 code execution, that could conceivably meet or exceed the highest level of performance delivered by contemporary high end x86 CPUs.

Apple seems to be content with targeting the content creation market. The x86 gaming market is safe for now. Intel and AMD can breathe easy. The only other company with the remotest chance of challenging Apple in performance could be nVidia with their ARM IP but that could take maybe 5 years or longer.

 

[Eug]

If you switch to a closed platform, you loose your "nerd" designation.

You've mentioned the closed platform thing 1000X now in this thread. We get it.

Way to cherry-pick a benchmark. 99.99% of people don't ever use Cinema4D.

No they don't, but you were the one saying M1 beats everything, which simply wasn't true.

Regardless, I think it's safe to say that the M1 is the fastest overall CPU for laptops. And we haven't factored in things like video encoding/decoding, image optimization, AI inference, which the M1 has dedicated accelerators for.

M1 is apparently totally amazing for Final Cut, beating out high end trash can Mac Pros and recent MacBook Pros.

But as mentioned, running through Rosetta, some people are saying it's terrible for Premiere Pro. Sure, that is not a "fair" test because Premiere Pro is not native, but that's irrelevant for them. For a lot of people (including YouTubers), Premiere Pro is the editor of choice. People need to get stuff done and for some stuff it's better than others. I agree it looks good for your usage though.

Apple seems to be content with targeting the content creation market. The x86 gaming market is safe for now. Intel and AMD can breathe easy. The only other company with the remotest chance of challenging Apple in performance could be nVidia with their ARM IP but that could take maybe 5 years or longer.

Macs aren't good as a gaming platform in general, but you should remember that M1 is Apple's low end chip. I'm 100% sure that the GPU will be much, much faster in the 2021 MacBook Pros and iMacs, as will the CPU.

 

[mikegg]

No they don't, but you were the one saying M1 beats everything, which simply wasn't true.

I didn't say the M1 beats everything in everything.

I just said it's the fastest overall laptop CPU in the world. I feel confident in that statement based on synthetic benchmarks and real application benchmarks.

 

[teejee]

M1 using an SLC cache? It's gonna wear out eventually. What then? Does the device go into a low performance mode? Or does it stop working altogether?

No, it doesn't use SLC as cache, but I think it has a LLC cache though (Last Level Cache).

 

[mikegg]

If you switch to a closed platform, you loose your "nerd" designation.

It's been a while since I used Android and Windows. How much customization do you really get nowadays? Isn't Google trending towards a standard Android experience anyway? And Google is trying its hardest to close any loop holes for its 30% cut in app revenue.

What are some things nerds like to do on Android that they can't do on iOS? Same question for Windows and MacOS.

 

[igor_kavinski]

No, it doesn't use SLC as cache, but I think it has a LLC cache though (Last Level Cache).

From AT https://www.anandtech.com/show/16252/mac-mini-apple-m1-tested:

The M1 also contains a large SLC cache which should be accessible by all IP blocks on the chip. We’re not exactly certain, but the test results do behave a lot like on the A14 and thus we assume this is a similar 16MB chunk of cache on the SoC, as some access patterns extend beyond that of the A14, which makes sense given the larger L2.

That could be a typo...

 

[mikegg]

$1000 Macbook Air plays back Cannon R5 footage smoother than $15,000 Mac Pro:

The magic of extra accelerators in the M1 plus Apple controlling the software stack to actually take advantage of the accelerators.

This is what I mean when I say it isn't totally fair to the M1 to test synthetic benchmarks since it packs so many different accelerators.

Thus, the M1 should be faster for most people than even the benchmarks and reviews show.

 

[tamz_msc]

If you switch to a closed platform, you loose your "nerd" designation.

This open-vs-closed ecosystem debate isn't as important for the end-user as you think it is.

 

[mikegg]

This open-vs-closed ecosystem debate isn't as important for the end-user as you think it is.

For 99.99% of users, it isn't a factor at all. It's the 0.01% nerds where it matters. A lot of them are on this forum.

 

[jeanlain]

And here's a result of the M1 beating the 4900HS by 5% in multi-core, and 54% in single core.

To be fair, even if Geekbench may be more representative than cinebench for general purpose computing, it doesn't scale well at all with the number of cores. So it naturally favours high-performance cores.
If you're interested in rendering or transcoding, which scale very well with the number of cores, geekbench MT scores are not good indicators or relative performance.