Discussion Apple Silicon SoC thread

[DrMrLordX]

Resorting to personal attacks on posters who disagree with you?

Do you even know which user I'm talking about? About the IPC chart threads he created in which he claimed Apple is "destroying the competition" etc.? Don't comment unless you understand what I'm talking about here. The guy has been banned more than once.

Keep that ego in check. I wasn't talking about you.

Just because I am saying it's the best overall mobile SoC/APU on the market doesn't mean it has to win every single esoteric benchmark.

Esoteric for whom? There are users on this forum that frequently make use of 3D rendering and encode software, which is why those two benchmark types happen to be very popular on these forums. That and gaming.

edit: for those wondering why GB5 is held to be an awful benchmark - I'll reiterate: in most cases, GB5 doesn't make my CPU actually do anything. At least Cinebench, y-cruncher, (in its day) w-prime, and Blender keep the execution units of my CPU working. I've watched the temps and the core activity levels while running GB5 MT and I can tell you that it doesn't put much load on the CPU. So from a purely technical point-of-view, benchmarks like GB5 don't really tell me anything about how well I can expect my CPU to work when I ask the most of it. Why am I supposed to care about GB5?

At least AnTuTu is up-front about having a "user experience" benchmark component.

edit: nice how we have an army of downvoters trying to crush dissent in this thread.

 

[therealmongo]

Here have a push back up

 

[guidryp]

Do you even know which user I'm talking about? About the IPC chart threads he created in which he claimed Apple is "destroying the competition" etc.? Don't comment unless you understand what I'm talking about here. The guy has been banned more than once.

Keep that ego in check. I wasn't talking about you.

You weren't referring to single individual, you said "unsavory posters", and "people hiding behind GB5 and SPEC numbers", those aren't singular references, and appear to be blanket attacks on those posting opposing viewpoints who mention those benchmarks.

And whether your attacks are aimed at a single individual, or a group of them, they are in poor form, and just drag down the debate.

edit: for those wondering why GB5 is held to be an awful benchmark - I'll reiterate: in most cases, GB5 doesn't make my CPU actually do anything. At least Cinebench, y-cruncher, (in its day) w-prime, and Blender keep the execution units of my CPU working. I've watched the temps and the core activity levels while running GB5 MT and I can tell you that it doesn't put much load on the CPU. So from a purely technical point-of-view, benchmarks like GB5 don't really tell me anything about how well I can expect my CPU to work when I ask the most of it. Why am I supposed to care about GB5?

That's just why you don't like it. You are only speaking for yourself. GB5 is not definitively held to be an awful benchmark, and even among those who have an issue with it, that doesn't seem to be the main issue.

Also your objections don't hold water. Keeping the activity monitor at 100% and heating up the CPU isn't the function of a performance benchmark. It's testing the relative performance of functional units.

Geekbench is a comprehensive, composite suite of benchmarks. It tests >20 different software areas. It runs test twice, once single threaded and once mult-threaded, making it over 40 individual ST and MT tests, so the activity meter will naturally be shifting constantly as the load varies constantly throughout the test.

The areas tested are detailed here.

https://www.geekbench.com/doc/geekbench5-cpu-workloads.pdf

Note that Geekbench also tests Ray Tracing as 1 of over 20 tests. So it's a superset that includes the only thing that Cinebench tests.

Given the comprehensive nature of GB tests, it would be more generally applicable, than the singular action of CB.

 

[IvanKaramazov]

I'll reiterate: in most cases, GB5 doesn't make my CPU actually do anything. At least Cinebench, y-cruncher, (in its day) w-prime, and Blender keep the execution units of my CPU working. I've watched the temps and the core activity levels while running GB5 MT and I can tell you that it doesn't put much load on the CPU. So from a purely technical point-of-view, benchmarks like GB5 don't really tell me anything about how well I can expect my CPU to work when I ask the most of it. Why am I supposed to care about GB5?

You make some good points and it’s absurd you’re being downvoted. I knew perfectly well who you were taking about, by the way, and I agree.

I do still disagree with you r.e. Geekbench, at least since GB 5 was released. I already mentioned that if you compare GB results with other general CPU benchmarks, they scale with each other very well, suggesting at the minimum that the results are measuring essentially the same things and usually fairly accurately. Have you seen the same things you mentioned in terms of utilization on modern versions of the benchmark?

Geekbench is a short-run benchmark of a few more or less common workloads, and it appears to do a good job of measuring relative performance for short-run, common general tasks. Nothing more nor less. If your primary use case is long rendering tasks, or compiling huge code bases, then GB is obviously not the best testimony to that. But it probably is a better measure of how the average user will experience the performance of their device then, say, Cinebench or Blender. And arguably that makes it a better fit for most device reviews for most people.

 

[jeanlain]

Why do you cut the quoted sentence in half to proceed to state the same?

Oops, sorry. I didn't read thoroughly.

 

[jeanlain]

I'll reiterate: in most cases, GB5 doesn't make my CPU actually do anything.

Are you sure it's not due to the pauses between individual tests?
In their tests, Nuvia did this:

we run Geekbench 5 without the default pauses between subtests to get an accurate power measurement of the active workload.

I'm not sure how they did though.

 

[DrMrLordX]

You weren't referring to single individual, you said "unsavory posters"

There was more than one. He was the primary culprit. Nevertheless, user callouts are not generally permitted and as such I will not mention names nor will I attempt to single out individuals through use of plurals or singulars. I already didn't want to call attention to the lamentable threads he created.

As far as GB5 is concerned, I agree with @IvanKaramazov. Yes, GB5 does have one or two MT test elements that do tax a CPU, along with many that don't. Attempting to mix them all into one benchmark and produce an aggregate score without even exposing the underlying source code for review is, in my mind, less-than-useful for judging the value of a particular CPU in particular workloads. And while M1 is also showing up well in encoding/decoding benchmarks where its fixed-function hardware is utilized, I would like to see it in some software encoding benchmarks that (hopefully) make proper use of NEON. Here's a benchmark based on Dr. Ian Cutress' 3DPM which is a nice, FP-heavy benchmark that will make use of the JVM to exploit NEON, AVX, AVX2, and possibly AVX512 (though I have no confirmed that):

3DPMRedux7122015.zip

Shared with Dropbox

www.dropbox.com

Source code:

3DPMReduxSource7122015.zip

Shared with Dropbox

www.dropbox.com

@jeanlain and @IvanKaramazov

As far as GB5 utilization goes, I look at the PPT for my 3900X @ 4.4 GHz (static). Max PPT is 144W during Structure from Motion, while average PPT across the MT segment is 87W. That's in GB5. In CinebenchR20, average PPT is 158W with peak 161W. That's one run.

When I run 3DPMRedux (which I linked above), I reach max PPT of only 146W, meaning it is not quite as intense as CBR20.

 

[Eug]

I feel for the poor sap who has to edit a 4k video with 16 GB of ram.

...

I think it was Linus Tech Tips complaining that Threadripper could only take 256 GB of ram, and that was just barely enough for his 4k video editing.

For some of the comparisons, FOR CERTAIN CONTENT, the M1 Mac was doing better than a 192 GB Mac Pro for 4K editing. YouTuber type 4K video editing, but 4K video editing nonetheless.

The big advantage here was that the M1 Mac was perfectly smooth in the actual editing process with butter smooth scrolling across the timeline, and perfectly clean playback. In contrast, the Mac Pro was stuttering through the same content.

As I mentioned before, it seems that Apple has purpose built the hardware accelerators to handle this sort of thing. It doesn't cover everything, and sometimes M1 fails hard once you hit 8K, but it's still remarkable what they've done... esp. considering this is a mobile SoC for ultrabooks.

Geekbench was used in Apple comparison, because it was the primary cross platform benchmark that would actually work on iPhones/iPads. I can't even think of anything else that easily fits the bill, unless you want to compile your own. Thus everyone used Geekbench. It's understandable that when it's pretty much all you have, you use it.

IMO scorn for Geekbench grew as performance of recent generations iPhone/iPad showed iPhone delivering desktop performance. The mindset developed that this was "Too good to be True" performance for a Smartphone SoC, and therefore Geekbench must be faulty.

Scorn for Geekbench began with its release, and lasted through Geekbench 3 IIRC. From what I gather, it gained a lot more respect with Geekbench 4 and then Geekbench 5.

The point you make is also probably true, but that is actually a later phenomenon.

Cinebench. I don't remember being prominent until Ryzen hit the scene. Since then it seems to be choice benchmark for AMD, and "AMD people" to show the core count advantage over Intel. If anything it's even less applicable than Geekbench to the real world. Geekbench is a composite benchmark. CB is a just one single task. In a way it's just one of the most simple embarrassingly parallel benchmarks out there.

Cinebench has been popular for just about forever even at AnandTech, at least when comparing Macs and Windows machines since it's cross platform and excludes the GPU. And yes, it's simple, which is actually one of its draws since anyone can run the benchmarks. In fact, there are databases out there which include scores of various Cinebench iterations, at stock clocks and overclocks.

Your experience with these benchmarks may be different though if for example you only started watching this stuff in say the last half-dozen years or so.

 

[jeanlain]

As far as GB5 utilization goes, I look at the PPT for my 3900X @ 4.4 GHz (static). Max PPT is 144W during Structure from Motion, while average PPT across the MT segment is 87W. That's in GB5. In CinebenchR20, average PPT is 158W with peak 161W. That's one run.

I my case, GB5 uses 48W approx during this test, while r23 uses 45W. This is on an i5 7600K, no hyper-threading. prime95 would consume 80W.
Different algorithms, different power usages.
Anyhow, all use 100% of the cores. Not counting the 1-second pauses between GB5 subtests.

 

[guidryp]

Shifting away from kerfuffle about some people attacking benchmarks they don't like:

Here is a new Arstechnica review of MBA. They were declined a review sample laptops, so this is his personal machine he bought to replace an Intel MBA:

Apple’s M1 MacBook Air has that Apple Silicon magic

Review: No piece of hardware is perfect, but man, this is a hell of a laptop.
...
Seriously, I just wasn't expecting the M1—I wasn't expecting it to be this ludicrously fast for the price and the wattage. I wasn't expecting the new chip to just work—though given Apple's previous architecture switches, I probably should have. I wasn't expecting the Air to kick as much ass as it does. Unlike most portables—including the i7-powered 2015 MBA I'm getting ready to retire—it gets the hell out of my way and doesn't make me wait on it when I want to do something.

It's great. And I'm excited to see what Apple does next.

The Good

• Fast as hell
• Battery lasts a long-ass time
• Common x86 apps seem to work perfectly under Rosetta 2
• To this casual laptop user, the M1 feels pretty dang amazing
• Checks almost every box I care about when it comes to hardware I want to own and use

The Bad

• No keyboard backlight adjustment keys
• You might run into Rosetta compatibility issues with less-common apps
• No Windows virtualization (not yet, at least)
• The loss of MagSafe still stings, even after literally years

The Ugly

• It's such an improvement over Intel-based MacBook Airs that you might find yourself spending a thousand unplanned dollars to join the Apple Silicon club

The bottom line
It's like I said in the beginning of the piece: it's fast. The battery life is great. The M1 seems like a hit, and given Apple's success at iterating on their silicon designs, it seems like things are only going to get faster. If you're looking for a portable, and you're not tied to Windows, the M1 MacBook Air is a pretty damn good use of your money.

 

[Eug]

Shifting away from kerfuffle about some people attacking benchmarks they don't like:

Here is a new Arstechnica review of MBA. They were declined a review sample laptops, so this is his personal machine he bought to replace an Intel MBA:

Apple’s M1 MacBook Air has that Apple Silicon magic

One of my top 3 features of all time for Mac laptops is MagSafe. It's such a shame they killed it.

BTW, I see a lot of consumers choosing the fanless throttle-risking MacBook Air over the MacBook Pro, just because of that irritating Touch Bar. Thank the gods though that Apple at least listened a bit and brought back the physical ESC key.

 

[guidryp]

One of my top 3 features of all time for Mac laptops is MagSafe. It's such a shame they killed it.

BTW, I see a lot of consumers choosing the fanless throttle-risking MacBook Air over the MacBook Pro, just because of that irritating Touch Bar. Thank the gods though that Apple at least listened a bit and brought back the physical ESC key.

Agree. If I was choosing between 13" MBA and MBP I would choose Air. No touchbar, and lower price. Having no fan to eventually fail or clog with dust is also a bonus.

I think it would be great if they also made a 15" MBA with the M1 chip. Just a bigger screen on the same internals to keep price down. M1 has plenty of power for most users. Just ran GB on my desktop for fun. It actually did better than I expected (ST 433, MT 1215). So a MBA is only about 4X ST to 6X MT more powerful on CPU side than my old Desktop.

 

[senttoschool]

You weren't referring to single individual, you said "unsavory posters", and "people hiding behind GB5 and SPEC numbers", those aren't singular references, and appear to be blanket attacks on those posting opposing viewpoints who mention those benchmarks.

And whether your attacks are aimed at a single individual, or a group of them, they are in poor form, and just drag down the debate.



That's just why you don't like it. You are only speaking for yourself. GB5 is not definitively held to be an awful benchmark, and even among those who have an issue with it, that doesn't seem to be the main issue.

Also your objections don't hold water. Keeping the activity monitor at 100% and heating up the CPU isn't the function of a performance benchmark. It's testing the relative performance of functional units.

Geekbench is a comprehensive, composite suite of benchmarks. It tests >20 different software areas. It runs test twice, once single threaded and once mult-threaded, making it over 40 individual ST and MT tests, so the activity meter will naturally be shifting constantly as the load varies constantly throughout the test.

The areas tested are detailed here.

https://www.geekbench.com/doc/geekbench5-cpu-workloads.pdf

Note that Geekbench also tests Ray Tracing as 1 of over 20 tests. So it's a superset that includes the only thing that Cinebench tests.

Given the comprehensive nature of GB tests, it would be more generally applicable, than the singular action of CB.

You’re too nice and logical. You’re arguing against people who are currently in denial. They’re no longer using logic at this point. You’re attacking their “manlihood”.

Lets just put it like it is. There are a lot of AMD fans here, and people who’s pride is on the line. They don’t want Apple to take the performance crown because they don’t use Apple products. And if they don’t use Apple, they can’t ever say they have the fastest again.

What do they do when Apple’s entry level chip is beating desktop Zen2 and Zen3 chips in many common tasks and applications? Deny the benchmarks. Claim that they’re not valid. Point to Cinebench numbers as the benchmark that best indicates performance despite it being the most AMD friendly benchmark out there and a super niche one at that.

 

[Roland00Address]

So let’s say the make a M1X chip for 16” with 8 big cores and a higher ram amount.

Do you think Apple would also use this same chip in the 13” pro even though it would be a different motherboard than the air and 13” M1 models?

 

[Qwertilot]

So let’s say the make a M1X chip for 16” with 8 big cores and a higher ram amount.

Do you think Apple would also use this same chip in the 13” pro even though it would be a different motherboard than the air and 13” M1 models?

Yes? Not the 13" pro which has the M1 in now, but they've got a load of 13" models with Intel in at the moment. Those are obviously getting replaced

 

[Doug S]

M1X or M2X is going to massacre everything on the market. My goodness.

Buy Apple stocks. They're clearly poised to increase their laptop market share from 10%. It won't surprise me if they can reach 50% laptop market share in the U.S. in the next 5 years as they have done with the iPhone and iPad.

That seems like wishful thinking. How many people are willing to switch OS to get faster performance? You have to 1) NEED faster performance to the point you are willing to pay more to get it (so people who are just browsing the net and reading email are out) and 2) what you want/need to run is available on macOS (so most gamers and those who run some Windows software that doesn't have an identical version available on Mac are out) and 3) willing to potentially re-buy any commercial software they have (if the vendor doesn't offer some sort of free license transfer from Windows to macOS)

I think Apple can reach double digits in worldwide PC marketshare. Not sure what their current US Mac marketshare is but if they can double it by 2027 (giving them until their entire line is on ARM) I would be frankly shocked. And I say this as someone who owns a huge amount of Apple stock who would like that to be true.

Heck, even if that was true it wouldn't move Apple's stock all that much. Go look at their financials and see how much of their overall revenue is iPhone, iPad, and other non-Mac hardware, and compare with the revenue that is Mac. Even doubling Mac sales worldwide wouldn't cause that much of a bump in their total hardware revenue. Some of their "lesser" products like the Watch and Airpods would rank pretty high in the Fortune 500 all by themselves. You don't understand the scale of Mac sales increase that would be required to move the revenue needle enough for Wall Street to notice.

 

[Doug S]

How are they going to massacre everything when their multithreaded performance is so far behind?

Where are they "so far behind"? Unless you are comparing them against PCs with 8 big cores (and probably SMT enabled as well) they aren't far behind.

And the "massacre" would come when new generations of the M* come with more big cores, eventually scaling up to the Mac Pro in a couple years. I expect at least 32 big cores in the high end there, maybe more. You're going to have to be comparing with some awfully big (and expensive) x86 hardware to put that "far behind".

 

[Doug S]

Are you sure it's not due to the pauses between individual tests?
In their tests, Nuvia did this:
I'm not sure how they did though.

Should be pretty obvious how they did it. Either they used a binary editor to patch out the delays, or more likely just asked the Geekbench author for the source or a special copy of the binary. If you're the Geekbench author and Gerard Williams III emails you asking for a favor, why are you going to refuse?

 

[jeanlain]

Should be pretty obvious how they did it. Either they used a binary editor to patch out the delays, or more likely just asked the Geekbench author for the source or a special copy of the binary. If you're the Geekbench author and Gerard Williams III emails you asking for a favor, why are you going to refuse?

Or maybe they just used Geekbench 5 pro. For $99, I sure hope it has more features than the free version.

 

[Roland00Address]

I think Apple can reach double digits in worldwide PC marketshare. Not sure what their current US Mac marketshare is but if they can double it by 2027 (giving them until their entire line is on ARM) I would be frankly shocked. And I say this as someone who owns a huge amount of Apple stock who would like that to be true.

Just a reminder. Apple is roughly 7% market-share of the PC Market (roughly 265 to 270 million pc computers a year.)



Note these numbers are constantly in flux, and are partly flawed for they are merely "estimates" from people that are doing survey data and thus they are not counting exactly but instead doing the inexact science of polling and prediction. I bring up the flux part for IDC Q32020 numbers think Apple's Market-share is 8.5% during that quarter, yet I provided the image link above which only has 6.6% for a year since that is a yearly number and not a quarter number. (For comparison the Q32019 numbers were 7.0%)

Traditional PC Market Delivers Another Quarter of Double-Digit Growth in Q3 2020, According to IDC

IDC examines consumer markets by devices, applications, networks, and services to provide complete solutions for succeeding in these expanding markets.

www.idc.com

Getting to 10% should be easy (double digit). That is a 50% increase from 6.6% (or 17% increase from 8.5%) but apple dominates in the higher price points, and the price point for the "total industry" is hovering around $630 per computer but that is including the cheapest atom / celeron devices and the most expensive gaming laptops and ultrabooks in a single ASP number.

 

[guidryp]

Just a reminder. Apple is roughly 7% market-share of the PC Market (roughly 265 to 270 million pc computers a year.)

View attachment 34774

Note these numbers are constantly in flux, and are partly flawed for they are merely "estimates" from people that are doing survey data and thus they are not counting exactly but instead doing the inexact science of polling and prediction. I bring up the flux part for IDC Q32020 numbers think Apple's Market-share is 8.5% during that quarter, yet I provided the image link above which only has 6.6% for a year since that is a yearly number and not a quarter number. (For comparison the Q32019 numbers were 7.0%)

Traditional PC Market Delivers Another Quarter of Double-Digit Growth in Q3 2020, According to IDC

IDC examines consumer markets by devices, applications, networks, and services to provide complete solutions for succeeding in these expanding markets.

www.idc.com

Getting to 10% should be easy (double digit). That is a 50% increase from 6.6% (or 17% increase from 8.5%) but apple dominates in the higher price points, and the price point for the "total industry" is hovering around $630 per computer but that is including the cheapest atom / celeron devices and the most expensive gaming laptops and ultrabooks in a single ASP number.

Alternately, Macs seem to represent ~15% of active laptops/desktops connecting to the internet:

Desktop Operating System Market Share Worldwide | Statcounter Global Stats

This graph shows the market share of desktop operating systems worldwide based on over 5 billion monthly page views.

gs.statcounter.com

10 years ago it was only about 5%, so significant growth.

There is also significant regional split.
Lower in Asia, Africa, South America, and India. (near 10%)
Higher in Oceana, and North America. (near 30%).
Europe in between. (near 20%)

 

[Qwertilot]

That seems like wishful thinking. How many people are willing to switch OS to get faster performance?

Definitely not disagreeing with your basic conclusion but the M1 is giving these devices *much more* than extra CPU performance.

Passive, or near silent (mbp), running.
Considerable extra battery life.
Instant restart from sleep.
Enough iGpu to play games at a reasonable level.
Mildly cheaper.
etc

They're all non trivial selling points. No reason they shouldn't pick up some more market share.

Obviously with their using iPads for the 'cheaper' end of the market they won't get a massive percentage of overall laptop sales.

 

[name99]

I thought the M1 was an 8 core CPU?

This is kind of dancing around the terminology if you ask me. If the smaller more efficient cores are being tapped during multithreaded workloads, then the M1 is an 8 core CPU no matter which way you slice it.

Well this is the difference between people whose goal is to understand technology and people whose goal isredacted



Inappropriate language for the tech forums.


esquared
Anandtech Forum Director

 

[guidryp]

That seems like wishful thinking. How many people are willing to switch OS to get faster performance?

I agree, negligible amounts will do that.

I don't think we will see a mass exodus, but Mac seems have been growing slowly but steadily, and I could see that rate improve with a highly compelling product.

Also it might not be a case of abandoning one ecosystem for another, but adding a low end M1 Mac as a test, so they can still use old SW on old machines, and might get it working under Windows Virtualization eventually.

 

[name99]

Why would you think I feel offended? I was just stating what I think is factual. The M1 is an 8 core CPU.

Are we going to say that Intel's Golden Cove with 8 big cores and 8 little cores is not a 16 core CPU as well? If it's doing work, it should be counted!

The very fact that you can say such a thing "If it's doing work, it should be counted!" shows how UTTERLY clueless you are.
Quick question. How many CPUs do you think exist on an M1? 8? 15? 50?
The same is somewhat true (though less extreme) for an Intel chip. No-one bothers counting how many M0 equivalents there are scattered throughout a modern SoC.
Hell, do you even know that there are (at least) THREE classes of Apple-designed cores on an M1? What's that third class doing...? Does Apple have their own Cortex M equivalent running sensor fusion and similar tasks, or do they still use an Cortex M3? Who knows?

Like I said, you can choose to be on the team that cares about *understanding* the technology. Or you can choose to be on the team that thinks scoring points with words counts as an important achievement.