News Apple M2 MacBook Air Beat By AMD-Powered Gaming Handheld

[Admin]

Phoronix compares AMD's Zen 4 processors against Apple's M2-based MacBook Air in Linux benchmarks.

Apple M2 MacBook Air Beat By AMD-Powered Gaming Handheld : Read more

 

[PlaneInTheSky]

Ridiculous clickbait article.

At least put the fact it was tested on Asahi Linux in the damn title.

Asahi Linux's GPU driver for Apple silicon is literally a few months old. Last time I tried it I couldn't even get USB to work.

What the hell does a Rog Ally or AMD even have to do with Asahi Linux, why would anyone even make that comparison. Of course it runs better, it has AMD support in the freaking kernel.

 

[Alvar "Miles" Udell]

It's a clickbait title from TH, but the original article clearly states the comparison was only done because forum members requested it, and Apple M GPU support under Linux was still very much a work in progress so the results should be interpreted in context.

 

[phitinh81]

Ridiculous clickbait article.

At least put the fact it was tested on Asahi Linux in the damn title.

Asahi Linux's GPU driver for Apple silicon is literally a few months old. Last time I tried it I couldn't even get USB to work.

What the hell does a Rog Ally or AMD even have to do with Asahi Linux, why would anyone even make that comparison. Of course it runs better, it has AMD support in the freaking kernel.

Salty ? Phoronix only tested CPU general purpose compute . As fair & square as it can be. Linux is the perfect middle ground to make comparison. At face value, Apple silicon's just overhype & overprice pieces of hardware.

 

[vinay2070]

Oh Really? Lets have a comparision once the laptops featuring similar chips with Higher TDPs are out in overeall scenarios. Would definitely like to see AMD win here both in performance and power consumption.

 

[syadnom]

This is a crap article, completely misleading.

Running native linux with good driver support and compiled for target CPU vs running an alpha release with nearly no drivers functioning with software compiled for generic arm64 on an 8 performance core ryzen vs a 4 performance core M2.

Many of these tests run happily on macos so the alpha linux environment is especially misleading.

A couple stand out as exceptionally bad, like the wireguard test where the macos implementation can push out over 500Mbps on the M1 and 700 on the M2 (my personal machines, daily drivers).

John the ripper is also a <Mod Edit> show. I actually use this on macos (macos/m* compiled) because it's faster on my MBP16 M2 Max (8 performance cores) than on my i7 13700 desktop.

The video encoding... wow. m2 x265 on macos absolutely destroys the ryzen per core. I'm not talking video toolbox/ hardware, just CPU. The examples in Asahi are an order of magnitude slower than on macos.

This is maybe the worst article I've seen on TH or Phoronix.

 

[vinay2070]

This is a crap article, completely misleading.

Running native linux with good driver support and compiled for target CPU vs running an alpha release ...

Welcome to the new Tomshardware!

 

[jilive]

Ummm ... folks, Serve The Home also says that the latest AMD Ryzen laptop chips beat the M2 and M2 Pro.

Minisforum UM790 Pro Review Big Upgrade to a Small AMD System

We check out the Minisforum UM790 Pro and see how the company takes a new-generation AMD Ryzen processor and does something different

www.servethehome.com

And they aren't alone, not by a long shot. Tons of actual tech enthusiast sites, instead of the big mainstream "tech" sites, say the same.

You folks need to realize that the American media is filled with people who use Apple products exclusively. They use iPhones, iPads, MacBooks, iMacs and even obscure stuff like the Apple TV and HomePod. The competition routinely beats their preferred Apple products but you never hear them report it. When ChromeOS outsold macOS from 4Q2019 - 2Q2022? They never acknowledged it. The latest MediaTek and Qualcomm flagship CPUs outperforming the latest iPhone chips, especially in graphics? Not a word. When Alder Lake Intel Core i7 and i9 beat the M1 Ultra and M2 Pro? They shifted the goal posts to power efficiency and - I am not making this up - performance when on battery.

Can't wait until September-October when 14th gen Intel chips launch. Their 3rd gen integrated graphics and 2nd gen efficiency cores will FINALLY force this crowd to acknowledge without caveats that 14th gen Intel is better than anything that the M1 and M2 can offer. Of course the M3 will come out soon after and allow them to get right back to playing their games but it won't last long ... AMD's Ryzen 8 will be right around the corner.

 

[NeoMorpheus]

Well…

One, as much as i like AMD, Phoronix clearly stated that its incomplete because Ahai Linux is reverse engineering apple code, because apple doesn’t give a f… about Linux.

Two, its a shame that such good SOC are at the hands of apple, since as stated above, will never get proper support/drivers for amy other OS.

Three, which means, EOL and ot will be e-waste.

Four, corporations rabid fanbois (like the ones from apple and ngreedia) are exhausting.

Five, shame on you toms. But then again, todays media and Tubers are all the same.

 

[palladin9479]

Umm guys ... this is a CPU test not a GPU or storage one, there absolutely doesn't need to be any special drivers involved, just the applications compiled for the native ISA's of the CPU's being tested.

An x86 CPU crushing an ARM CPU is nothing new, they do it all the time because x86 processors tend to be designed for performance while ARM is usually designed for power efficiency. Apple's M2 only looked good early on because they were the first to have access to a new TSMC processing node and therefor got to harvest those benefits while everyone else was stuck on the older nodes. Apple paid a ridiculous amount of money to TSMC for that privilege then controlled what media outlets were allowed to say.

And holy cow the Apple faithful are salty as heck...

 

[KaihatsuJai]

Absolutely howling here.
Mac fans whining about clickbait, obvious whatabouters.
Topic is M2 is old hat, stay focused.

 

[purpleduggy]

why no Thunderbolt/USB4? The chip costs like $15.

 

[purpleduggy]

Ummm ... folks, Serve The Home also says that the latest AMD Ryzen laptop chips beat the M2 and M2 Pro.

Minisforum UM790 Pro Review Big Upgrade to a Small AMD System

We check out the Minisforum UM790 Pro and see how the company takes a new-generation AMD Ryzen processor and does something different

www.servethehome.com

And they aren't alone, not by a long shot. Tons of actual tech enthusiast sites, instead of the big mainstream "tech" sites, say the same.

You folks need to realize that the American media is filled with people who use Apple products exclusively. They use iPhones, iPads, MacBooks, iMacs and even obscure stuff like the Apple TV and HomePod. The competition routinely beats their preferred Apple products but you never hear them report it. When ChromeOS outsold macOS from 4Q2019 - 2Q2022? They never acknowledged it. The latest MediaTek and Qualcomm flagship CPUs outperforming the latest iPhone chips, especially in graphics? Not a word. When Alder Lake Intel Core i7 and i9 beat the M1 Ultra and M2 Pro? They shifted the goal posts to power efficiency and - I am not making this up - performance when on battery.

Can't wait until September-October when 14th gen Intel chips launch. Their 3rd gen integrated graphics and 2nd gen efficiency cores will FINALLY force this crowd to acknowledge without caveats that 14th gen Intel is better than anything that the M1 and M2 can offer. Of course the M3 will come out soon after and allow them to get right back to playing their games but it won't last long ... AMD's Ryzen 8 will be right around the corner.

that's cool, now try connect some Thunderbolt 3 devices to that minisforum minipc.

 

[bit_user]

Salty ? Phoronix only tested CPU general purpose compute . As fair & square as it can be.

It's not "fair & square", because Asahi's support for the M2 is still in pretty rough shape.

For instance, its kernel doesn't even know the difference between the P & E cores, apparently. If you know this, and use numactl, then you can manually run the lightly-threaded jobs on just the P-cores and get much better performance. However, the Phoronix benchmarks don't do that.

These Phoronix benchmarks are, at best, a snapshot of the current state of Asahi's M2 support. At worst, they present a false narrative which people can latch onto without delving into the details, if their preconceived or preferred narratives are affirmed.

 

[bit_user]

Ummm ... folks, Serve The Home also says that the latest AMD Ryzen laptop chips beat the M2 and M2 Pro.

Except those benchmarks don't actually seem to cover the M2 or M2 Pro, leading me to wonder what they're basing that on.

The baseline M2 should be easy for it to beat, since that has just 4 P-cores and 4 E-cores. However, the M2 Pro has 8 P-cores and 4 E-cores. It's not too big of a stretch to believe that Zen 4 can simply ramp up its clocks high enough to get the job done, but nobody can touch Apple on power-efficiency (as STH acknowledged).

You folks need to realize that the American media is filled with people who use Apple products exclusively. They use iPhones, iPads, MacBooks, iMacs and even obscure stuff like the Apple TV and HomePod.

Sorry, we don't deal in conspiracy theories, here. Post actual data or drop it.

The latest MediaTek and Qualcomm flagship CPUs outperforming the latest iPhone chips, especially in graphics?

Then link to some credible benchmarks.

When Alder Lake Intel Core i7 and i9 beat the M1 Ultra and M2 Pro? They shifted the goal posts to power efficiency and - I am not making this up - performance when on battery.

It's not goalpost shifting. Apple has always been in the lead, on efficiency. What was remarkable was when they also managed to take a lead in absolute performance. So, merely reverting to talking about efficiency isn't goalpost shifting, but just pointing out the same thing that's been true all along.

 

[bit_user]

there absolutely doesn't need to be any special drivers involved, just the applications compiled for the native ISA's of the CPU's being tested.

1. You actually do need the OS scheduler to know the difference between the P-cores and E-cores, which apparently it doesn't.
2. You need the encryption benchmarks to use the native acceleration, which isn't happening on the M2.
3. You make a good point about ISA - in this case, they compiled for the generic ARMv8 from 12 years ago, rather than enabling all of the ISA extensions the M2 supports.

An x86 CPU crushing an ARM CPU is nothing new, they do it all the time because x86 processors tend to be designed for performance while ARM is usually designed for power efficiency.

Sort of true. x86 cores are designed to clock higher, which lets them outperform Apple's cores at the expense of burning a lot more power.

As you say, since Apple's existing cores are primarily aimed at phones and laptops, they've prioritized efficiency at the expense of clock speeds.

Apple's M2 only looked good early on because they were the first to have access to a new TSMC processing node

No, it's made on TSMC N5, just like the M1 was. You can easily fact-check this, yourself.

Meanwhile, we're comparing it against an AMD CPU made on TSMC N4. So, the process node advantage is the reverse of what you claim.

Fanboys and haters operate on the basis of faith and tribalism. Fact-based individuals look for good data and follow the logical conclusions based on what it reveals. Which are you?

I don't like Apple as a company or a brand, which makes it annoying that their CPU tech is so good.

​

"Know thyself and know thy enemy = a thousand victories."​

-- Sun Tzu, The Art of War​

​

 

[palladin9479]

1. You actually do need the OS scheduler to know the difference between the P-cores and E-cores, which apparently it doesn't.
2. You need the encryption benchmarks to use the native acceleration, which isn't happening on the M2.

Not really, since they are testing generic multi-threaded CPU performance, which is what the article was about, the OS just schedules it on all cores, both E and P included. And again we're not caring about encryption acceleration, we are just testing the raw compute performance of both chips. We all knew this back during it's initial release, Apple was very serious about what and how reviewers did testing and reporting.

The Apple faithful simple can not tolerate this, it's like discussing the negatives of a BMW around a BMW driver, or the negatives of a Jeep around a Jeep driver, exact same reaction.

The entire issue can be summed up in the phrase "I don't use a PC, I use a Mac", which is like someone saying "I don't drive a car, I drive a BMW". Apple is absolutely genius with it's marketing, like next level genius.

 

[bit_user]

Not really, since they are testing generic multi-threaded CPU performance, which is what the article was about, the OS just schedules it on all cores, both E and P included.

Many are not multi-threaded. For instance, simdjson and QuantLib are both single-threaded. However, I don't know what the percentage breakdown is between lightly-threaded and heavily-threaded.

Even when you have a multi-threaded test, it's often the case that not all of the threads will finish at the exact same time. If the OS scheduler knows which are the P-cores, it can migrate the remaining threads as needed, to make sure they're fully-occupied.

And again we're not caring about encryption acceleration, we are just testing the raw compute performance of both chips.

John The Ripper and some of the OpenSSL test cases are using hardware crypto support, on the AMD CPUs.

I'm curious why you're using "we", in this case. Do you have anything to do with these benchmarks, Phoronix, PTS, or OpenBenchmarking?

We all knew this back during it's initial release, Apple was very serious about what and how reviewers did testing and reporting.

Sorry, what did we know?

The Apple faithful simple can not tolerate this, it's like discussing the negatives of a BMW around a BMW driver, or the negatives of a Jeep around a Jeep driver, exact same reaction.

I think you'd be surprised just how clear-eyed some people can be about their vehicles. At least, the serious ones, that is, because they actually know all the faults and short-comings of them. On the car forums I've frequented, there's a lot less tribalism than I would have expected. Then again, many of them have vehicles of more than one make, so they're already not purists. Also, seeing one thread after another about each gremlin, defect, recall, etc. affecting your model acts as somewhat of a reality-check on those inclined towards brand-supremacy.

Truth be told, the biggest car snob I've met in real life was a Prius owner. By far. And I wouldn't say he was particularly knowledgeable about cars. Maybe the same sentiment exists among Tesla owners, I can't speak from experience but I've heard things to that effect.

The entire issue can be summed up in the phrase "I don't use a PW, I use a Mac", which is like someone saying "I don't drive a car, I drive a BMW". Apple is absolutely genius with it's marketing, like next level genius.

You lost me. Did you mean to say PC? Did you see my update to the post you replied to? I never used an Apple product, nor do I ever plan to. I'm just not so insecure about my decisions around compute platforms and operating systems that I'm threatened by anything positive coming up on the other side of the fence.