MacBook Air Teardown Reveals M2 Chip and Single Storage Chip for 256GB Model

[cupcakes2000]

The point is, non-geeks don't quit apps in macOS because it's not intuitive to do so. The fact that you need a shortcut to do that is relevant. As a result, memory usage will be higher for new macOS users, especially those who choose 8/256.

It's relatively easy to force quit apps in iOS as well, but most people don't it because it's not intuitive.

You don’t need a shortcut to do it though. You click quit. It’s identical to what you do in windows, but it’s in a different place. you can use the shortcut if you like, I assume windows has one to.

I would argue that windows is less user friendly, because if you click that giant red X in error, the app quits completely rather than just closes like in macos.

 

[Analog Kid]

Those non-geeks are the same people that would buy 8/256 and then turn to swap because they like to leave tabs and programs open. Unlike Windows, macOS doesn't make it easy to quit an app.

This is the problem with people taking one benchmark, essentially a Blackmagic throughput test, and extrapolating all kinds of real world usage impacts.

The difference in throughput, for massive sequential read accesses is roughly 1.5GB/s versus 3GB/s. There's less of a difference for writes, which are in the background anyway for swapping. That means that if you're reading a contiguous 1GB, it'll take 0.3 seconds longer to change tabs.

Let's count to 0.3 together, shall we? Ready? One, two... dang, let's try that again...

But how much is actually being swapped because you have a tab or unused application open? If I look at my Activity Monitor, I see a rare few that might be as much as 300MB. So it would take 0.1 seconds longer to read that back from swap.

Except that's not how the virtual memory works on MacOS-- long before the system starts paging to disk, it starts compressing the memory and then, when it has to, it pages the compressed memory to disk. When I look at the compressed VM size in Activity Monitor, the biggest I'm seeing is on the order of 70-80MB. So about 25 milliseconds different.

For reference, with a 60Hz display, the screen refresh time is 16 milliseconds. How often are your non-geek users switching tabs and apps that they're noticing the added disk access time is slowing down their day?

Of course, the VM doesn't page out the entire memory space at once, it does it 16kB at a time based on the algorithm's estimate of which memory is least likely to be accessed. So how much of the above analysis is even relevant? What do we know about small random accesses which are dominated by latency rather than throughput? Nothing that I'm aware of.

And that's the point. Using a throughput benchmark to make generalizations about the impact on Granny's Chrome performance when she forgot to force-quit Pages is senseless.

The challenge for swap isn't too many browser tabs or unused applications that you haven't closed-- that's an access rate measured in human time. The challenge for swap is when a single application is processing a single dataset too large for memory and needs to constantly thrash to disk to access all of it-- this is an access rate measured in CPU time.

So let's not start to go too crazy on what the implications of the slower throughput of the 256GB SSD are. It's a pretty complex interaction of factors that drive performance-- throughput, latency, access patterns, free space in flash, erase times and user workflow. The tradeoffs are different for what you'd expect a 256GB drive and a 2TB to be used for.

 

[BornAgainMac]

I am glad this won't affect me. I never would get the base model.

 

[Ethosik]

It's so funny seeing the fanboys defend Apple like this trillion dollar CORPORATION is their family member. ⚰️

The fact is MOST people get base model Macbooks and the new "supercharged" base model should NOT have slower speeds in any aspect than the 2 year old model, especially considering they bumped up the price by $200.

Stop sucking up to a corporation. They care about their shareholders and not you.

~$1,100 laptops from competitors have way worse issues than a slower SSD.

 

[Lounge vibes 05]

It's so funny seeing the fanboys defend Apple like this trillion dollar CORPORATION is their family member. ⚰️

The fact is MOST people get base model Macbooks and the new "supercharged" base model should NOT have slower speeds in any aspect than the 2 year old model, especially considering they bumped up the price by $200.

Stop sucking up to a corporation. They care about their shareholders and not you.

I have not seen a single person "defend" this move.
I have not seen someone say apple was "right" for doing this or Apple "should've" done this.
I'm pretty sure everyone is in agreement that this isn't the best thing to happen to the MacBook Air, and it would've been "better" if they stuck to the original lay-out, or started at 512, or at least noted it in the tech specs.
What people *have* done is stated a fact, the fact being that, no matter how crappy this move was, 99.9% of customers will not notice or care.
The new design, WebCam, screen, etc is more important to the majority of customers than SSD speed, at least in their mind.

 

[CarlJ]

I’d think the color matched braided cables that don’t even come with the pro, or the high impedance headphone jack or something like that would’ve been the first thing to go before the fast SSD.

I would expect the cables and headphone jack come from quite differently equipped factories than the SSD’s (which are neither color matched or braided to my knowledge).

 

[compwiz1202]

What's so bizarre? It's not about the money they are saving on the chip, it's the money they are making from people upgrading to 512, and then the people upgrading to 14 pro because the price difference is so close. The pricing ladder is working as intended.

Exactly. I don't understand people who keep saying 256 is of course worse than 512 etc. No one debated that. The issue is they put a worse component of the same level into a newer version of the same line. And I wouldn't put it past the board of doing it on purpose to "force" people to upgrade.

 

[Ethosik]

One big caveat there..

If they buy a base model with only 8GB ram, it's absolutely easy for them to load it up and the swap comes into play...and that 50% slower than M1 SSD speed will be a bottleneck.

My wife has a base M1, and it's even been an occasional issue for her.

If you are that concerned about performance due to swap, that is evidence alone that you need more than 8GB of RAM.

 

[TheGenerous]

If it's not iJustine telling me this, I simply won't watch

 

[CarlJ]

Is there a way to block a channel on YouTube from showing you recommendations? Because I totally want to block this guy.

Click the “…” menu next to the video, then click “Don’t recommend this channel”.

 

[MaxYuryev]

How do know it will be noticed?

macOS will write the swap before its necessary so that when you run low of memory, the memory content has already been written.

Added:

The M2 MacBook Air 256GB SSD has a sequential write speed of 2260 MB/s when writing 1GB blocks (The Verge)
The M2 MacBook Air 512GB SSD has a sequential write speed of 2760 MB/s when writing 1GB blocks (The Verge)

So writing 1GB of swap till take 0.44 seconds vs 0.36 seconds.

Writing isn't much of an issue as it happens in the background, its when you have to read from the ssd and that happens as soon as you click onto a tab, or photo that you had opened in photoshop, or something else. The M1's had double the read speed. If you want to see how it can make a noticeable difference we have real-world test. Or check out Greg's gadgets latest video for file transfers with the M2 Air. a 35GB file took 14 seconds to transfer from an OWC thunderbolt 3 drive to a 512GB M2 Air, and 55 seconds to a 256GB M2 Air. That's a substantial difference.

 

[Ethosik]

Yeah ok but this is absolute basic computer usage. MacOS also has the dots to let you know it’s still running turned on by default. Also, Windows 11 just added those, and made it harder to open Task Manager. Right click icon -> close works the same on both.

Besides all that, it’s not universally true that hitting the X quits an app on Windows. Many applications use a notification area icon to stay running. Granted, most Windows users don’t know that either. Also, File->Quit is remarkably similar to Application Name -> Quit.

It’s always hard to say what “users know” because they’re all different.

My only point is that it’s not actually any harder, just different.

Yep. Teams, Slack, stores like Steam keep running in the background even if I close the window. Sometimes it even removes the indicator in the task bar leaving you to hunt in the notification area to see like you said.

 

[cupcakes2000]

This is the problem with people taking one benchmark, essentially a Blackmagic throughput test, and extrapolating all kinds of real world usage impacts.

The difference in throughput, for massive sequential read accesses is roughly 1.5GB/s versus 3GB/s. There's less of a difference for writes, which are in the background anyway for swapping. That means that if you're reading a contiguous 1GB, it'll take 0.3 seconds longer to change tabs.

Let's count to 0.3 together, shall we? Ready? One, two... dang, let's try that again...

But how much is actually being swapped because you have a tab or unused application open? If I look at my Activity Monitor, I see a rare few that might be as much as 300MB. So it would take 0.1 seconds longer to read that back from swap.

Except that's not how the virtual memory works on MacOS-- long before the system starts paging to disk, it starts compressing the memory and then, when it has to, it pages the compressed memory to disk. When I look at the compressed VM size in Activity Monitor, the biggest I'm seeing is on the order of 70-80MB. So about 25 milliseconds different.

For reference, with a 60Hz display, the screen refresh time is 16 milliseconds. How often are your non-geek users switching tabs and apps that they're noticing the added disk access time is slowing down their day?

Of course, the VM doesn't page out the entire memory space at once, it does it 16kB at a time based on the algorithm's estimate of which memory is least likely to be accessed. So how much of the above analysis is even relevant? What do we know about small random accesses which are dominated by latency rather than throughput? Nothing that I'm aware of.

And that's the point. Using a throughput benchmark to make generalizations about the impact on Granny's Chrome performance when she forgot to force-quite Pages is senseless.

The challenge for swap isn't too many browser tabs or unused applications that you haven't closed-- that's an access rate measured in human time. The challenge for swap is when a single application is processing a single dataset too large for memory and needs to constantly thrash to disk to access all of it-- this is an access rate measured in CPU time.

So let's not start to go too crazy on what the implications of the slower throughput of the 256GB SSD are. It's a pretty complex interaction of factors that drive performance-- throughput, latency, access patterns, free space in flash, erase times and user workflow. The tradeoffs are different for what you'd expect a 256GB drive and a 2TB to be used for.

Come on! You know speaking sense doesn’t matter in this forum. You fan boy Apple apologist, you.

 

[Ethosik]

Thing is, that’s not always true on Windows either. It’s true you shouldn’t need to use Task Manager to quit applications often, but you shouldn’t have to use Force Quit often either. I was just pointing out another option.

Also since Vista there were memory improvements like Superfetch. First time opening Photoshop after a cold boot takes about 10 times as long as if I close it and come back an hour later (if my RAM has been low enough to keep it it memory). It’s nearly instant the second time.

 

[3xBoom]

Jane's Addiction "Nothing's Shocking" comes to mind... MaxTech is blowing smoke as per usual

 

[nylon]

Bought base model for my Mum who only uses for general productivity, browsing and media consumption. I think the SSD speed issue will only affect those whose computing needs exceed basic generic usage.

 

[turbineseaplane]

Bought base model for my Mum who only uses for general productivity, browsing and media consumption. I think the SSD speed issue will only affect those whose computing needs exceed basic generic usage.

No 16gb Ram puts the SSD issue in play for even basic users due to SSD swapping

 

[cupcakes2000]

No 16gb Ram puts the SSD issue in play for even basic users due to SSD swapping

By absolutely no noticeable metric for such a use. Stop spreading fud.

 

[Lounge vibes 05]

I would expect the cables and headphone jack come from quite differently equipped factories than the SSD’s (which are neither color matched or braided to my knowledge).

They are, I have one.
The Midnight MacBook comes with a braided Blue cable, the gray comes with a gray, etc.
And that literally is my point.
If supply constraints truly were not a factor, and Apple is simply just using slower SSDs to put more coins in Timmy's pockets, there is plenty of other things they could've left out to save money, like the colored braided cables and the better jack

 

[ApplesAreSweet&Sour]

Exactly. I don't understand people who keep saying 256 is of course worse than 512 etc. No one debated that. The issue is they put a worse component of the same level into a newer version of the same line. And I wouldn't put it past the board of doing it on purpose to "force" people to upgrade.

I agree. But what makes this particularly terrible is that there's no way you'd figure this out if you go to purchase a new 8/256GB M2 machine, which happens to be the configuration that will be most readily available and the config that will see most discounts in stores.

The fact that Apple's product pages only make statements about more expensive configurations is also not an argument. The omission of information on the 256GB SSDs does not equate to transparent info that any consumer can readily access.

This is the second generation of a product -Every key spec has to be either the same or better. That's how it's always been.

 

[lesalps]

This is the problem with people taking one benchmark, essentially a Blackmagic throughput test, and extrapolating all kinds of real world usage impacts.
...[snip]...
So let's not start to go too crazy on what the implications of the slower throughput of the 256GB SSD are.

Are you sure you are in the correct forum, on the correct internet?

What you said is

• reasoned
• informed
• well articulated
• possibly even true

That's simply not how things are done around here.

I think you may have logged into the wrong site by mistake?

 

[turbineseaplane]

.

 

[ApplesAreSweet&Sour]

They are, I have one.
The Midnight MacBook comes with a braided Blue cable, the gray comes with a gray, etc.
And that literally is my point.
If supply constraints truly were not a factor, and Apple is simply just using slower SSDs to put more coins in Timmy's pockets, there is plenty of other things they could've left out to save money, like the colored braided cables and the better jack

Obviously, Apple isn't going to leave out the comparably inexpensive and immediately noticeable features, like braided cables and a better headphone jack.

There's a lot to save by switching SSDs and next to nobody is opening up their MacBooks to check the parts.

The harddrive is hidden, all the exterior aesthetics, keyboard, display, color options, cables and ports are not.

 

[issacj]

No, because there would be legal repercussions of doing that.

Apple would need to update their M1/MBA press release which specifically states "up to 2X" faster storage performance. Apple can't simply delete a CPU core or reduce SSD performance without taking other action.

Besides, the M1 MBA is widely available across the world.

The "up to" is quite vague, though, right? I found the original press release and bizarrely, under footnote 1 (which follows a paragraph about the MBA) it says this was according to tests done with a pre-production MacBook Pro.

Similarly, on this page on the website, claims are based on "pre-production MacBook Air systems with Apple M1 chip and 8-core GPU, configured with 8GB of RAM and 512GB SSD."

I can't see why any of that would stop Apple from changing the SSD setup in the 256GB model. A MBA M1 configured with 512GB would still be "up to" twice as fast.

But yeah, full disclosure, I don't have legal training and it was just a thought that anyone buying an M1 today based on YouTube videos saying the M1 MBA has faster storage speeds might be disappointed.

 

[lesalps]

I am just enjoying that the entry level Air is again crap.

Which is as things should be.

And have been since 2008.

So that power users like me can rest assured our Macbook Pros are a justified purchase.

With 5GB/s read and write on the SSD.

Which is very important as I use it a lot for Macrumors.

Sometimes in 2 tabs at once.