Should I go for the Macbook air m1 or Macbook air m2?

[Heriotic08]

I am currently quite confused as I am having a difficult time choosing over MacBook air m1 and MacBook air m2. The main problem is that the base MacBook air m2 only comes with one 256GB nand chip, and I don't have the money to go for the 512 GB model. I am planning to use my MacBook for mainly coding purposes, including running some ml models. Although m2 is faster than m1, I speculate that it would hinder the performance if I'm running a heavy memory task like an ml project, and hence I'd pay for the same, if not worse, performance. The new MacBook Air has got some new chassis and stuff, but it might not be that good of incentive to pay additional 300ish dollars (as MacBook air m1 is discounted). Can anyone give any suggestions and some graphs pointing out what would be a real-world difference when it comes to coding as I couldn't find some relevant data?

[hishnash]

I would suggest it would be worth getting the M1 (you can get it quite a bit cheaper) with a 512 SSD with 16GB of ram rather than getting the base model M2 (might well be able to get this at around the same price as the entry level M1).  If your working with ML you will find yourselves very very quickly filling up your disk these things can get very large very quick, and if your doing dev work depending on what your working on you might well find the resource files etc rather large.  256 GB from a size perspective will end up being your biggest concern.  

Coding really depends on what you will be writing for and what target you want it to run.  If your doing anything that is web based your very likly going to want to have a few docker containers running in a VM so getting an M1 and using the savings to get more RAM would be worth it.

 

For ML tasks you are correct the low memory will be an issue but also the ssd size will be an issue so I would again suggest getting a good deal on an M1 if you can with 512 GB and 16GB of ram rather than base model.  Depending on were you are in the world you can find good deal on these with resellers like amazon etc or on the apple refurbished store. 

 

If you cant find a good deal of a 512 GB ssd + 16 GB ram get the 16 GB M1 since not needing to swap will have a massive impact you can always get a fast external USBc/TB drive later.  Amazon are shipping a 16GB RAM, 256GB SSD for $1,185  https://www.amazon.com/Apple-MacBook-13-inch-256GB-Storage/dp/B08QMB94YW

Also depending on were you are in the world you might be able to get educationally discount (you might have a friend/family member in education or working in education) if you order from apple directly.  In the US with education pricing 16 GB ram + 512 SSD is 1,259.00. Or you can save $180 and get the 256 GB SSD and spend that $180 on a large fast external TB3 drive to add extra storage.

[saintlouisbagels]

I would easily recommend buying used on eBay. You’ll get a great deal, and if someone sends you lemon then there’ll be buyer protection for you to get your money back.

 

Edit: It’s really not worth looking at the M2 models. It’s a minor performance upgrade for higher thermals, the SSD’s are worse, and they’re still Thunderbolt 3. Find a 16GB, 512GB M1 Air (or M1 MBP!) on the used market and be happy

Edited July 21, 2022 by saint_louis_bagels

[Imbadatnames]

48 minutes ago, Heriotic08 said:

I am currently quite confused as I am having a difficult time choosing over MacBook air m1 and MacBook air m2. The main problem is that the base MacBook air m2 only comes with one 256GB nand chip, and I don't have the money to go for the 512 GB model. I am planning to use my MacBook for mainly coding purposes, including running some ml models. Although m2 is faster than m1, I speculate that it would hinder the performance if I'm running a heavy memory task like an ml project, and hence I'd pay for the same, if not worse, performance. The new MacBook Air has got some new chassis and stuff, but it might not be that good of incentive to pay additional 300ish dollars (as MacBook air m1 is discounted). Can anyone give any suggestions and some graphs pointing out what would be a real-world difference when it comes to coding as I couldn't find some relevant data?

It depends. You are extremely unlikely to see an issue with the memory speed, it’s rather over blown and the M2 is actually faster in random reads. It’s still around 2GB/s too in sequential. 
 

Personally MagSafe would push me to the M2 along with the better keyboard (with full height function keys) and screen. 
 

 

[Heriotic08]

53 minutes ago, hishnash said:

I would suggest it would be worth getting the M1 (you can get it quite a bit cheaper) with a 512 SSD with 16GB of ram rather than getting the base model M2 (might well be able to get this at around the same price as the entry level M1).  If your working with ML you will find yourselves very very quickly filling up your disk these things can get very large very quick, and if your doing dev work depending on what your working on you might well find the resource files etc rather large.  256 GB from a size perspective will end up being your biggest concern.  

Coding really depends on what you will be writing for and what target you want it to run.  If your doing anything that is web based your very likly going to want to have a few docker containers running in a VM so getting an M1 and using the savings to get more RAM would be worth it.

 

For ML tasks you are correct the low memory will be an issue but also the ssd size will be an issue so I would again suggest getting a good deal on an M1 if you can with 512 GB and 16GB of ram rather than base model.  Depending on were you are in the world you can find good deal on these with resellers like amazon etc or on the apple refurbished store. 

 

If you cant find a good deal of a 512 GB ssd + 16 GB ram get the 16 GB M1 since not needing to swap will have a massive impact you can always get a fast external USBc/TB drive later.  Amazon are shipping a 16GB RAM, 256GB SSD for $1,185  https://www.amazon.com/Apple-MacBook-13-inch-256GB-Storage/dp/B08QMB94YW

Also depending on were you are in the world you might be able to get educationally discount (you might have a friend/family member in education or working in education) if you order from apple directly.  In the US with education pricing 16 GB ram + 512 SSD is 1,259.00. Or you can save $180 and get the 256 GB SSD and spend that $180 on a large fast external TB3 drive to add extra storage.

Yeah, but the main concern would be how much performance difference would I see with the new faster GPU as that'd be a game changer as ML (with the conflicting memory speed) would depend heavily on that. I also wanted to ask if I buy an external fast drive, will the bottleneck still be an issue? Your points are quite helpful, the 16gb ram looks enticing. Thank you for that!

[Heriotic08]

55 minutes ago, saint_louis_bagels said:

I would easily recommend buying used on eBay. You’ll get a great deal, and if someone sends you lemon then there’ll be buyer protection for you to get your money back.

 

Edit: It’s really not worth looking at the M2 models. It’s a minor performance upgrade for higher thermals, the SSD’s are worse, and they’re still Thunderbolt 3. Find a 16GB, 512GB M1 Air (or M1 MBP!) on the used market and be happy

Yeah, that is fair but I have never bought anything from eBay, so am a bit skeptical, plus it might not come with any warranty (not does apple provide the best warranty services, ig). I guess the only factor now to consider is the performance difference due to the GPU. Thank you!

[Heriotic08]

39 minutes ago, Imbadatnames said:

It depends. You are extremely unlikely to see an issue with the memory speed, it’s rather over blown and the M2 is actually faster in random reads. It’s still around 2GB/s too in sequential. 
 

Personally MagSafe would push me to the M2 along with the better keyboard (with full height function keys) and screen. 
 

 

It might be true, but read/write speed tests have shown that ssd speed of the m2 base model is 50% slower. The additional features are great, but I guess, I have to make some compromises whether it be with GPU performance or SSD speed and ram size. Thank you, for the suggestion!

[Imbadatnames]

15 minutes ago, Heriotic08 said:

Yeah, but the main concern would be how much performance difference would I see with the new faster GPU as that'd be a game changer as ML (with the conflicting memory speed) would depend heavily on that. I also wanted to ask if I buy an external fast drive, will the bottleneck still be an issue? Your points are quite helpful, the 16gb ram looks enticing. Thank you for that!

You should be fine and I believe the M2 has updated ML

[hishnash]

26 minutes ago, Heriotic08 said:

would depend heavily on that. I also wanted to ask if I buy an external fast drive, will the bottleneck still be an issue? Your points are quite helpful, the 16gb ram looks enticing. Thank you for that!

So for ML tasks your going to need more than 8GB of ram so you will end up swapping to disk, that will be the internal disk so you will get the performance impact hit as the swap really starts to chug against the SSD.  
 

11 minutes ago, Imbadatnames said:

You should be fine and I believe the M2 has updated ML

For training yes the GPU is faster but only if the data fits in RAM. 

 

24 minutes ago, Heriotic08 said:

plus it might not come with any warranty (not does apple provide the best warranty services, ig)

With respect to apples warranty this is attached to the device apple do not care if you purchased it second hand if it still has a warranty then it moves over they just care about the serial number. https://checkcoverage.apple.com/ if you can get them to share the serial number with you before purchase then you can validate how much coverage it still has. It is also possible (in some regions of the world) to extend this (as long as it has not expired) if you want a longer coverage. 
 

 

[saintlouisbagels]

7 hours ago, Heriotic08 said:

It might be true, but read/write speed tests have shown that ssd speed of the m2 base model is 50% slower. The additional features are great, but I guess, I have to make some compromises whether it be with GPU performance or SSD speed and ram size. Thank you, for the suggestion!

Just keep in mind that the extra GPU speed won't be much a difference due to thermal throttling. I don't know how intensive ML is, but if you're running intensive programs for long hours, it'll do you better to have the better SSD for swap, than to have a GPU that will throttle during runtimes.

[Just that Mario]

12 hours ago, hishnash said:

 If your doing anything that is web based your very likly going to want to have a few docker containers running in a VM so getting an M1 and using the savings to get more RAM would be worth it.

What? Why?

I never understand why people on forums ALWAYS claim you will be running multiple VMs and such. In real life you will not be doing anything like that on a mobile workstation or at all. Most usually you'll just be running a browser, code editor of your choice and then whatever is required to test your output. It also seems OP is just starting some kind of education in the field, which means they won't be doing anything crazy anyways. Probably picked up few keywords such as machine learning from teacher and expects to be writing next Google Bot or any other AI, when in reality they'll probably have 30 minute look at what ML fundamentally is and then move on.

[Obioban]

The drive speed and thermals are just internet FUD, not real issues, for all but very specific/isolated use cases. I wouldn't pay any attention them.

 

That said, 8gb is pretty tight for all but pretty casual use-- if you're going to have this for a while, I'd get 16gb.

 

And, 256gb of store is pretty tight in the modern era from a straight up storage perspective. I'd think very carefully about if 256gb is going to be sufficient to meet your space needs for however long you're going to own this, since you can't upgrade later.

 

If those push the M2 out of your budget, the M1 is still a spectacular machine.

 

Most of the ML stuff on Apple silicon runs on the ML cores, not the GPU cores-- so you might be over emphasizing how important those will be to you.

 

The M2 is a decade newer chassis design, so, if you can increase your budget a bit.. I don't think you'd regret it.

 

If you're doing this for school, don't forget to get educational pricing, whatever route you go.

[hishnash]

2 hours ago, Just that Mario said:

I never understand why people on forums ALWAYS claim you will be running multiple VMs and such. In real life you will not be doing anything like that on a mobile workstation or at all. Most usually you'll just be running a browser, code editor of your choice and then whatever is required to test your output.

depends on what sort of dev work you are doing, sure if your just doing frontend JS dev you might note be using a docker container but even then having the ability to run the backend as close to prod is very useful, if your doing backend dev going to docker pathways for database, caches etc is just so much easier than configuring all of these things and then dealing with multiple conflicting versions of them for each project.  

 

2 hours ago, Just that Mario said:

It also seems OP is just starting some kind of education in the field, which means they won't be doing anything crazy anyways.

Yer if the use case is just some software dev for edu then specs are not really to important ad your right the project size and complexity will be kept to a minimum.
 
 

[wolverine104]

M1 Macbook air is a fantastic machine. M2 is mostly a chassis upgrade. 

 

While there are definite performance upgrades you can have just by getting the base model M2, the M1 with some small upgrades is going to be a killer machine for a while /