M1 8GB or 16GB?

Anecdotal evidence incoming: I’m an academic and spend most of my time on my M1 MBA reading journal articles, writing articles, compiling statistical data and browsing, along with some media (music and the occasional podcast when not in Deep Work mode). Not very intensive uses of the M1 architecture, to be sure. So in my use case scenario, the 8GB of RAM has been more than sufficient and I have had nothing but good experiences with my device so far. YMMV.

1 Like

I don’t, think excess capability is ever regretted. This is the only reason why I have my 2009 iMac running - increased its memory to a maximum of 16 gb. I am hoping my 16 gb M1 Mac mini will last as long.


I had the 8Gb m1 mbair for a while. It was quite wonderful for most of the day, until it wasn’t which is when it became utterly slow.
I’d notice this at the end of the work day. I generally have a couple of browsers with several Tabs open along with quite a bit of text - markdown editors, note apps, spreadsheets, spotlight, launchbar and the like.
My current laptop is a 16 inch with 6 cores 16gb ram and 1 tb ssd with 160gb to 200gb of free space. This will also towards the end of the day start slowing down but not as much as the m1 air.
So in summary - the m1 air was much better than the 16inch for most of the day until the m1 became very bad. Past that point the 16inch has been better.

I’ve now placed an order for an m1 air with 16gb and 1tb ssd (previous one had 512gbb) and yeah it’s about a month to get it delivered.

I really only have the larger laptop because I’ve tended to get screen claustrophobic with smaller screens but the delight of using the m1 will hopefully fix that. Also the lack of fan noise was amazing.

Haven’t tried the new mini. I’m a little intrigued. Maybe get a 8Gb with the smallest ssd and use my external ssd as an adjunct with the hope of that ssd’s 480mbps transfer speed not affecting the experience negatively too much.


I maxed out my 15" MBP, buying the most expensive computer I’d ever bought and had keyboard issues, battery problems and now the display died. That’s nothing to do with the fact it was specced up, but it makes the problem an expensive one. It would have cost a bit more than a base model M1 MacBook Air to fix, so I held off and I have it hooked to a TV for occasional gaming.
So, do I regret maxing it out?
Very much so. The performance was great, but I stretched my budget and bought a dud. It was a very good dud for a while :smiley:

Obviously I don’t expect that kind of bad luck this time, but $200 is $200 that is not being spent on other things, or in savings. 16GB of RAM is twice as many resources being mined etc as 8GB.

I say that if you know you need/want to spec up the computer then do it because having one computer last you 8 years, and getting all that joy out of it, is better than it lasting less time. But likewise, if you know a lesser model is enough, then make do for those 8 years and enjoy the sufficiency.

And to the OP, a wait time of 1 month is very little in the scheme of things. I had to fight myself to stop from buying a standard 512GB SSD right now, instead going up to 1TB. I know I’ll get use out of that, so 1 month to wait will get me a more satisfying machine in the end.

1 Like

I’m intrigued by your problem of the computers slowing down during the day.
Do you know what is causing it as I have only intermittently noticed it on a Mac; even my 8GB, 128GB little MBP doesn’t have that sort of issue with what I do.


It’s easy to just tell everyone to get 16GB, but $200 is not a small price; it’s 15-22% of the cost of the base laptop (ranging from education Air to non-education Pro.) Similarly to AppleCare, the extra cost needs to be weighed against the utility and the risk that something bad will happen without it. Excess capacity is never regretted only by people who have excess money.

Incidentally, there’s a piece of advice that a PhD friend of mine gave me. I was asking about some science-y stuff related to medicine, supplements, etc. and she said:

“Google whatever you’re looking for, and add the words ‘method of action’. That’ll help you sort out the legit info from the random stuff people claim.”

I’m inclined to think that sort of advice applies to the 8 GB vs 16 GB question.

I’ve heard a lot of people say that the 8 GB is fine, and that “8 is the new 16”, and that sort of thing - but I (so far - maybe I’m reading the wrong sources?) haven’t seen anybody explaining why or how the M1 could conceivably do things just as well with half the RAM.

There are lots of potential partial explanations. A faster processor, by itself, would decrease RAM usage somewhat as any application that runs and then quits would do so in less time, reducing memory pressure. Increased SSD speeds would decrease the cost of using virtual memory, which would help significantly. There are lots of little things like that.

But those are partial explanations. At a certain point, if you’re loading data into memory that you’re working on in memory (video, audio, etc.), you need the RAM to store that stuff, don’t you?


This is my favourite dive into it by Erik Engheim.

The gist is that unified memory is not the same as shared memory, which is what I remember from years past. It allows all parts of the system on the chip to access data without shuttling said data around. That’s a massive speed saving.

Swapping is still going to be needed, but less than before. Anyway, in this architecture, with today’s speedy SSDs, maybe swapping is not the enemy it once was. The SSD, to some extent, is just slower memory now. My 1TB SSB at 2400 MB/s is a tad faster than the old DDR 266 memory modules I think. DDR4 is 10x faster and probably has other benefits to do with latency, but hey, that seems pretty incredible.

For me, the great unknown is what software will look like in 5 years time. Currently, nothing seems to really be testing the M1 hardware. Once developers get used to the new system, some are sure to push it to its limits. The big, big question is, will I care about those programs or will they be niche/super high end?
No one knows, so early adopters are risk takers, as always.

Here’s another thought, that I considered before pulling the trigger: what will the next MacBook Air look like? Will the hardware improvements (new tech such as touch screen, not just faster chips) make the first models feel obsolete too early?

1 Like

It’s more that many people didn’t need 16GB on Intel either. :wink:


Oh how I would love to find out the reason! Every once a while I’ll stumble upon something that would could should resolve such issues but no long standing luck so far. My latest anecdotal hunch is that there’s something going on with Spotlight and indexing and this even affects launchbar but I really don’t know.

WindowServer is often one of the highest cpu hogs. Safari often gets to a point where I have to do more than close all tabs to alleviate issues.

I’m also suspecting the issues have something to do with my iCloud or Apple ID account. I never seem to get the battery others claim to get from any of my Apple devices.

1 Like

It can’t. Or, more accurately, there are workloads that require > 8GB of RAM that cannot be accomplished, or will incur unacceptable performance penalties on machines with less than the RAM that they require. I sometimes run multiple VMs that can use 32GB of RAM combined; that would be a painful experience on an 8GB (or even a 16GB) machine.

That being said, for combined workloads that might occasionally exceed the physical RAM on a system, but don’t need to be in RAM at the same time, the architecture of the system and speed of its components can do a lot to hide the fact that the RAM is running low. By all accounts the M1 based Macs are doing a very good job of that. It may be that some people who found 8GB too performance limiting on Intel Macs would be okay with that on M1 Macs. I do remain sceptical that cutting back on RAM (going to a lesser amount on the new vs the previous computer) is a good idea unless one never was hitting swap with the greater amount.

This is also a very good point, especially for the class of lower-end/mass consumer computers that the current crop of M1 Macs belong to.


I think that if you have the doubt and you are not on a strict budget, purchase the 16 GB. There is a chance you will be fine with 8 GB but probably you will be anxious all the time for not having 16 GB. My advice on this kind of decisions is that if your budget allows, go get the best you can and enjoy it. I think it is better to feel anxious that you spent more than you need, than feel this way because you do not have the device you really wish. If you have a limited budget it looks like most people is still very impressed with the results of the 8 GB M1 so it is really not that bad.


I bought an 8gb in March and took a £150 hit to replace it with an M1. It’s second machine and I’ll be replacing my iMac at some point and would rather spend the cash on extra memory for that.

You should get the 16gb if you can but …

Whilst I’m not saying 8gb is the new 16gb, I will observe that I found the Intel 8gb quite sluggish. By contrast I find the M1 to be much zippier for the same tasks.


This is the question that I have so far failed to find an answer to.
Has anyone actually demonstrated a real world situation that the 8GB machine cannot handle?

The common sense assumption would be that if someone wanted to work on an image or a virtual machine that was more than 8GB in size then the limited RAM would be an issue.
So far I have not seen anyone demonstrate this to be the case in reality.

What I have seen, and I’d forgotten about, is that exporting video or doing long Xcode builds is a lot quicker with 16GB. My current assessment from others’ experiences is that there seems to be close to zero difference in usage, but 20-200% difference in the speed of sustained loads. If you do a lot of that, 16GB seems worthwhile but, oddly, not a necessity.

I remember the days when hard drive prices of $1 per megabyte were an absolute steal, and looking at 250 megabyte hard drives and wondering what on earth somebody would ever be able to do with all that space. :smiley:

I think it’s a safe bet that software will expand to avail itself of the available hardware. :slight_smile:


I feel like the people for whom the difference may be significant are either (a) not ordering the 8 GB (and thus not a useful data point), or (b) holding off until early 2021 in the hopes that the M(2?) is on the way.

1 Like

My 2011 MBP had 8gb and ran Mac app fine but when I loaded Parallels and Windows, paging numbers went through the roof. Upgraded to 16 and had almost no paging. The 2015 MBA I have now only has 8gb but the memory pressure has never gotten out of the green. Helps that I don’t run Parallels anymore. When I upgrade to a M1 or successor in the future I will be very tempted to get 16gb because the incremental cost is significantly less than the cost to upgrade if I find I need more memory. However if cost is a significant constraint I will opt for a larger SSD over more memory.


I mentioned it: Running multiple virtual machines. My real world is different from your real world though, so this isn’t likely something that you’d run across.

When I’m travelling it’s nice to have a machine with me that I can throw some of that work at, should the need arise. That’s why I ordered my Air with 16GB but it’s not an attempt to convince you that you need it :slight_smile:

(I’m taking a chance though, as neither of the major virtualization solutions exist yet for the M1 Macs. I assume that they’ll be ready by the time I’m ready to travel again)

I have every confidence that in when the lockdown finally ends in 2023 or so, Parallels or VMWare will have released updates for M1 / Big Sur. :smiley:


I haven’t seen any evidence this is actually a problem with the current architecture, though, that’s my point.
If this mattered to me then of course I’d wait and see, or get the headroom; like you said everyone’s real world is different. I’d certainly get 16GB if I was doing what you’re doing with it and buying now, just in case.

But at the moment I think it’s still conjecture that it won’t work well.