I spent 2 months on a base M2 Mac Mini

[Allen_Wentz]

That's a non-issue as neither Apple is trying to gain market shares of the PC(Windows) gaming and gaming console demographic, nor is the average PC(Windows) or gaming console consumer in any meaningful way incentivized to switch away from their abundant AAA game libraries and infinitely customizable hardware setups over to Apple's which, in most ways, is almost the exact opposite of that.

Gaming on Mac will never take off no matter how well current of future Mac hardware can run high-end, AAA games.

The hardware is practically inaccessible to the largest parts of the gaming demographic, kids, teens, and young adults, simply because of Mac hardware pricing.

Additionally, the bulk of Apple's gaming library of mobile and very limited selection of AAA titles can already be played on devices that consumers are far more likely to already own or upgrade frequently than desktop computers, namely iPhones and iPads.

Apple would need a library of highly attractive AAA games, based on known and popular ips, that's exclusive to Macs and only runs on Macs, not iPhones, iPads, AVP, etc.

Apple is never really moving into gaming and thus 8GB is fine for many years to come.

Your post was well thought out until the last few words: "...8GB is fine for many years to come" which is just wrong.

 

[Allen_Wentz]

I do agree. I think it's the "it's fine" aspect of it all that gets people riled up, compounded with the extortionate upgrade costs. People don't spent $1,600 (plus tax) for specs that are merely fine (current base MBP). The processors are lightning fast, the devices are impeccably machined, the screens are generally excellent, the keyboards are great, the trackpads are brilliant, yet the storage and RAM are "meh" as you say.

Allowing customers leeway to use more RAM heavy software 3 years from now (I'm looking at you, LLMs), or play RAM intense games, or allowing customers space to save said games and videos should they decide to, without fear of running out of storage, would be great. It's Apple penny pinching, which is a bit naughty for premium gear and dirt cheap aspects of them.

You are simply whining about wanting more for less. <sigh>

 

[ric22]

1) superb joke.
2) "The ONLY reason the base spec exists is to NUDGE you into upsells." If the base only exists to force an upsell, then why do Apple's retail partners not typically offer the upgraded units? Sorry, that just doesn't make sense.

They get bigger discounts from Apple on base models than BTO models. That's why they can frequently put the base models on sale.

 

[Allen_Wentz]

If it wasn’t the right fit for the vast majority (and not “everyone”) then they (Apple) would see it in returns and poor feedback. Surely, right? This forum would also be filled with disgruntled purchasers who cannot get their basics done in an acceptable timeframe. But it’s not.

Are we seeing that? I don’t think we are.

No. The fact that the vast majority do not return their purchases simply means that the base RAM works well enough for the first year or two. It does not mean that base RAM was an optimum life cycle purchase decision. That important truism is what many fail to grasp.

Spending $thousands on a computer to compute with, computing sub-optimally for much of the life cycle of the box for wont of +$400 for +32 GB RAM usually represents bad planning.

 

[ric22]

You are simply whining about wanting more for less. <sigh>

I'm not. Why are you accusing me of whining? That's not a very intelligent way of disagreeing with some part of my post.

Anyone that doesn't feel a bit peeved that Apple charge more for an 8GB upgrade than other companies charge for complete laptops that come with 8GB of RAM is an odd person I'd argue.

 

[LevorgPenmancho]

The eternal RAM chat reignites.
Base model RAM will always be just enough for normal use cases. If you have a use case for more, buy more. If you don’t have a use case for more, and you’re going to change your machine in 2 years to keep with the trends, don’t buy more. If you want to run the machine for 10 years, pay for more. I’m still running a MB Air 11inch 2013 with 8GB. It’s fine.

 

[Blair110]

No. The fact that the vast majority do not return their purchases simply means that the base RAM works well enough for the first year or two. It does not mean that base RAM was an optimum life cycle purchase decision. That important truism is what many fail to grasp.

Spending $thousands on a computer to compute with, computing sub-optimally for much of the life cycle of the box for wont of +$400 for +32 GB RAM usually represents bad planning.

Actually, you CAN'T make the assumption that it means whether the 8GB only works well enough for the first year or two. Everyone's use cases are different, and everyone's definition of how well their computer works differs also. It may not be optimal for YOU but it does not mean it's not optimal for everyone else.

My father purchased a base M1 Mini when it first came out. He uses it to surf, email, and light gaming. It's now years 3 1/2 years later, and he's never had an issue. Thing runs great. He'll probably get another 2-3 years for his use case before it even starts to slow down. There are tons of users out there that have similar use cases, and I'd argue that $500 for 5-6 + years of quality computing is pretty darn good.

 

[Allen_Wentz]

That's mostly R being a steamer that is. The datasets themselves don't usually have to sit in RAM. R does like to parse, copy, copy, copy, forget to GC etc all the time.

Anything which requires any heavy lifting usually gets tested in R on small datasets then ported to Go and my own toolkit which is somewhat more memory efficient. Basically uses mmap and a dynamic reader then streaming aggregates instead of parsing the entire thing into tibbles in RAM and then reading out of there. I can quite happily work on 1TB sized datasets and keep under 50-70MB of RAM. This is required because we deal with thousands of datasets that large on a daily basis and we want to keep costs very very low.

So just because it's there doesn't mean you need it and doesn't mean you need to want to pay for it...

Edit: to add that framework was mostly developed while SSH'ed into an AWS t2.small instance (2 vCPU / 2GB RAM) with a 2TB EBS disk attached to it.

I disagree. You suggest modifying workflow to conserve on RAM usage, which is IMO very unproductive when more RAM can be available. In the old days we did all kinds of tricks to cope with unavoidable RAM limits, but today it is far smarter to spend $800 for +64 GB of RAM and just let your brain be creative at work product instead of being creative at how to waste brain cycles fitting workflow under less than optimal RAM

 

[Howard2k]

No. The fact that the vast majority do not return their purchases simply means that the base RAM works well enough for the first year or two. It does not mean that base RAM was an optimum life cycle purchase decision. That important truism is what many fail to grasp.

Spending $thousands on a computer to compute with, computing sub-optimally for much of the life cycle of the box for wont of +$400 for +32 GB RAM usually represents bad planning.

Then where are all of the posts from people saying that their machines only work for the first year or two?

 

[Howard2k]

They get bigger discounts from Apple on base models than BTO models. That's why they can frequently put the base models on sale.

But again, if the only reason the base model exists is to force an upgrade, then why are they pushing the base model through the retail channel and not the upgrades? If the base model doesn't work for the majority, why would they lead their entire retail channel with that?

 

[Closingracer]

The project is over.

About 2 months ago now, a friend of mine sold me a base 8/256 Mac Mini M2 for a stupid low price which I couldn't turn down. It was his first ARM Mac after his old 16" Intel lap burner gave up. He bought an M3 iMac to replace the mini with so wanted to dump it off. I wasn't going to say no because I can sell it for more than I paid for it anyway 🤣

So my daily driver machine before that was a 2021 14" MBP with M1Pro + 16Gb RAM and 512Gb disk. This has been an absolutely amazing machine but I thought I'd see if I can get away with using a base thing should the day come I can't afford to keep spending loads of money on Apple crap. I've already got a Studio Display, magic keyboard and logitech MX Master so I just hooked them up to the M2 and used it.

Anyway findings so far. My use cases are fairly wide. I use the usual Apple apps (Mail, messages, safari, maps, photos, calendar, contacts, numbers, reminders, notes, apple music, weather, chess etc) for most things. On top of that I have some more specialist workloads. I wanted rid of Adobe from my existence from the photography side of things so I actually moved my stuff out of Lightroom and into Apple Photos and Pixelmator. This is perfectly adequate for my light editing needs from my phone and mirrorless camera. On top I do mathematical typesetting with LaTeX via TeXshop, fairly heavy stuff in R with RStudio, Maxima CAS, various bits of work on AWS from the Terminal and RDP into a windows box occasionally. The datasets I'm working with in R are around 1Gb a piece and it had no problem with that. I may transcode a video or two for my iPad so I can watch stuff somewhere else as well on top of this.

Conclusions:

• At a high level I can't actually tell the difference between the machines. They feel completely identical.
• There's probably a 5-10% gain in Handbrake if I use the VideoToolbox H.265 codec compared to the M1 Pro.
• I didn't really see or notice any memory pressure issues. Had a couple of yellow graphs but quite frankly I'm paying for the RAM so using it is a win actually using it.
• The storage bandwidth is a lot less but I honestly couldn't tell.
• Connecting bluetooth devices to it initially is a pain in the ass and I had to shove a PC keyboard and mouse into it.
• It was utterly boringly reliable and I had no issues at all.
• Good enough is good enough!

Alas I need to go back to the MBP now as the primary machine as I need to actually do some stuff while travelling but that is the only reason. I don't think that the memory paranoia resulting in moving to 16Gb is valid for most users. I mean I'd like 16Gb as a baseline but meh, it's fine.

Well that was boring. Sorry if you read to here and are asleep.

8GB is fine but it's the principle. For $1099 for the Base model M2 Air it should have 16GB of memory. Just because it's fine now doesn't mean it shouldn't have 16GB min. Apple knows they will make good money on people getting the upgrade. Myself a prime example with all the options I ended up with a Base M3 Pro MacBook Pro. I figured for $1,849 I get the better screen, 2GB of more ram than just 16GB and active cooling. I went from hmm upgrade the ram and storage of the M2 Air to hmm the M3 base MBP looks nice with the nice screen and active cooling to the M3 Pro MBP is better and I get the headroom plus extra USB Type C and Space Black for $200 more.

 

[Closingracer]

I don't think anyone who's interested in the kind of gaming you're talking about is going to buy an 8GB anything.

Lol yes. I have 32GB on my Gaming desktop... Wouldn't think of running 8GB in single channel on windows.

 

[Closingracer]

If it wasn’t the right fit for the vast majority (and not “everyone”) then they (Apple) would see it in returns and poor feedback. Surely, right? This forum would also be filled with disgruntled purchasers who cannot get their basics done in an acceptable timeframe. But it’s not.

Are we seeing that? I don’t think we are.

Most people just don't go to tech forums. Also most people don't thin that far ahead they just see the people saying 8Gb ifs fine you'll be ok people and going with it. Will the 8GB memory be ok for a few years ? yes. But I am thinking of having this laptop for a long time and 8GB isn't where it's at down the line.

 

[Allen_Wentz]

Actually, you CAN'T make the assumption that it means whether the 8GB only works well enough for the first year or two. Everyone's use cases are different, and everyone's definition of how well their computer works differs also. It may not be optimal for YOU but it does not mean it's not optimal for everyone else.

My father purchased a base M1 Mini when it first came out. He uses it to surf, email, and light gaming. It's now years 3 1/2 years later, and he's never had an issue. Thing runs great. He'll probably get another 2-3 years for his use case before it even starts to slow down. There are tons of users out there that have similar use cases, and I'd argue that $500 for 5-6 + years of quality computing is pretty darn good.

What are you shouting about? And misquoting my post as well. What I said was:
The fact that the vast majority do not return their purchases simply means that the base RAM works well enough for the first year or two. It does not mean that base RAM was an optimum life cycle purchase decision.

I did not say that "that it means whether the 8GB only works well enough for the first year or two." The key word that you slid in rewording the gist of my comment was only. We agree that some light users can get by with underpowered boxes; I have argued against the mainstream many times here for Apple to continue to provide the cheaper 8 GB RAM option for those low usage folks.

You shout that "It may not be optimal for YOU but it does not mean it's not optimal for everyone else." What I argue is that for most here (not "everyone else") a careful look at RAM demands history and desired life cycle will show adding RAM above basic to be most cost effective.

Or one can compute sub-optimally forever. It is a choice. But IMO spending $thousands on a box to compute with and then forcing sub-optimal computing by configuring it with less RAM is just wrong.

 

[Allen_Wentz]

Then where are all of the posts from people saying that their machines only work for the first year or two?

C'mon, lose the reiterative circular reasoning based on what you say. I did not say "machines only work for the first year or two," I responded to your claim that "they (Apple) would see it in returns and poor feedback."

 

[Allen_Wentz]

Then where are all of the posts from people saying that their machines only work for the first year or two?

Who said anything about "people saying that their machines only work for the first year or two?" Optimal versus sub-optimal computing and life-cycle computer configuring is what I am discussing. If that is too subtle for you, so be it.

 

[ric22]

But again, if the only reason the base model exists is to force an upgrade, then why are they pushing the base model through the retail channel and not the upgrades? If the base model doesn't work for the majority, why would they lead their entire retail channel with that?

Because customers will return to buy replacement device sooner ? As they can't be upgraded after purchase like in the past...

Win win for Apple.

 

[LavaLevel]

Newbie here. First post.

I found the OG post enlightening, encouraging and rather true.

I still have a Mac Mini (2012) and couldn't compile my App because I can no longer install latest X-Code for High Sierra 10.12.6 which i'm stuck on.

So I've been forced to using my wife's M1 8gb/256 Macbook Air and I am blown away by the speed in which it compiled my App. buttery smooth. Perfectly fine. Can't wait to put my App on the App store. Built on a Mac Mini(2012) compiled on an M1. 😂

Going to get a bare bones M2 asap.

What I got out of this thread is, know your needs & buy accordingly.

16gb base is not going to make the PC world all the sudden respect Apple.

 

[Blair110]

What are you shouting about? And misquoting my post as well. What I said was:
The fact that the vast majority do not return their purchases simply means that the base RAM works well enough for the first year or two. It does not mean that base RAM was an optimum life cycle purchase decision.

I did not say that "that it means whether the 8GB only works well enough for the first year or two." The key word that you slid in rewording the gist of my comment was only. We agree that some light users can get by with underpowered boxes; I have argued against the mainstream many times here for Apple to continue to provide the cheaper 8 GB RAM option for those low usage folks.

You shout that "It may not be optimal for YOU but it does not mean it's not optimal for everyone else." What I argue is that for most here (not "everyone else") a careful look at RAM demands history and desired life cycle will show adding RAM above basic to be most cost effective.

Or one can compute sub-optimally forever. It is a choice. But IMO spending $thousands on a box to compute with and then forcing sub-optimal computing by configuring it with less RAM is just wrong.

First off, I'm not shouting. So calm down. Second, I didn't mischaracterize your reply. You fail to recognize that what you deem an "optimal life cycle decision" is a subjective term and means different things to different people. The fact remains that most (and most doesn't mean MR forum users) can use base models effectively for years to come. For what they consider optimum for them. Adding RAM is not cost effective for them, because either they can't, won't or unable to do that. So again, for those people, it's fine.

They've made an optimal purchase for their use case and needs, and Apple is happy to provide that.

 

[turbineseaplane]

To me the issues would be solved by Apple:

1. Bumping all Macs to 16GB RAM as the "base"

2. Stop overcharging so much for SSD upgrades

3. Related to #2 - Apple: do whatever you want with base SSD and upgrade prices, but add a customer accessible NVMe slot, so folks can pop in their own second SSD stick

Sony did with this with the PS5 (user accessible NVMe slot)

If game console users are capable and "allowed" to do such an upgrade themselves, all computer users should be able to do the same.

Considering Apple likes to fancy itself an "environmentally friendly company" (sort of laughable for any MegaCorp), they should build some more longevity into these devices.

 

[progx]

Why would he get any hate for his post?

You should look at posts from 10-20 years ago on here. My favorites were always the ones where Steve Jobs should be fired for releasing the G4 Cube and iPod.

I made a blunder by saying the Intel Macs were going to be crap in 2005. So, I’m not immune to making some mistakes myself.

 

[AlmightyKang]

I disagree. You suggest modifying workflow to conserve on RAM usage, which is IMO very unproductive when more RAM can be available. In the old days we did all kinds of tricks to cope with unavoidable RAM limits, but today it is far smarter to spend $800 for +64 GB of RAM and just let your brain be creative at work product instead of being creative at how to waste brain cycles fitting workflow under less than optimal RAM

At some point you can't just throw more RAM and money at the problem. You either don't have enough of it or you reach the limits of the technology. We have a crap load of money at work so we throw 2TB of RAM into some of our machines. With a couple of days of optimisation work and changing how things worked internally (mostly batching of workload) I got that down to 1TB, halved the execution time and saved us $500,000 a year in AWS fees. That was after eating that cost for 3 years.

The resources are cheap mentality does not scale well in practice. Point is pick your tools wisely and make sure that they will scale with the workload.

If you do a linear regression model of cost vs performance over time, the wall is a lot further away anyway when you spend less and think more early on.

 

[Analog Kid]

Your post was well thought out until the last few words: "...8GB is fine for many years to come" which is just wrong.

I feel like everyone going to the future and coming back with RAM recommendations are missing the better opportunity... Much better to come back with stock tips. If I save $400 on RAM today and invest it, what will give me the best return over the next 5 years or so?

 

[AlmightyKang]

I feel like everyone going to the future and coming back with RAM recommendations are missing the better opportunity... Much better to come back with stock tips. If I save $400 on RAM today and invest it, what will give me the best return over the next 5 years or so?

Well I'd spend the $400 on floozies and beer personally.

 

[Allen_Wentz]

First off, I'm not shouting. So calm down. Second, I didn't mischaracterize your reply. You fail to recognize that what you deem an "optimal life cycle decision" is a subjective term and means different things to different people. The fact remains that most (and most doesn't mean MR forum users) can use base models effectively for years to come. For what they consider optimum for them. Adding RAM is not cost effective for them, because either they can't, won't or unable to do that. So again, for those people, it's fine.

They've made an optimal purchase for their use case and needs, and Apple is happy to provide that.

First off, your use of all-caps is shouting, by definition.

Secondly, you keep rewording what I say. I discuss optimal computing so you misquote me as "optimal life cycle decision." Sorry but such discussion is just going in circles.

At some point you can't just throw more RAM and money at the problem. You either don't have enough of it or you reach the limits of the technology. We have a crap load of money at work so we throw 2TB of RAM into some of our machines. With a couple of days of optimisation work and changing how things worked internally (mostly batching of workload) I got that down to 1TB, halved the execution time and saved us $500,000 a year in AWS fees. That was after eating that cost for 3 years.

The resources are cheap mentality does not scale well in practice. Point is pick your tools wisely and make sure that they will scale with the workload.

If you do a linear regression model of cost vs performance over time, the wall is a lot further away anyway when you spend less and think more early on.

?? I was referencing Macbook Pro RAM choices and the workflows that run on them, not 2TB of RAM solutions. IMO for most heavy MBP use cases as described it is indeed cost effective to throw +64 GB of RAM at operations rather than spending 5+ years compromising every day to fit operations under lesser RAM.