New M2 Pro and M2 Max MacBook Pro + New M2 Mac mini

[kerochan]

1 minute ago, David Nahmani said:

Then doing the same work you would likely never hear the fans on the Mac mini M2 as well. 

possibly!

[analogika]

5 hours ago, onewave said:

Sure is tempting to look at the 12 core M2 pro machines, since nobody on this thread has mentioned it, how about considering the Lpddr5 memory speed on M2...

" The biggest change of which is support for much higher memory clockspeeds; based on Apple’s figures, the M2 is running at 6400Mbps/pin (LPDDR5-6400), which is up significantly from the 4266Mbps/pin (LPDDR4x-4266) memory clockspeeds of the original M1. The net result is that, on the SoC’s 128-bit memory bus, the M2 has 100GB/second of memory bandwidth to play with, a 50% increase over the M1 (~68GB/sec)."

https://www.anandtech.com/show/17431/apple-announces-m2-soc-apple-silicon-updated-for-2022

I've heard this is a much faster spec and makes the whole M2 infrastructure move data faster.

It will indeed be interesting to see what results M2 12-core Pro users will have with plugins once the machines have been used in productions in the coming weeks. 

Always the big downside with updating a new computer is drawback of dealing with the usual third party software issues. For this reason alone I've stuck with the M1 base model, at least till all of my hardware / software will work on Ventura, this could take much longer still... So yes M1 pro / max is a great value !!!

This was apparently incorrect. I've since found (and verified on notebookcheck) that just the basic M1 used LPDDR4x.

The M1 Pro and Max both switched over to LPDDR5-6400 on the same 256/512 bus widths.

[Migs 60]

Hi Folks, 

New to the forum and looking for some viewpoints please.

It's time to upgrade and I was thinking M1 Max Mac Studio with 10 core and 64g RAM as I hear RAM is king when it comes to Logic. But now the M2 Pro Mini is here, I could spec that up with 12 cores but only 32g RAM. Which is likely to perform better? I am guessing the extra 2 cores/M2 performance won't offset that extra 32g of RAM when it comes to multiple CPU intensive soft synths and plug ins?

Then again, is 64gig overkill? (I am upgrading from a 2013 MBPro, so anything will be a significant improvement!) 🙂

Waiting for an M2 Studio isn't an option as I need to upgrade now. 

What would you guys do? I'm leaning towards to the Mini and it's a little cheaper which is bonus.

Thanks in advance.

[analogika]

Assuming that the Mac mini is the same design internally as the MacBook Pro, I would opt for the Mini with 32 GB. 

The M1 Max will be slower by about 20% for pure CPU tasks, and it has the same number of performance cores (8 of them). 

Its Neural Engine (which Logic apparently uses today for things like beat analysis, and which will likely be more and more important going forward for all things flex, loop indexing, and anything involving pattern recognition) is apparently about 40% faster on the M2 Pro series, and disk speed is also boosted. 

The RAM would be most important if you're using huge sample libraries, I'd think. 

[mrgeeze]

I also own a Macbook Pro  2013. Intel Quad Core. I think I paid $4k back then. Quite simply the best computer (of any brand) I've ever owned. I do miss my old Compaq Portable Lunchbox running MsDos (ouch!).

I plan to purchase the Mini with 32GB.  I don't see the point for 64GB RAM.

I'm still on the fence about internal vs external SSD capacity. I know AAPL wants a premium for its storage but my last mac lasted ten years. Once I have it I'll be set for quite some time.

[Migs 60]

14 minutes ago, analogika said:

Assuming that the Mac mini is the same design internally as the MacBook Pro, I would opt for the Mini with 32 GB. 

The M1 Max will be slower by about 20% for pure CPU tasks, and it has the same number of performance cores (8 of them). 

Its Neural Engine (which Logic apparently uses today for things like beat analysis, and which will likely be more and more important going forward for all things flex, loop indexing, and anything involving pattern recognition) is apparently about 40% faster on the M2 Pro series, and disk speed is also boosted. 

The RAM would be most important if you're using huge sample libraries, I'd think. 

Helpful, thanks. I do use some large sample plugins and over sampling but not a lot. 

[Migs 60]

2 minutes ago, mrgeeze said:

I also own a Macbook Pro  2013. Intel Quad Core. I think I paid $4k back then. Quite simply the best computer (of any brand) I've ever owned. I do miss my old Compaq Portable Lunchbox running MsDos (ouch!).

I plan to purchase the Mini with 32GB.  I don't see the point for 64GB RAM.

I'm still on the fence about internal vs external SSD capacity. I know AAPL wants a premium for its storage but my last mac lasted ten years. Once I have it I'll be set for quite some time.

My 2013 MPro has definitely served me well but it will be nice not to have to bounce to audio as much! I’m thinking 512 storage will do to run day to day and offload to projects to external drives I already have and by the time I need more external SSD it will be even cheaper. 

[fisherking]

ram is managed differently (& far more efficiently) on the silicon chips; 32GB is a lot.  personally, i'd go with the M2 and that, than 64GB ram... unless you're planning on editing 3 hour 8K movies... with 12 channels of audio, while working on a huge orchestral score in logic.... (which might actually be fun)

[analogika]

9 minutes ago, fisherking said:

ram is managed differently (& far more efficiently) on the silicon chips; 32GB is a lot.  

While I agree that 32 GB is a lot, I believe the idea that RAM is managed differently, or that Apple Silicon somehow reduces RAM requirements for things like large sample instruments, has been debunked as a myth. 

It *seemed* more efficient in the early days of Apple Silicon, because Apple had optimised the hell out of everything, and it was mostly Apple's own software that was native and gave those insane benchmarks, what with 120 instances of Space Designer and such.

Yes, RAM is managed differently due to the shared graphics RAM, but that mostly benefits graphics-heavy processes. 
a 9-GB Kontakt instrument will apparently still require those same 9 GB under Apple Silicon. 

I hear this from a guy who builds MainStage setups and has done some not-entirely-formalised testing in the process. 

[Migs 60]

2 minutes ago, analogika said:

While I agree that 32 GB is a lot, I believe the idea that RAM is managed differently, or that Apple Silicon somehow reduces RAM requirements for things like large sample instruments, has been debunked as a myth. 

It *seemed* more efficient in the early days of Apple Silicon, because Apple had optimised the hell out of everything, and it was mostly Apple's own software that was native and gave those insane benchmarks, what with 120 instances of Space Designer and such.

Yes, RAM is managed differently due to the shared graphics RAM, but that mostly benefits graphics-heavy processes. 
a 9-GB Kontakt instrument will apparently still require those same 9 GB under Apple Silicon. 

I hear this from a guy who builds MainStage setups and has done some not-entirely-formalised testing in the process. 

That is how I understood things to be. It’s great to see bench tests of 300 tracks with a logic synth and reverb but as soon as you go for the more CPU heavy third party stuff like that of Native Instruments or Roland Cloud, it’s a different ballgame.

I think for me though 32gig should be enough, certainly way better than I have today. Hopefully there won’t be any teething issues with these new machines as others have called out previously. 

[fisherking]

18 minutes ago, analogika said:

While I agree that 32 GB is a lot, I believe the idea that RAM is managed differently, or that Apple Silicon somehow reduces RAM requirements for things like large sample instruments, has been debunked as a myth. 

It *seemed* more efficient in the early days of Apple Silicon, because Apple had optimised the hell out of everything, and it was mostly Apple's own software that was native and gave those insane benchmarks, what with 120 instances of Space Designer and such.

Yes, RAM is managed differently due to the shared graphics RAM, but that mostly benefits graphics-heavy processes. 
a 9-GB Kontakt instrument will apparently still require those same 9 GB under Apple Silicon. 

I hear this from a guy who builds MainStage setups and has done some not-entirely-formalised testing in the process. 

yeah, i stand by what i wrote, despite what some 'guy who builds mainstage setups' says. you can research this easily.

[des99]

29 minutes ago, analogika said:

While I agree that 32 GB is a lot, I believe the idea that RAM is managed differently, or that Apple Silicon somehow reduces RAM requirements for things like large sample instruments, has been debunked as a myth. 

I've talked about this a bit before so I won't go on about it, but from my perception, this all started to happen when the first M1 Macs were released, as 8GB and 16GB models.

While initially disappointed with those memory choices, as people thought, reasonably fairly, that 8GB Intel Macs weren't that great in terms of performance, and you generally needed more to do much of any note smoothly, people found that even the 8GB M1 models ran smoothly, didn't hiccup, didn't seem slow or constrained as they might expect, and started reporting "these 8GB models are really nice, they behave more like I'd expect a previous 16GB machine to behave" - which then slowly chinese*-whispered into "8GB on M1 is the same as 16GB on other computers" and people started to get all confused, thinking that somehow M1 memory was magic or something...

Memory on these machines is fast, and shared - which has some great performance benefits, and some tradeoffs - but it's not magic. If you need to store 8GB of data in RAM, you can't do in on an 8GB machine (the system already takes up a chunk of that), you can on a 16GB machine.

I think 32GB is a good performance point for people doing serious work on their machine, and expecting a good lifetime from it. If you need more, you're a demanding user, and you probably know you need more.

I have a 16GB machine, as I needed one and it was all I could get hold of during the worst of the supply constraints (otherwise I'd be waiting months). I would have preferred 32GB - but honestly, if my machine was 32GB instead of 16GB, and I think of what I've done with it - it would have had *zero* affect on my work. I've been on 16GB for the last decade, it doesn't seem to be a limiting factor for me going forward. Everything I've used this computer for so far has been a delight...

* Is there a more politically-correct term for this these days? I should woke up a bit on some of those old, worn-out phrases...

17 minutes ago, Migs 60 said:

as soon as you go for the more CPU heavy third party stuff like that of Native Instruments or Roland Cloud, it’s a different ballgame.

The Roland Cloud plugins* are horribly unoptimised on Apple Silicon. Their engines are multicore which doesn't work great on M1 (it's optimised for Intel) and so far, the engine rewrites they'd need to do to get their plugins performing as they should haven't been forthcoming. Hence even the lighter plugins on Intel are CPU heavy on M1.

So basically, bad code isn't a good way to benchmark a system - but it *is* worth knowing about if you use those plugins a lot...

* Most of them, it depends on their engines. The ACB ones are mostly all very bad, and the PCM engine ones like the 1080/5080/SRX are passable but still way heavier than they should be. Some of the others seem to work ok - newer engines possibly.

Edited January 26, 2023 by des99

[analogika]

On 1/26/2023 at 1:06 AM, fisherking said:

yeah, i stand by what i wrote, despite what some 'guy who builds mainstage setups' says. you can research this easily.

I suppose Ethan Deppe is indeed "some guy who builds mainstage setups", as per his credentials. You, however, are just "some dude on an internet forum" — not reddit at least, but, well… 😜

In all seriousness: If you have verifiable sources that claim otherwise, please share them! 

Because I have researched this, and the experience of Ethan's is the first tangible "study" I've found on this. 

[fisherking]

8 minutes ago, analogika said:

I suppose Ethan Deppe is indeed "some guy who builds mainstage setups", as per his credentials. You, however, are just "some dude on an internet forum" — not reddit at least, but, well… 😜

In all seriousness: If you have verifiable sources that claim otherwise, please share them! 

Because I have researched this, and the experience of Ethan's is the first tangible "study" I've found on this. 

@analogika, i quoted you on 'some guy who...', which would equal me being 'some dude on an...' 👍

anyway:  ram on silicon

after which, you can of course draw your own conclusions.

[analogika]

53 minutes ago, fisherking said:

@analogika, i quoted you on 'some guy who...', which would equal me being 'some dude on an...' 👍

anyway:  ram on silicon

after which, you can of course draw your own conclusions.

You seriously just sent me a google search link for "ram on silicon" and left it at that?

WTF. Okay, I'll bite...

 

First link: MacWorld. Explaining shared memory, SoC. Nothing about needing less RAM.

Second link: a MacRumors thread that starts off with "RAM is RAM. It doesn't matter whether you're accessing memory from an M1 Mac or ZX Spectrum - the principles are still the same." But hardly a sound technical analysis. Pass. 

Third link: a Quora link that regurgitates a bunch of stuff from the initial keynote, with a sickening amount of hand-waving about "efficiency" and Apple products "needing far less RAM because Apple expended millions in R&D to make their products more efficient" — pass.

Fourth link: Eclectic Light Company. Ah! NOW we're getting some real…nope. Just great for video, but GPU usage eats into available RAM. With measurements, analysis and graphs. 

AH! Another Eclectic Light Company… blah blah GPU - wait: "If demands made on unified memory are more variable, and could require that a high proportion of physical memory is used for graphics and the display, this might result in increased use of virtual memory, and CPU cycles lost to caching." Damn. That sounds like the exact opposite of what you're claiming. 

A little further down, a "techjourneyman" link. Never hear of him, but it seems fairly competent. Oh wait: "However, there are limits of optimizing the hardware to use less RAM. The datasets never change and if the data set is more than available RAM, you still need to swap from storage to RAM. There are stats that show a 16GB RAM M1 will do better for video encoding in 4K from 8K because the datasets is too big for 8GB of RAM."

 

So… no?

 

All of the links that DO talk about seemingly more efficient usage of RAM explain that it's due to predictive swapping and extremely fast SSDs. 

Problem is, extremely fast swapping doesn't keep your sample libraries in RAM when you're actually actively using them. 

For time-critical stuff (like, say, playing live instruments), RAM is RAM is RAM, and libraries don't shrink or become magically less cumbersome just because Apple Silicon is a lot better at predicting when static files may be needed next. 

 

Again: If you have ANY sources to support the conclusion that it might be otherwise, PLEASE SHARE THEM. 

But don't just throw a Google search link in my face. I've been around a lot longer than Google has. 

Edited January 26, 2023 by analogika

[fisherking]

7 minutes ago, analogika said:

But don't just throw a Google search link in my face. I've been around a lot longer than Google has. 

 idon't want to get combatitive! not at all... so, will let it go here.

anyway, am planning the 'top-line' M2 mini with 16GB ram; will be interesting to see how it all goes, after my 2019 imac.

[analogika]

I'm not really trying to be combative, sorry. I'm trying to get to the bottom of this. 

I was really hoping to find support for your claim, but all you've done (besides making me feel mildly insulted) has been to have me forage links supplied by your search, all the credible ones of which disagree with you. 🤷‍♂️

[fisherking]

3 minutes ago, analogika said:

I'm not really trying to be combative, sorry. I'm trying to get to the bottom of this. 

I was really hoping to find support for your claim, but all you've done (besides making me feel mildly insulted) has been to have me forage links supplied by your search, all the credible ones of which disagree with you. 🤷‍♂️

not sure why you feel 'mildly insulted', but sorry to hear that. really tho, all we can do, at this point, is go back & forth, and i can't see any benefit to that.

here's another article, but considering the internet, there are a million articles, about everything.

unified memory

[David Nahmani]

5 hours ago, fisherking said:

not sure why you feel 'mildly insulted'

I've felt that way before when being asked to do a google search when posting in a forum (for a completely different topic mind you). If I'd wanted to do a google search, I'd have done a google search, but the reason we post in forums is because we want to hear about real users who are part of a community, meaning that they may relate to or have been in the same situation we are. 

So I agree with @analogika that it's never a good idea, on this forum, to answer to Google something or to direct someone to the results of a Google search. 

As for the subject at hand, here's a pretty good video that explains how memory works on Apple Silicon machines when using Logic Pro: 

[fisherking]

11 minutes ago, David Nahmani said:

I've felt that way before when being asked to do a google search when posting in a forum (for a completely different topic mind you). If I'd wanted to do a google search, I'd have done a google search, but the reason we post in forums is because we want to hear about real users who are part of a community, meaning that they may relate to or have been in the same situation we are. 

So I agree with @analogika that it's never a good idea, on this forum, to answer to Google something or to direct someone to the results of a Google search. 

As for the subject at hand, here's a pretty good video that explains how memory works on Apple Silicon machines when using Logic Pro: 

i meant no harm, and thought a link was ok (as opposed to saying 'google it'). still, i won't do that again.

[lookatthisguy]

Not to flare anything up here, but @analogika is right. And I saw Ethan's post on Facebook sharing that information with MainStage users, because as one of the owners of one of the biggest companies out there creating MainStage files for musical theatre pit orchestras, his experience carries a lot of weight. But we can also think about this logically—to play back quickly enough, samples traditionally need to be loaded into RAM to be able to be triggered at the right moment. Computers with Silicon processors are using some state of the art, fast components, but try as they might, Apple can't beat the laws of physics. Your average bedroom producer may never hit a ceiling with that throughput, but someone in my shoes, scoring films with decently-sized orchestral sessions, is much more likely to. And if that fancy new Mn x Mac can't stream all those samples from the drive in time, what's it going to do? Load them into the RAM. The samples aren't going to magically change size—they're going to put the same weight onto the RAM now as they would in an Intel Mac. Which is why even with my Intel i9 with 64 GB of RAM, seeing the benchmarks demonstrating how much even the M1s are slapping it around, I'm still holding out until I can get a Mac Studio with 128 GB of RAM. Hopefully by that time, M2 Ultra is already out…

[Migs 60]

I remain of the view that having as much RAM as can be applied/afforded is going to make the single biggest difference over a M1/2 or a couple of extra cores when it comes to Logic sample/VST/Bad Roland code etc. That said, it seems likely it would be more of a consideration at 8g or 16g rather than at 32g or 64g and in all likely hood, 64g is overkill for what I do (no Orchestras here!). By going with the M2 mini I get other benefits i see being called out around latest chip benefits/neural engine/being quieter over the Mac studio etc. So my decision is made and the Mini M2 is getting ordered. 

Thanks again for the views and debate and the video link, really helpful and appreciated. 🙂

[analogika]

6 hours ago, fisherking said:

i meant no harm, and thought a link was ok (as opposed to saying 'google it'). still, i won't do that again.

Thanks for that.

By way of explanation:  
You posted that in response to my comment stating that I had already researched it, and that I had found nothing that supported your claim, and statements from, arguably, an expert, who claimed the opposite.

So I asked you for your sources, and you threw me a google link — even though I'd just written that I'd already researched it. It didn't help that nothing in the Google search supported your claim, either, and several of the links outright stated the opposite.

 

I suppose I did kind of deserve it for joking about you being "just some guy on the internet". 😇

No hard feelings. ✌️

[redgreenblue]

I just have to add that this as about as contentious as I’ve ever seen this forum get, and no one was being a troll or a jerk here. 
 

I maintain that this is one of the best places on the interwebs. 

[David Nahmani]

11 minutes ago, redgreenblue said:

I maintain that this is one of the best places on the interwebs. 

Agreed. 😅

[skillzmcgavern]

There are differing reports of noise under load coming out for the M2 Pro Minis. It seems likely that maximum GPU stress is more likely to cause it than anything audio (CPU) related, but I'm wondering if it is also less likely on the regular M2. If the M2 is pushed to the limit, of course its performance ceiling is lower than the M2 Pro, but at what noise level? Recording a vocal or anything acoustic in front of a fully loaded session is the ultimate ease-of-workflow scenario, so even though great results can be had specifically preparing for this kind of situation (freezing tracks, bouncing a quick 2-track mix so you can turn off all processing during recording, etc.) it is definitely great to get the furthest down this path possible, and something I'm very interested in since I'm upgrading with this generation.

So, if anyone has experience doing this (especially to ridiculous degrees), and their results with both noise and performance vary between M2 & M2 Pro, please share your findings.

[des99]

2 minutes ago, skillzmcgavern said:

It seems likely that maximum GPU stress is more likely to cause it than anything audio (CPU) related, but I'm wondering if it is also less likely on the regular M2.

The low end chips have low core counts, and will therefore generate less heat. Those chips are also in machines that do not have fans at all, so their thermal load is not that great and nothing to worry about.

Honestly, I don't expect much behaviour differences in terms of noise for the M2's over the equivalent M1s - they're all very power efficient, thus generate less heat and require less cooling than the previous Intel chips we were all using. And all heard our fans with those, under varying circumstances.

No serious noise concerns have happened with *any* of the M1s, except for some unit-tolerance issues on some Mac Studios where the fan itself in some case was making some annoying noise - that too, seems to have settled down.

Like I say, I've been using an M1Pro for a year now, and I work this machine quite hard sometimes, and the fans have come on *once*. Fan noise is so much improved on this new generation of Apple machines, that concerns of it being intrusive have been shown to be largely without much merit and not worth worrying about it.

Is there is an unexpected and severe noise issue with these new machines, we will hear about it through reviews and user reports.

[skillzmcgavern]

4 minutes ago, des99 said:

Is there is an unexpected and severe noise issue with these new machines, we will hear about it through reviews and user reports.

The incidents reported so far are only on the M2 Pro, which in the Mini apparently may not have enough thermal architecture to run as silent as the M2. On a teardown of the M2 Pro Mini, it looked like the fan had to be repositioned to barely clear the enclosure edge, but IDK what other compromises come into play - besides the additional CPU/GPU cores that you mentioned.

[des99]

Just now, skillzmcgavern said:

The incidents reported so far are only on the M2 Pro, which in the Mini apparently may not have enough thermal architecture to run as silent as the M2.

Have you got a link to that report? I haven't used an Arm Mac Mini - do the fans run all the time (like the Mac Studio) or do they only turn on when needed, like the laptops? Or is it a case that the M2 fan only turns on when needed, but the M2Pro fan runs all the time?

[mrgeeze]

1 hour ago, David Nahmani said:

Agreed. 😅

it does seem much better for decent discussion than some of the other music forums i visit.

Excellent quality of understanding