[ghporter]

All of the current MacBook Air and Pro models come standard with "8GB Unified Memory".

Let's consider an MBP with the M2 chip. If this is a "primary" computer, and the user runs photo editing software, Xcode, and a few other supposedly "processor intensive" apps, is that 8GB enough memory?

And from the other direction, is it worth $200 to bump these machines up to 16GB?

[subego]

Spend the $200, without question.

For the actual coding part, check out Nova. I waaaay prefer it to Xcode.

https://nova.app/

[reader50]

Agreed, 8 GB is not enough today. And it's more like 7 GB - when they say "unified memory" they mean the graphics chip doesn't have dedicated RAM. Instead it grabs up to 1 GB of system RAM.

In Mojave I typically use 10-11 GB with just browsing, Mail, VLC, Finder, etc. After upgrading to Monterey, I'm using 13 GB with the same ordinary apps. No movie editing pushing things.

[ghporter]

Interesting. I kind of figured that the software would catch up and use all the available RAM.

[Spheric Harlot]

Originally Posted by reader50 

In Mojave I typically use 10-11 GB with just browsing, Mail, VLC, Finder, etc. After upgrading to Monterey, I'm using 13 GB with the same ordinary apps. No movie editing pushing things.

To be fair, that is a pretty useless figure, because it might just mean that Monterey is BETTER at using the available resources, while Mojave is WASTING available RAM by leaving it unused.

[ghporter]

Serious question: I’m curious how having more RAM than is needed at any specific moment is “wasting” that RAM. I know that the OS guys at Apple changed their thinking in 2013 to reflect “unused RAM is wasted RAM”, but I don’t get how it’s “wasted”. If RAM isn’t saturated by App A, then App B might be more responsive (or more efficient, etc.) because there’s RAM available and neither app needs to page out.

Also, available RAM/virtual RAM should allow more running apps to work “faster” in the background, again because they don’t page RAM contents in and out at hard disk/SSD speeds. That would also allow more processor/RAM-intensive apps to be quicker, more responsive, more efficient, etc.

Additionally, since we’re stuck with Apple’s offerings of 8GB, 16GB and 24GB, if it turns out that a particular machine runs way better using “more than 8GB but less than 16GB”, how could “wasting” the RAM between the 8GB and 16GB steps be avoided? With reader’s examples, he uses more than 8GB, but there’s no option for what we might call “optimal” RAM sizes between 8 and 16GB. I don’t know if this is even related to the same point; which kind of emphasizes how this not-very-recent computer science grad just doesn’t understand either the post 2013 “wasted RAM” paradigm or the Apple Silicon memory management concept.

So as I said, what does “wasted” mean? Does it mean something when only one program is running, but something different when there are more running? I understand that Apple’s processors have MUCH faster access to that “unified” memory, and how that implies a new way of managing RAM. What I don’t get is how any resource that isn’t 100% saturated can be wasted without some specific parameters regarding the context, processor work load, number of apps in foreground/background/suspended, etc.

Finally, are there any benchmarks that can demonstrate how Apple Silicon RAM quantity impacts performance? I’m thinking about tests like the old “max FPS” or similar tests that stressed the processor/RAM/video systems of PCs to demonstrate differences between processor speeds, RAM size, and so on.

[Spheric Harlot]

My limited understanding is that the system can and will load whatever it may deem potentially useful into RAM, based on what it thinks you might use, what it might/will need, and what you've used in the past.
This may mean caching parts of the OS, applications, whatever.

If you're not running enough stuff for the OS to be able to fill up RAM, you've got too much RAM for what you're doing —> wasted resource.

The more effectively the OS can fill up the RAM, the less of it is wasted.

I'm NOT saying that Monterey is necessarily more efficient — that could just be bloat. I AM saying that RAM usage is not an indicator for how much memory is NEEDED. It MAY just be an indicator that macOS engineers have found more ways to load stuff from disk into the much faster RAM, when available.

[ghporter]

“More ways to load stuff from disk into much faster RAM” sounds likely. I wonder if they are interacting with the disk cache directly instead of handing off the data then requesting it later.

OS people tend to act like “mere mortals” can’t possibly understand what they do - or more likely can’t understand what their code does. This is especially true with the OS parts that are specific to fancy hardware. So there aren’t a lot of “executive summaries” about how Apple Silicon manages RAM because executives can’t conceptualize it. Supposedly…

Anyway, if I skip a step in fetching data - get it straight from RAM and not from cache - I may not notice the difference, but my apps will certainly notice. I can comfortably spend a bit more for more RAM, and feel my engineer persona has awakened, over-spec’d the RAM, and then dozed off again.

[ghporter]

I’ve just run into a great deal on a June, 2022 M2 MBP from the Apple Refurb store, so I pounced on it. This particular machine has 24GB of RAM and a 512GB SSD. It’ll do.

Yes, I still feel that “more than absolutely necessary RAM” is better than “just enough RAM”. Now I’ll actually have a chance to see how that plays out.

[OreoCookie]

You did the right thing, Glenn. I am typing this on a 2013 13" MacBook Pro with 8 GB RAM and a 256 GB SSD. 8 GB means the machine is constantly struggling. With my browsing habits, even 16 GB is not enough. 24 GB should give you some breathing room and e. g. allow you and your wife be logged in simultaneously.

[ghporter]

While I can’t tell if it’s “only” due to the larger amount of RAM or “only” the more efficient M2 processor with its more efficient connections to basically everything, but this machine is fast. And frankly I don’t care which one makes the difference, though it’s probably the combination of the two.

I’m running a couple of Intel-only apps that are graphics-intensive (Canon’s EOS Utility and their Digital Photography Pro), and they are pretty amazingly fast so far.

It’ll probably have to wait a bit, but I’m going to dig into challenging them, and a universal app (RawTherapee - an app that reads and manipulates Canon Raw 2-format images).

My camera produces some pretty darn large RAW files, and my work flow is to tether the camera to the computer using the EOS Utility. This lets me see exactly what the sensor sees, and make adjustments (white balance, etc.) before capturing the image. With my 2015 MBP, there was a distinct lag while the image was saved, and it “felt” slow in implementing camera setting changes from the computer.

Plus, since the app doesn’t actually edit images, but instead applies edits to how the image is displayed, the computer does a bunch of processing every time an image file is opened. With existing, already edited files, this really felt transparent. I’m looking forward to seeing how quickly new images are displayed and how quickly edits show up.

[OreoCookie]

[QUOTE=ghporter;4430951]While I can’t tell if it’s “only” due to the larger amount of RAM or “only” the more efficient M2 processor with its more efficient connections to basically everything, but this machine is fast. And frankly I don’t care which one makes the difference, though it’s probably the combination of the two.[/quote
The machine I am typing on is a cautionary tale: it was the entry-level 13" MacBook Pro at the time. The machine itself is solid, but 9 years later being RAM and disk space starved is worse than before. The 24 GB will give your machine a longer life.

Originally Posted by ghporter 

I’m running a couple of Intel-only apps that are graphics-intensive (Canon’s EOS Utility and their Digital Photography Pro), and they are pretty amazingly fast so far.

Yeah, the M-series processors are really fast. In many benchmarks they are faster (running in emulation mode) than on the last Intel-based MacBook Pro while consuming less power.

Originally Posted by ghporter 

Plus, since the app doesn’t actually edit images, but instead applies edits to how the image is displayed, the computer does a bunch of processing every time an image file is opened. With existing, already edited files, this really felt transparent. I’m looking forward to seeing how quickly new images are displayed and how quickly edits show up.

Loading lots of images requires lots of RAM. I remember learning that when I first got Aperture way back when.

[ghporter]

I haven’t had a chance to do any timing tests, but as I mentioned above, the M2 machine feels hugely faster when opening the same image file with the same app.

And since I can add external storage through the Thunderbolt 4/USB C ports, the built in 512 GB flash drive isn’t a real issue related to future usability. But RAM is another matter. I had expected to buy a new machine with 16GB of RAM, but this refurbished MBP showed up out of the blue, and I’m glad I grabbed it.