Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M3 Max MBP Excessive RAM Utilization
- Thread starter Mity
- Start date
- Sort by reaction score
Are you experiencing performance issues? Other symptoms? If not, what’s the problem exactly?
As others have said, Macs will aggressively preload and cache stuff if there is plenty of RAM available. The more RAM you have the more it is likely that the system will decide to use it in this way.
As others have said, Macs will aggressively preload and cache stuff if there is plenty of RAM available. The more RAM you have the more it is likely that the system will decide to use it in this way.
Unless you're having performance issues, why do you care?
Apple Silicon is designed to utilize as much RAM as it thinks is necessary to optimize performance. RAM, GPU, and processing all happens on the same chips now, so it just takes advantage of what it has at any given time. This is a non-issue and there's nothing you can do to change this correct behavior from your computer.
It’s your choice what you worry about. There is no objective reason to worry about RAM usage from the data you have given us. You cannot measure the system state under irrelevant circumstances and think you have a reliable extrapolation.
Run vm_stats command to see the details of current memory system. I’m sure most of the pages will be inactive or purgeable.
What difference does it make if someone gets 64GB or 128GB? The point is that nothing taxing is being done. Utilization with these tasks should not approach 16-21GB.
For example, I'm able to watch the same videos on an M1 MBA 16GB and total RAM utilization is about 12GB with ~10-15 tabs open. I have watched the same videos on my MBA so why doesn't the RAM push up memory pressure?
Many have tried explaining it to you; if your mindset is of a culprit, it won't be much of a help. Apple silicon is more optimized with unified memory. It will cache and even dump non-frequent memory to swap, it's a good practice; the last thing you want is memory offloaded to swap under load. My 2019 MB 16 couldn't even run decent-sized data sets; anything more than 5-8 GB in Python was hit-and-miss. I am gonna leave this snapshot from my 64GB M1 max, which uses unified memory, and swapping to run a lot more than my Linux workstation. You bought one of the amazing machines available in the market today to be bogged down by one of the worst intel MBP.
The 2019 64 GB MBP was one of the worst Macs I ever owned. There was nothing normal about it: heavy throttling, loud, and poor performance. I usually upgrade every 5 years, but I couldn't wait to get rid of it for a 64 GB M1 Max.
The pictures do show that @Mity correctly did not count the buffer cache towards used memory.
Is there any real indication or proof that Apple's Xnu paging and swapping algorithms have ever changed with the processor architecture?
I am gonna try one last time; if your Mac has more memory, I will try to use it as much as possible. My family uses an 8GB Base model MBA M2. The base model runs fine for its use, and my MBP marks more memory because it has more available. Let OS do its job on managing memory.
You're getting what you paid for - your computer will put more resources toward every task, making everything *snappier* 🙃. If, at some point, you launch a bunch of apps or processes that need more RAM, it'll reallocate resources to those as needed.
If you bought a bunch a RAM just so you can stare at unused RAM in activity monitor all day, you're going to be disappointed. Use your computer and stop wasting your time.
If you bought a bunch a RAM just so you can stare at unused RAM in activity monitor all day, you're going to be disappointed. Use your computer and stop wasting your time.
RAM usage is a good thing. We buy these boxes to compute with, and RAM usage is a very, very good way to compute.
Reading up on Apple's Unified Memory Architecture (UMA) is recommended. Those who understand UMA will make sure that they put plenty of RAM in the boxes they buy to compute with.
Yes there is. It's being used for caching.
Yes. After the SSD Swap overwrite issue in 2021, the swap usage on M1 Macs was significantly different after MacOS 11.4. More changes and security measures were added in 2022. Apple Xnu github page has updates.
I know for a fact that my Intel machine did not. Never. I had activity monitor open every day and watched.
One of the reasons I was so paranoid was that I experienced extreme swap on my base 8GB M1 MBA. I wanted to see how Apple Silicon managed memory. And I can definitely say that Apple Silicon is far more aggressive than x86.
Suppose you have 64GB of total ram and you run a 4K video and use 16GB (as I did in my case). And then you run another 4K video, should caching cause utilization to increase linearly, meaning, 32GB?
Further, suppose that I open other apps. Will Apple Silicon keep utilizing more RAM until I hit 64GB? Or will it start to reduce caching of other processes? And at what point will it start utilizing swap?
Brother. Lot of people have told it to you. System caching is not just simple, it maybe cookies, photos, videos in those website have been cached so you dont need to reload when you scroll or switching tabs.
You use microsoft word then switch to web? Boom. Its fast thats why ram exist and empty ram is waste.
🤷 beats me. This sounds like you're trying to bait me into an argument, and I'm not interested. Good luck.
All I see is two screenshots, both with zero swap. There is nothing to worry about.
Funny way of saying, "I don't know." I'm not trying to bait anyone. It's perplexing to see such a difference from x86.
I'm not worried about utilizing swap now. When will it? From your usage, does it happen when you hit 100% of RAM? Or earlier? If it's earlier, how much so? I have never had an M1 or M2 Max with 64GB so I don't know. I can only go off what I experienced with my Intel Mac.
Last edited by a moderator:
I think you need to open your mind to the explanations being provided here, instead of trying to *fight* the explanations. Nobody is coming here with hostility but your insistence is tough to square with the reality, which is resounding post after post in the replies. Try to let go of your pre conceived notion about how this system works and be open to learning a new way of understanding it.
Interesting, do have a hint on where to look? Checking out some of the 2022 diffs is not the fastest way to find any architecture specific VM changes.
It shows the explicit cache, which is not what I was referring to: the use of memory to save things that are no longer needed but left in case they are recalled.