Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
My 48Gb m4 pro has 31Gb of memory used. Most of it is cache currently and by design it will just fill up. Most of it will be very old data that might not be called up again and it will just start to recycle space.

I do work with very large files and lots of open files, but I'm virtually certain 24Gb would be enough for most people.

Please could you elaborate on what Apple Intelligence is doing in the background to consume so much memory for normal tasks? (please excuse my ignorance here).
 
  • Like
Reactions: phrehdd
I was going to buy the highest M4Pro, with 1TB drive and possibly 64GB RAM - but Ive decided to hold for the Studio instead. I feel at that pricing I'm close to M4 Max Studio territory, and I'm in no rush......
 
  • Like
Reactions: phrehdd
I was going to buy the highest M4Pro, with 1TB drive and possibly 64GB RAM - but Ive decided to hold for the Studio instead. I feel at that pricing I'm close to M4 Max Studio territory, and I'm in no rush......
Ditto
 
  • Like
Reactions: loby
Daft question, but which one is the binned one?
Technically I'd say they are both binned (just differently), but this thread considers the 12 core Pro part to be binned.

I've generally seen people refer to the higher spec part as "binned" in the PC space, but that doesn't seem to be the language around the Macs.
 
The act of "binning" a processor is to test it for performance vs. the baseline (the baseline is ALWAYS the full-featured processor running at maximum performance).

If a processor runs in-spec, then it goes into the top-tier pile.

If a processor fails a test, is slower, or straight-up doesn't work, then they look at what parts of the die can be disabled and produce a viable lower-tier processor from the failed device. If it still runs in-spec with cores/GPU disabled, then it gets "binned" as a lower-tier processor spec.

If a processor has a problem that they can't get around through disabling features, then that silicon is wasted and disposed of.

Something many people don't seem to appreciate is most tiers of CPU are never designed. The manufacturer designs one processor (the top-tier model) and then monitors yield in production. If 90% of the yield is a fully working processor that is in-spec, then it is party time as most systems can be of the higher spec, and little money is wasted.

If most of the yield is defective in some way, then they get very stressed, and this causes processor prices to rise significantly (looking at you, Intel).

To squeeze money out of the last 10% of the wafer that was defective, they produce the lower-tier processors that have bits disabled to make them functional. Over time, they learn what they can get away with in terms of spec, and over time lower and lower tiers of processor appear until they are practically useless (hello i3). This can take some time, and depends on how many of the processors are able to be the top-tier product.

There are other considerations, such as demand, sales forcasts for the various processors, cost, etc., but this is a brief overview of how processor models are "made".

In the case of M4 vs. M4 Pro, these appear to be two distinct processor designs, as the die layout is different between the two models of processor.

In terms of the difference between 12/16 and 14/20 core versions of the M4 Pro, this is where the disabling of parts of the processor die comes into effect, to create the 12/16 version from the 14/20 version.

It is worth noting that as part of the design, they build-in the ability to disable parts of the processor. Two approaches are taken during manufacture: they can blow fuses in testing, or they can cut the die to break links. Some are done in microcode, which is why with certain models of AMD processors it was possible to turn a hex-core into an octa-core.
 
Last edited:
Whether it is worth going for the 14/20 or sticking with the 12/16 model depends on your use-case.

You may not see any raw performance gain from the 14/20 version, but it does offer the ability to run more tasks in parallel, which may or may not be useful.

For me, it is a benefit, because I intend to run VMs, and they require cores, so more cores = more VMs running concurrently, or more cores per VM for increased performance in those VMs while keeping the host machine snappy.

Memory matters for a lot of things, so more is always good.

The memory is shared between the CPU and GPU, so you need to allow for this in your calculations. For my use, 48 GB is a sweet-spot. This is equivalent to 16 GB of VRAM and 32 GB of system memory on a typical PC, which for my uses is more than adequate.

With the way macOS can be flexible with this memory, it actually gives me even more memory than I have with my PC in typical workloads, as most of my stuff is not VRAM heavy.
 
My 48Gb m4 pro has 31Gb of memory used. Most of it is cache currently and by design it will just fill up. Most of it will be very old data that might not be called up again and it will just start to recycle space.

I do work with very large files and lots of open files, but I'm virtually certain 24Gb would be enough for most people.

Please could you elaborate on what Apple Intelligence is doing in the background to consume so much memory for normal tasks? (please excuse my ignorance here).
When choosing memory, the key factor is whether you're considering it as "VRAM" in your decision-making. If you're like me, an AI professional planning to deploy AI projects on this machine, then the 64GB memory (which also acts as VRAM) is an absolute must. Similarly, opting for a 14-core CPU and a 20-core GPU is equally important. If you go with the 20-core GPU, you’ll need to pair it with the 14-core CPU, not the 12-core.

If you think Apple's memory is overpriced, just compare it to the cost of Nvidia's VRAM, and you’ll realize how reasonable Apple's memory (VRAM) actually is.

As for the SSD, I think there’s a good chance that third-party, replaceable SSD modules for the Mac mini M4 will hit the market in the near future. The 2024 Mac mini M4 is one of the hottest topics globally, and the entry-level model is expected to see a lot of buyers. However, its 256GB SSD is likely to be an upgrade target for many users. Considering that engineers in Shenzhen’s Huaqiangbei district are already working on circuit boards, molds, and product testing, it’s only a matter of time before replacement options become available.
 
When choosing memory, the key factor is whether you're considering it as "VRAM" in your decision-making. If you're like me, an AI professional planning to deploy AI projects on this machine, then the 64GB memory (which also acts as VRAM) is an absolute must. Similarly, opting for a 14-core CPU and a 20-core GPU is equally important. If you go with the 20-core GPU, you’ll need to pair it with the 14-core CPU, not the 12-core.

If you think Apple's memory is overpriced, just compare it to the cost of Nvidia's VRAM, and you’ll realize how reasonable Apple's memory (VRAM) actually is.

As for the SSD, I think there’s a good chance that third-party, replaceable SSD modules for the Mac mini M4 will hit the market in the near future. The 2024 Mac mini M4 is one of the hottest topics globally, and the entry-level model is expected to see a lot of buyers. However, its 256GB SSD is likely to be an upgrade target for many users. Considering that engineers in Shenzhen’s Huaqiangbei district are already working on circuit boards, molds, and product testing, it’s only a matter of time before replacement options become available.
Are you just copying and pasting this same response in every thread?
 
What’s the use case for that? Sure if you’re just looking to load the machine, but you can only watch one video at a time?
No particular use case, just was curious what is the difference since i have both machine at hand. Basically if the workflow requires more RAM, then the device with more RAM would be faster. So I figured if one is driving 2 4k displays, then better have something with Mx(1,2,3,4) pro cpu.
 
Are you just copying and pasting this same response in every thread?
I'm sorry about that, and my mistake was not explaining clearly why I did it.
A few days ago, after reading some different discussions about choosing Mac Mini configs, I had those thoughts and wrote them down. So I posted the same thing on all the threads I saw.
 
  • Like
Reactions: phrehdd
Agreed. The base model Mac mini is plenty for nearly everybody's use case, sure add more storage (or hang an SSD off the back and save a boat load of money) if you need it - but more compute or RAM is lost on pretty much everyone. Those who need it know they need it and don't ask on MacRumors forums.
its basically the same on reddit where everyone seems to be a photographer/video editor lol
 
Or programmer/developer. People who really are those things know what they need. They don’t ask on Reddit.
Please forgive me for disagreeing with your point. Not every programmer out there is a seasoned expert. Every day, new developers are learning and growing, and they need to gather information from multiple sources—whether that’s MacRumors, Reddit, or YouTube. The real world is constantly changing, every minute and every second, and so are we as individuals. Thank you all.
 
  • Like
Reactions: phrehdd
Please forgive me for disagreeing with your point. Not every programmer out there is a seasoned expert. Every day, new developers are learning and growing, and they need to gather information from multiple sources—whether that’s MacRumors, Reddit, or YouTube. The real world is constantly changing, every minute and every second, and so are we as individuals. Thank you all.

As I said, people who need the extra compute or RAM know why they need it (I can afford it and I just want it is a valid answer of course). Everyone else can buy the base model (add more storage if you insist) without worry.

Even people who are beginning developers are more than fine with a base model Mac today.

That said, the definition of the word developer has really been watered down in recent years. I see people who learned how to use a Bootstrap template calling themselves "full stack developers" all the time.
 
Last edited:
Please forgive me for disagreeing with your point. Not every programmer out there is a seasoned expert. Every day, new developers are learning and growing, and they need to gather information from multiple sources—whether that’s MacRumors, Reddit, or YouTube. The real world is constantly changing, every minute and every second, and so are we as individuals. Thank you all.
For me being a programmer and not being able to choose a proper machine is like being a cavalryman and not being able to choose the right horse.
 
  • Like
Reactions: sarga
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.