Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Which doesn’t make sense, when the entry-level chips can outperform the supposed high-end chips. Given that reality, even less people will buy high-end machines, making them even less cost effective/more expensive.

What you say is not entirely accurate. The M4 chips do outperform M2-Ultra in peak performance on short tasks in single-core performance and neural engine throughput, they lag behind in multi-core and GPU performance. However, there is a wrinkle due to the thermal design of the different Mac models. Speed-based throttling in machines with an M4 chip, such as MacBooks or compact systems with limited cooling, significantly impacts performance relative to an M2 Ultra in a Mac Studio, which has a robust cooling system that enables sustained full-speed performance. In sustained tasks, the throttled performance of an M4 chip can be between 60% to 75% that of an M2 Ultra.

I saw the effects of speed-throttling clearly when I ran GPU-compute-bound AI tasks comparing my MacBook Pro M3-Max against my M2-Ultra. The M3-Max was about 80% the speed of the M2-Ultra. That's pretty impressive, nonetheless, but it means my M2-Ultra will still be more capable than other Mac models until it is replaced by an M4-Ultra.

I don’t see a reason why Apple can’t refresh the whole lineup within the span of a few months so that there is a clear value proposition for each product. The current rollout strategy is all messed up.

Logistics, engineering and manufacturing schedules, market demands, etc. I'm sure Apple sells a ton more machines with M and M Pro level chips than M Max and M Ultra chips. It makes sense for them to spend more time and effort on the much larger market for the affluent consumer than on the high-end content creator and software engineering niches.
 
Last edited:
This happens every year. People were buying M2 Mac minis instead of the 8 month old Mac Studio M1 and right now people are buying M4 Mac minis instead of the old M2 Mac Studio.... Apples release schedule kinda blows for high end users. I'm thinking I may just work from low end machines going forward and update them for frequently. The $499 Mac mini m4 is more powerful than any computer I've used so I'm wondering why I should spent 4+ times the cost on a new studio with older chipsets.

The price difference between a low-end ($599) M4 Mac mini and a low-end ($2000) M2-Max Mac Studio is significant, but there are some real differences that affect performance. The Mac mini will beat the Studio on single-core CPU performance according to benchmarks. But on multi-CPU performance, the 10-core mini is only about 83% of the speed of the 12-core Studio.

GPU performance is a much bigger difference. The 10-core mini is about 40% the speed of the 30-core studio. That drops to about 30% of the speed under throttling for thermal management. That's a huge advantage to the Studio if you are a professional processing image, video, graphics, or AI data.

Memory bandwidth on the mini is 120 GB/s and 400 GB/s on the Ultra. Again, this is a huge advantage on a studio for professional workloads.

Ethernet on the mini is 1GB. It is 10GB on the studio. If your work requires you have multiple machines and move data between them, you should have standardized on 10GB ethernet five years ago.

The SSD on the mini is half the size of the Ultra. IOW, the low-end mini is nearly useless for any task requiring lots of media or large AI models unless you buy external storage.

Now, if you aren't a professional or you are a programmer doing Mac or iOS development that doesn't require large data or sustained computation, the mini is a fine machine and will cost you less than the Studio. But, if you are a professional who needs the greater resources of the Studio, the mini is a false bargain.
 
Last edited:
  • Like
Reactions: bzgnyc2
What about the M6 chips?
I’ll wait for the M Quantum. Then we can argue over how many qubits it has. But the usual crowd here will declare that Qualcomm’s quantum chip is better, of course.
 
I got my M2 studio max in late '23' and I'm very happy with it.

At the time when I got it, many people on the forum told me that I was crazy since the M3 studio would be out in spring '24' :).

I'm glad that I didn't wait.
I waited and wound up gettin an m4 mini. I use an M2 Max studio at work and it actually feels laggier to me lol
 
The price difference between a low-end ($599) M4 Mac mini and a low-end ($2000) M2-Max Mac Studio is significant, but there are some real differences that affect performance. The Mac mini will beat the Studio on single-core CPU performance according to benchmarks. But on multi-CPU performance, the 10-core mini is only about 83% of the speed of the 12-core Studio.

GPU performance is a much bigger difference. The 10-core mini is about 40% the speed of the 30-core studio. That drops to about 30% of the speed under throttling for thermal management. That's a huge advantage to the Studio if you are a professional processing image, video, graphics, or AI data.

Memory bandwidth on the mini is 120 GB/s and 400 GB/s on the Ultra. Again, this is a huge advantage on a studio for professional workloads.

Ethernet on the mini is 1GB. It is 10GB on the studio. If your work requires you have multiple machines and move data between them, you should have standardized on 10GB ethernet five years ago.

The SSD on the mini is half the size of the Ultra. IOW, the low-end mini is nearly useless for any task requiring lots of media or large AI models unless you buy external storage.

Now, if you aren't a professional or you are are a programmer doing Mac or iOS development that doesn't require large data or sustained computation, the mini is a fine machine and will cost you less than the Studio. But, if you are a professional who needs the greater resources of the Studio, the mini is a false bargain.
Im just saying the mini is a great bargain for me coming from a maxed out 2017 iMac . it blows it away speed wise. yes I could use more ram for sure but 16 is functional. It no longer feels crippled like the 8gb base models were. Most of my progressional work load is In the Adobe Design suite which uses more CPU than GPU. GPU would be more helpful for processing RAW photos I suppose but I'm not like blown away by how fast my Mac Studio at work processes anything versus my base mini at home. SSD is a moot point cus I'd never pay apples prices. I buy the highest speed and largest NVME I can buy and put whole whole home folder on that. It's a desktop so no need to not stick with external storage. I do use the 10 gig ethernet on my studio at work but I actually have a 10 gig external thunderbolt card sitting in a drawer from my old iMac that I can always use if needed at home
 
  • Like
Reactions: vantelimus
I am not holding out hope that there will be an M5 iMac in 2025.

It's been nearly half a decade since Apple released the M chips and despite the good news this year of the base RAM increase, I just no longer have hope that Apple will make an iMac with a base 1TB of storage (like it used to) in a 27" screen size (like it used to), with a more powerful — a la M5 Pro — chip option (like it used to).
 
  • Like
Reactions: turbineseaplane
Golly, I may have to update my Intel Mac mini. It’s quite happy without AI tho … The MacBook Pro circa 2015 still lives, speakers are blown, but gets OS updates now & then, Firefox often, but given how little I use it… fine.
 
Considering their storage and RAM upgrade costs, it seriously is becoming cheaper to get a bunch of Minis. If your workload can be split/clustered, makes way more sense to get a couple maxed out Minis than a single Pro.

I think eventually this will be what we get. Apple released some open-source software (their openELM LLM) and if you read the code, there is a comment inside that says "We use Liniux because SLURM does not yet run on macOS."

See here https://en.wikipedia.org/wiki/Slurm_Workload_Manager

So it seems Apple is working on porting SLURM to macOS and intends to use it at least internally. I think eventually we might see user-facing clusters. Likely not clustered $500 minis but you would cluster a few Mac Pros, not with Eithernet but rather Infiniband networking (three terabits per second.) Networking with Minis even on 10G Ethernet is too slow.

Use cases would be AI training and video rendering. But we know for certain that Apple plans to do this internally.
 
  • Love
Reactions: CarAnalogy
Why would anyone buy a M4 Max or M4 Ultra Mac Studio, if the M5 versions are coming out shortly after the M4 Mac Studios launch?
A mac studio user here.

Our company needs desktops, as computers are working 24/7 and laptops arent designed for that and the cost is too high for no advantage at all. So we need Max or Ultra but desktop.

Beyond desktops needs, if you need huge GPU performance, your only option is the Ultra, so worth the wait.

This is a niche market for apple, so they know clients wont go too far. Until clients needs the latest and greatest and switch to Pc. (As happen with IA needs)

Apple is selling the cheaper Max option quite late, as it come up together with the Ultra favor, harming overall desktop users, but I guess, production wise, they cant do it better.

What amazes and bothers me me is that having the Ultra ready takes Apple 8 months.
 
.Logistics, engineering and manufacturing schedules, market demands, etc. I'm sure Apple sells a ton more machines with M and M Pro level chips than M Max and M Ultra chips. It makes sense for them to spend more time and effort on the much larger market for the affluent consumer than on the high-end content creator and software engineering niches.
No, it doesn't. Coming out with the high end first would force early adopters to buy the most expensive model. Their whole pricing and product strategy is built to promote upselling, yet they shoot themselves in the leg by offering the more expensive ones with older tech. It's no coincidence that everyone else does this. And the mac studio is low volume for this exact reason.
 
  • Like
Reactions: mode11
The price difference between a low-end ($599) M4 Mac mini and a low-end ($2000) M2-Max Mac Studio is significant, but there are some real differences that affect performance. The Mac mini will beat the Studio on single-core CPU performance according to benchmarks. But on multi-CPU performance, the 10-core mini is only about 83% of the speed of the 12-core Studio.
Benchmarks usually scale well on multi core, because that's they are designed for. Real life workloads don't always behave like this. Multi core workloads with less parallelism will favor the m4. Benchmarks are the best-case scenario for scaling, not the expectation.
 
This happens every year. People were buying M2 Mac minis instead of the 8 month old Mac Studio M1 and right now people are buying M4 Mac minis instead of the old M2 Mac Studio.... Apples release schedule kinda blows for high end users. I'm thinking I may just work from low end machines going forward and update them for frequently. The $499 Mac mini m4 is more powerful than any computer I've used so I'm wondering why I should spent 4+ times the cost on a new studio with older chipsets.
True, but there are a few types of high-end users for whom older but high-end generation chips are still better than newer lower-end ones: any type of professionals that need the GPU power for their work (3D modeling, AI etc. - I'm not mentioning gaming, because on chips like Apple Silicon even AAA game titles can work very very well as long as they are optimized for Apple Silicon, and games are just a bonus if the user is buying the machine mainly for their work anyway 😄 -).
For example, as far as I understand, a Mac Studio with an M2 Max is still better for that type of work than a new Mac Mini with M4 (i.e., non-M4 Pro, because I've seen test results showing M4 Pro surpass M2 Max for GPU work for both single-core and multi-core tasks).
 
A mac studio user here.

Our company needs desktops, as computers are working 24/7 and laptops arent designed for that and the cost is too high for no advantage at all. So we need Max or Ultra but desktop.

Beyond desktops needs, if you need huge GPU performance, your only option is the Ultra, so worth the wait.

This is a niche market for apple, so they know clients wont go too far. Until clients needs the latest and greatest and switch to Pc. (As happen with IA needs)

Apple is selling the cheaper Max option quite late, as it come up together with the Ultra favor, harming overall desktop users, but I guess, production wise, they cant do it better.

What amazes and bothers me me is that having the Ultra ready takes Apple 8 months.
Tying the studio to the ultra was a strategic mistake. Not being able to get the m4 max in a desktop is sad.
 
  • Like
Reactions: mode11 and cocoua
M9 will be even more impressive, but who knows what they will do after that?!
I know! It will be called MX Apple marketing will isist it is pronounced M Ten.

And it will be most powerful mac ever made! Until the following year MX jaguar comes out and that will be the most powerful mac ever made...etc...etc....etc
 
  • Like
Reactions: Porco
Kuo expects the standard M5 chip to enter mass production in the first half of 2025, followed by the M5 Pro and M5 Max chips in the second half of the year. Then, he expects the M5 Ultra chip to enter mass production in 2026.

This mostly checks out to me.

But. The MacBook Air with M4 is coming this spring. That raises the questions:

  • why not skip the M4 on the Air?
  • what device will come with the M5 first? Only the iPad Pro seems to make sense, and… I honestly don’t get it. Who’s pressuring Apple to make the iPad Pro faster?
 
I really wish Apple would release these chips starting at the top and work down. Release the M5 Max and Ultra first in the Studio and Pro in summer 2025, then the M5 Pro in the fall with the MacBook Pro and Mac mini, then the M5 MacBook Airs and iMacs in the spring 2026 and then just continue that trend. It would make so much more sense.

That doesn’t make sense from an engineering (lack of experience doing the most complicated SoC) or economics (lack of economics of scale) perspective, only from an “I want it” perspective.

No CPU vendor starts at the high end.
 
Last edited:
I would like to hear about some people who REALLY CAN use all the power of these chips. I know I live a sheltered life, but my M1 machines can play 4 videos like YouTube simultaneously. Downloading and converting a 10 minute video to another format takes on my a minute or two.

What the heck are people doing that need this kind of processing power? These things are like Ferrari’s on a road with a speed limit.

Same goes for iPhone. I have 11 pro max, and I have yet to find anything that makes me wait while it processes.

I’m not talking about the pros who use their devices to generate revenue, just the rest of us.
That is why I think the Studio and Pro should be first. Why does a Mac Mini or MacBook Air etc get the latest?
 
That doesn’t make sense from an engineering (lack of experience doing the most complicated SoC) of economics (lack of economics of scale) perspective, only from an “I want it” perspective.

No CPU vendor starts at the high end.
No CPU vendor has their X+1 gen laptop chip beat the best of the best desktop X gen either. In ANY workflows.
 
  • Like
Reactions: cocoua and ryanarrr
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.