First review about sound... fan noise

[IvyKing]

I got some fan noise out of my M4 Pro Mini by with a couple of installs from MacPorts. It wasn't so much the building of the port itself that was taxing the Mini, but all three of the ports (nedit, gnuplot, gerbv) requiring new compilers for the ultimate build. The gerbv build required the Rust compiler, which took up ~5 GB of storage space. Memory usage peaked about 19GB for apps and 16GB for file caching - the 64GB came in handy.

As for the fan noise, it was low enough not to be annoying, less than an Intel MacBook Pro.

NB: I ran a floating point heavy test program on the M4 Pro that I previously ran on my M1 3 1/2 years ago. Speed up appears to be ~40% for a single core/single thread task going from M1 to M4 Pro. Speed up from a 200 MHz Pentium was about a factor of 400 with the same source code and input file, with Pentium test done late 1998.

 

[Confused-User]

I think I have what I came here for to make the right choice for a silent machine that will sit all day feet away from me in my living room. When you compare the speeds (geekbench), the non GPU performance of the base m4 is on par with the m2pro, and the m4pro is only about 1.5x faster CPU and 1.8x GPU. Coming from a 7th gen i5 intel nuc, I’m still getting a 5x bump in cpu and GPU and so I’ll take the silent machine.

If you're looking at a media center machine, the notion that you need more than the base m4 in the first place seems highly misguided. By right paths or wrong, though, I think you came to the right conclusion. The base M4 will feel like lightning compared to your Intel.

 

[Patrickf73]

I’ve read all the threads here and watched many videos about the Mac Mini M4 Pro. From what I understand, it tends to get loud when the GPU is under heavy load.

If you need a powerful GPU, the Mac Mini might not be the best choice—you should probably wait for the Mac Studio M4 Max or Ultra.

Music producers, on the other hand, rely more on CPU and RAM. I don’t think the fan will kick in while using Cubase, Logic, or Pro Tools. And if someone ever manages to max out the CPU with a music program (which seems almost impossible to me), they’ll likely need a Mac Studio or a Mac Pro anyway.

It’s always important to buy a computer that fits your needs and has enough headroom to handle future demands. This ensures that your projects run smoothly without pushing the system to its limits, giving you more reliability and room for growth. That’s why I don’t pay much attention to benchmark tests. Of course, they’ll push the Mac Mini to its limits, causing the fan to spin up—but that’s not a realistic use case for most people.


I’ve ordered the Mac Mini M4 Pro (48 GB RAM and 1 TB SSD), and I don’t think I’ll ever hear the fan. I’ll let you know.

 

[SkweeBop]

Taking any of the YouTube "influenzas" input into anything that goes beyond how to make a-thumbnail-with-the-silliest-face-without-looking-like-you're-not-well is not the best idea. I'll wait for real user feedback from large sample of users before I decide whether to pull the trigger or not.

I agree with this in principle, but I may have seen what I was hoping for in the video, as its primary use would be for tracking instruments and singing in a home studio environ, and the reason I had to stop using my 2015 MBP for studio work. The fan just simply ruined takes.

It would seem that the M4 standard stood up to the dude's cacophonous monster track, which is exponentially more over the top than any template or session I'd be recording.

My M1 has been perfect, and I'm not in a rush to replace it, at any rate.

 

[ravenofdoom]

If you're looking at a media center machine, the notion that you need more than the base m4 in the first place seems highly misguided. By right paths or wrong, though, I think you came to the right conclusion. The base M4 will feel like lightning compared to your Intel.

It’ll be doing more than serving as a media center. I will occasionally be doing some light image and video re-encoding, large file transfers to and from NAS on 10Gbe, hosting/sharing of some files, emulated games (older consoles). There is no need for all the added GPU cores on the m4pro though, which seems like is one of the primary sources of the added heat and fan noise when maxed out. I primarily wanted TB5 and the faster read/write speeds on the SSD for direct transfers of an m2 sata down the line and some level of future-proofing, but in practice TB4 is probably sufficient since most my data transfers won’t likely be a plug-in on TB but over NAS.

 

[Confused-User]

It’ll be doing more than serving as a media center. I will occasionally be doing some light image and video re-encoding, large file transfers to and from NAS on 10Gbe, hosting/sharing of some files, emulated games (older consoles). There is no need for all the added GPU cores on the m4pro though, which seems like is one of the primary sources of the added heat and fan noise when maxed out. I primarily wanted TB5 and the faster read/write speeds on the SSD for direct transfers of an m2 sata down the line and some level of future-proofing, but in practice TB4 is probably sufficient since most my data transfers won’t likely be a plug-in on TB but over NAS.

The reencoding could, possibly, push it enough to spin up the fans, depending one what software you're using. But you could always configure it not to do that. Nothing else you're talking about, especially the I/O (to disk and network), would ever get you close no matter how hard you pushed it. TB4 is very definitely more than fast enough for anything you're talking about. In particular, it vastly outstrips the performance of your network and your internet connection.

Futureproofing with TB5 is not worthless but it's a badly false economy. Even if you discard your new M4 in two years (instead of selling or repurposing) and replace it with a base M6 you'll have TB5 and more cash in your pocket. You'll also have substantially better SC performance, and though MC will probably be inferior, you'll probably never notice.

 

[Fishrrman]

One review that seems to examine "the noise issue" quite well:
(apologize if it's been posted already)

Apple Mac Mini M4 review - Smaller, faster and louder

Notebookcheck has tested the new Apple Mac Mini with the M4 SoC, 16 GB RAM and a 256-GB SSD.

www.notebookcheck.net

 

[ravenofdoom]

One review that seems to examine "the noise issue" quite well:
(apologize if it's been posted already)

Apple Mac Mini M4 review - Smaller, faster and louder

Notebookcheck has tested the new Apple Mac Mini with the M4 SoC, 16 GB RAM and a 256-GB SSD.

www.notebookcheck.net

This is the most detailed review I've seen about heat and fan speed -- especially on the base m4. What I'm concerned about is this is the first review that says they were able to generate very audible fan noise without manual trigger, which is counter to people here who have tested and other reviewers (albeit of the youtube variety):

"... we were able to push the Mac Mini harder, which was met with a maximum fan speed of 2,950 rpm—or a clearly audible 45.1 dB(A). However, there is still a lot of room for improvement in this respect, as the fan can reach a maximum of 4,900 rpm (59.4 dB(A)) which we could only trigger manually"

Moreover... in their testing, they've indicated the m2 mac mini reached a peak of 29.4 db(A) as compared to the m4 mac mini's 45.1

EDIT: it's not clear to me what "load maximum" means -- what specific tests / what was done to trigger that level of wattage and fan noise. Playing Cyberpunk 2077 ultra I would imagine would max out every GPU and CPU core (39.6 W), so I don't know what additional test would eek out additional wattage to get past that and into this 45.1 db(A) tier / 62.5 W

 

[EdwardC]

One review that seems to examine "the noise issue" quite well:
(apologize if it's been posted already)

Apple Mac Mini M4 review - Smaller, faster and louder

Notebookcheck has tested the new Apple Mac Mini with the M4 SoC, 16 GB RAM and a 256-GB SSD.

www.notebookcheck.net

That's a very detailed review!

 

[AAPLGeek]

This is the most detailed review I've seen about heat and fan speed -- especially on the base m4. What I'm concerned about is this is the first review that says they were able to generate very audible fan noise without manual trigger, which is counter to people here who have tested and other reviewers (albeit of the youtube variety):

"... we were able to push the Mac Mini harder, which was met with a maximum fan speed of 2,950 rpm—or a clearly audible 45.1 dB(A). However, there is still a lot of room for improvement in this respect, as the fan can reach a maximum of 4,900 rpm (59.4 dB(A)) which we could only trigger manually"

Moreover... in their testing, they've indicated the m2 mac mini reached a peak of 29.4 db(A) as compared to the m4 mac mini's 45.1

I simply never trust any YouTubers or "content creators", especially when they can never bother to publish any actual numbers on their "created content".

Half of them just keep doing these silly FCP or Premiere exports without even realizing that it's never going to push the CPU/GPU as the media engine does most of the heavy lifting.

 

[!!!]

I got the base model M4 mini, when pushing it hard (gaming through Whisky/wine) it draws ~46W, the fans spin up to around 2500-3000RPM. 20cm away the Apple Watch 7 reads about 41dB, noise floor is 31-32dB. Ambient 25.4C. AVG CPU is ~89C, hottest CPU (performance core) 103C. At this level I'd call it noisy/obtrusive, you're probably gonna want to slap on headphones if you had to keep it at this level and work long term.

I'll call the machine quietly audible at 33dB (non obtrusive wooshing), drawing 32W, fan ~1450rpm. (I capped fps to 45, 4 P cores maxed and gpu at around 70-80%) This would be a pretty moderate load.

The machine is practically silent at anything below say 27W, fan <1250RPM, it disappears into my noise floor. (4 P cores still 90%, GPU at 15%) It's about as silent as the hissing of my Neumann KH80 studio monitors, at 70cm away its a struggle to hear anything from the mac or my speakers over my mild tinnitus.

I let the machine idle and the fan reports 1000RPM. With my ear up against the machine you can still hear the fan hum (almost like quiet mains transformer humming) but when say moving my mouse around while watching a 4k60 YouTube video there's no sharp coil whine noise or anything, so it's very well behaved.

I've tried raising it so the intake/exhaust can breathe better, but the increase in noise level more than offsets any gain in cooling efficiency. Hope this helps.

Thanks for your review. That's very disappointing. I can't wear headphones, and 45 fps is very sad in 2024.

 

[ravenofdoom]

@loliii123

This aligns to the detailed review with heat and dB. Sadly it is starting to look like Apple can’t outrun simple physics of making the box smaller without pumping that fan at louder rpms. The peak load max watts on the m4 is 62.5 compared to 51.5, so that may be what is accounting for the db delta of 45.1 to 29.4. Nonetheless, that’s a big swing in fan noise tolerances for the max heat output of the box — I have no doubt that if they made the box bigger and added a larger heatsink and larger fan that could push more volume at lower rpm they would have gotten the noise down at peak (which comes on even sooner for m4pro). I think they made a conscious choice of form over noise function here when it comes to peak load.

 

[AAPLGeek]

This aligns to the detailed review with heat and dB. Sadly it is starting to look like Apple can’t outrun simple physics of making the box smaller without pumping that fan at louder rpms. The peak load max watts on the m4 is 62.5 compared to 51.5, so that may be what is accounting for the db delta of 45.1 to 29.4. Nonetheless, that’s a big swing in fan noise tolerances for the max heat output of the box — I have no doubt that if they made the box bigger and added a larger heatsink and larger fan that could push more volume at lower rpm they would have gotten the noise down at peak (which comes on even sooner for m4pro). I think they made a conscious choice of form over noise function here when it comes to peak load.

I think Apple's goal is to drive Mac Studio sales for intense workloads and the thermal limitations of the new Mini are a small but important step in driving those sales.

 

[ravenofdoom]

I think Apple's goal is to drive Mac Studio sales for intense workloads and the thermal limitations of the new Mini are a small but important step in driving those sales.

I think some of this is also marketing. Can they really continue to call the m4 mini a "mini" if the chassis, circa m2 variant, is almost that of the studio (i believe they had the same w x l, and only differed in height by 2.25"

 

[SmallDane]

I think some of this is also marketing. Can they really continue to call the m4 mini a "mini" if the chassis, circa m2 variant, is almost that of the studio (i believe they had the same w x l, and only differed in height by 2.25"

Apple has long had an obsession with making products smaller, thinner, lighter. For some products it makes sense. For others not so much.

 

[AAPLGeek]

I think some of this is also marketing. Can they really continue to call the m4 mini a "mini" if the chassis, circa m2 variant, is almost that of the studio (i believe they had the same w x l, and only differed in height by 2.25"

Yes, because Studio is still quite larger by volume and weight.

Even then, a smaller design like 6.2x6.2x2.4 would have mostly achieved that same marketing effect while still maintaining the thermal capabilities of the prior design.

 

[streetfunk]

I do audio work. (not DAW production related)
i have a M4Mini pro, 14c, 24GB RAM // Room Temp is around 23° // running two screens, 2K

i heard the fans at a CPU temperature of around 63°.
My main app runs single threaded only.
That main loaded core was under a ~40% CPU load. ( the read out of the plugin host).

The M4pro definitly runs hotter than a M2pro. It heats up way quicker than my M2pro, running same projects under a way higher cPU load on the M2pro. Fans kick in very early on the M4.
Interestingly the power drain is under my load somewhat higher. While the temterature raises way more in relation.
Which tells me something. Running the M4pro under low load, -just some internetting-, the temperature is higher too. Around 52° here. (some apps beeing open in the backgroundm running idle)

I just saw here some folks guessworking that they might not hear the fans at all doing their common DAW work. I Personally i would not bet on that.
The type of the fan noise is ok. Aslong as it´s a steady controlled noise it´s sorta -halfways- ok for me.
And that´s the case in my usecase szenario.
imagine we had a similar CPU load on 80% of the cores ? Then ? i don´t know.

if you want to be sure to have a mostly quite M4desktop under most conditions, consider a coming M4Studio max to be a saver bet.

my M4 sits behind my Masterkeyboard on the same board, but i have a 2nd level board directly above, overlapping very slightly with the Masterkeyboard, 5cm above the masterekyboard.
So there is only a small gap -2 fingers in width- the noise can come thru to me "directly".

consider the M4pro to be a slightly noisy one in your studio.
Don´t take my words as something final. Just the impression from one guy vs. his (my) setup and usecase.
( semi calm noise conditions outside. No permanent audible streetnoise outside, as it was in my old place)

the cooling situation seems not adequate enough in the M4pro.
ckeck if your workloads would require sustained high loads, or just some peaks shorter in time ?
i would walk from there vs. audio work.


my CPU was running steadily at around 3.9Ghz. The loads are quite constant.
keyboard playing with very complex patches with permanent -FX based- realtime manipulation of the sound.
Thats not a very high load at all. But i hear the fan.
( i broke my fan speed setting in Istats, no clue how to get it back.....so, can´t tell you fan speed numbers. i think i saw them bevore at around 1700rpm)

my sum up: i´d not put too high bids on the M4minipro to be a hush computer for audio work.
inform yourself first in case "that one" counts !
Strangly enough, the reports are quite wide spread vs. how noisy the M4/pro is.
The last words on this are not spoken yet, in my opinion.

 

[ravenofdoom]

I'd be really interested in seeing similar thermal and db reports for the m2max studio -- anything out there that comes close to the level of benchmarking details as the most recent one shared?

EDIT:

Found it here -- same benchmarking site:

Apple Mac Studio 2023 review - The M2 Max also impresses in a desktop device

Apple has updated its desktop computer, the Mac Studio, which is now fitted with the faster M2 Max or M2 Ultra SoCs. We have tested the base model with the M2 Max and the 30-core GPU as well as 32 GB RAM.

www.notebookcheck.net

m2Max studio -- under max load, came out to 25db at 111.5 w !!! That's some effective thermal management with appropriately sized heatsinks and quiet fans at low rpm. Again for comparison:

	m2 mini	m4 mini	m2max studio
idle min	24.7 db / 2.2 W	24.1db / 2.6W	23.1 db / 7.2W
load avg	24.7 db / 28.7 W	24.1 db / 31.5 W	23.1 db / 63.5 W
load max	29.4 db / 51.5 W	45.1db / 62.5W	25db / 111.5W

Apple has previously done a great job (with the m2 mini) at keeping the unit quiet even under max load -- the same true for the m2max studio... the acoustics from idle to max load are nearly the same. They simply could not sustain this acoustic profile with the directive to shrink the mini when put under a high performance / max load.

I may still be tempted to buy the m4 base and not a mac studio because I suspect I will never exceed the average load scenario, so it will likely hover in the 24-25db range. Also, I intend to put this box inside my media cabinet, and so I want the unit that will minimize the miniature heater effect inside the cabinet to not heat up my other electronics, and so while a mac studio will always be quieter, it will always output more net heat in there given it consistently runs at higher wattage. I am curious what the frequency curve looks like for these units... is the noise in higher or lower frequency ranges... since these are in my media cabinet, lower frequency would be more annoying and act more like a subwoofer.

 

[Confused-User]

EDIT: it's not clear to me what "load maximum" means -- what specific tests / what was done to trigger that level of wattage and fan noise. Playing Cyberpunk 2077 ultra I would imagine would max out every GPU and CPU core (39.6 W), so I don't know what additional test would eek out additional wattage to get past that and into this 45.1 db(A) tier / 62.5 W

You could load other parts of the SoC as well. I don't know of any real-world application that can do this, and it's a little hard to imagine a real-world use case for doing it with multiple apps either, but you could run code that additionally exercises the NPU and the ISP (not sure how, since there's no built-in camera, but I'll bet it's doable), runs the media encoders and decoders, and pushes memory bandwidth to the max. But beyond that you'd also push the 10Gbps ethernet to the max (that alone should eat ~2.5-3W, maybe more?) and attach peripherals to the TB ports that each suck up the maximum power they can, thereby driving the internal power supply to its maximum (since you should expect 15-20% loss there, with that loss translating directly to heat). Make sure there are three external displays attached, to run all the display controllers.

That's probably the worst you can do.

 

[ravenofdoom]

worse than m2 starting in low-mid to mid range

 

[Hallonskalle]

Maybe some of them have better or worse thermal paste job?
I have swapped CPU on PCs a lot of times and it does make a difference if you dont get the pasting right.
Maybe the latest Thermal Grizzly Kryosheets would be nice? Doesnt look like much thermal paste from the teardown videos..

 

[ravenofdoom]

I'm starting to think the move to shrink the mac mini was also driven by the need to say it's carbon neutral and to reduce quantity materials in the process, e.g. aluminum. The energy argument should be to reduce not only production energy burden but overall energy usage on the end-user... they've probably achieved this if you compare processing task times per wattage form the m2 to m4 mini; however, the non-idle watts consumed seems to be going up... which is not unrelated to increased fan noise. Again, marketing drivers. I'm not saying the m4 mini isn't a great machine and won't suit the majority of non-power users, just that these design changes have multiple drivers -- design changes can often be tipped for form/promotion over function. I do think that a less aggressive change to the dimensions and tweaks to the heatsink and fan in a slightly larger box would have resulted in an improved thermals, acoustics, and performance -- and especially so in the m4pro. They went with an aggressive dimension/redesign change that tracks with their current marketing campaign. I think we have some sense now on the impact this has on heat and fan noise.

 

[Confused-User]

worse than m2 starting in low-mid to mid range

FINALLY some good hard data. Where is this from and what fan/performance level(s) does this cover?

If this is representative of the average M4 mini fan, that's unfortunate and a little surprising. It's still a great machine, but that would limit its use (well, mostly the Pro's use) in a few cases.

 

[ravenofdoom]

FINALLY some good hard data. Where is this from and what fan/performance level(s) does this cover?

If this is representative of the average M4 mini fan, that's unfortunate and a little surprising. It's still a great machine, but that would limit its use (well, mostly the Pro's use) in a few cases.

This was from the existing data leverage from the Notebookcheck website:

Apple Mac Mini M4 review - Smaller, faster and louder

Notebookcheck has tested the new Apple Mac Mini with the M4 SoC, 16 GB RAM and a 256-GB SSD.

www.notebookcheck.net

If you scroll 2/3 the way down there's a noise/emissions section and you can actually select multiple machines to compare the noise curve to. You can also select the M2Max Studio, which has a better curve than either of these two. It's a little cluttered / hard to read what the gradients are -- presumably these are measured at different workloads and averages at the right are a combination of all.

EDIT: I wish the publisher/author also did the same tests for the m4pro mini, but I think we'd just see higher wattages and fan noise on both the Cyberpunk and Load Maximum scores... would be curious though if "load average" would show a higher than baseline db.

I think you really see a difference in quality and transparency of reviews from a site like the above as opposed to the first 7-day media onslaught of insta/youtubers that flood social media whose livelihoods are solely driven by getting free test machines to always seem wowed by certain manufacturers and get those like-button smashes. It's hard to look at all those social media outlets in aggregate and not draw conclusions, but there's no substitute for consistent and methodically generated measurements from those who do this across multiple manufacturers' products.

 

[!!!]

Being able to push the fans loud is actually good: it means the CPU/GPU can be pushed to higher clocks while maintaining temperatures within bounds. This can be adjusted with the different modes, which is actually great (instead of needing a third party utility to adjust the behaviour, except if for any reason you need more finegrained control). In short if fans were not working full at no scenario it would basically mean that apple could push the mac mini more, but it did not.

It actually isn't a good thing. It's a desktop, it should remain quiet, regardless. I have a desktop with a 5800x and a 3080. This PC, while idling, consumes more power than the base M4 at max power. Even when gaming I can barely hear anything, and most of the noise comes from the fact that the 3080 is consuming somewhere around 300W.

As for "fine-grained control", well my use case is only that I do not want to hear it. So I want the fan to ramp up as much as possible so long as it's not audible to me. Apple apparently does not provide that option: low power is too little performance, and "automatic" seems to still make quite a bit of noise, from what others have said.

I can understand that if there's no fan at all or if the system stays silent all the time, that we need to accept the need for CPU trottling. but ramping up the fan(s) and throttling the CPU's clock speed is a total nogo in my book.

Indeed. It's a desktop, there's zero reason for there to be this kind of compromise. I thought the M-series minis would finally end the long-suffering of the Intel mini's awful fan noises, but alas that doesn't seem the case. I mean I can do what I used to do and lock the fan speed to just barely inaudible, but that's not very "Apple-like" and I really expect better at this point.

Note:
A big problem with this discussion involves the fact that people's tolerances for noises are different, and people sit at different distances from their minis.