One of the studio YouTubers I follow mentioned Nuendo and Cubase (not sure which versions - probably 13 if not 14) were running poorly in stress tests with an M4 Max Mac Studio from Apple.
He wasn’t sure why, but also mentioned that the DAWs ran fine on M4 Pro and regular M4 CPUs.
Has anyone here had similar issues with the M4 Max CPU?
I’m compelled to first point out that “YouTubers who make statements followed by ‘I don’t know why’” are probably not the best choice for reliable, unbiased, objective information.
Not that it really matters, but the “stress tests” I’ve seen are multiple tracks with the exact same plug-ins, without any analysis of innate plug-in instance memory-management, or even something a critical as thread management. There was one dude who did some variations between tests, which does “help,” but it doesn’t give the user any “real” information other than “that’s what happened when he did it on what he said the systems were.”
What I have found personally in different tests on different base M series processors (and even Intel Xeon-based MacOS systems) is that for Cubendo, performance will vary based on the performance:efficiency core ratio of the processor model. The M4 Max is a 16 core 12p:4e model, similar to the M3 Max. This has been a rock-solid, high-performance (stunning, really) solution for me. One should note I opted for 128G RAM on my main MBP which I use as my main rig; RAM matters.
As a corollary, @Ben_H and I over on the SpectraLayers forum were testing performance differences between Windows and MacOS on different processor types in regard to FFT efficiencies. My M3 Max performed quite well even with large (32k) FFT frames. But this was in part due to the PFFT (that’s real, despite the tongue-in-cheek name) processing on the Mac. In fact, and here’s the interesting part, my Xeon W 8-core 3.2 in my “backup” iMac Pro (64G ram) performed AT THE SAME SPEED with our test pink-noise project.
The reason I bring all this up is that when making CPU model decisions, I think it’s important for you to do as much personal research as possible, particularly when building for a specific application. In the case of SL, the best performance came from using an OS that explicitly supported libraries required to boost performance and NOT the CPU.
Just giving you my buck-o-five 
I saw that one. “A lot of errors and I don’t know why.” Sounds to me like installation or configuration issues rather than a Cubase incompatibility with the M4 Mac Studio Max. I will find out for myself soon enough…
Same - ordering my new Mac Studio M3 Ultra fully loaded tomorrow. What a beast! And just to drive home this YouTube stuff, these guys say whatever they have to say to get clicks. I didn’t even go looking for one, but in validating some specs I saw a link to some dude talking about “The HUGE Problem with M3 Ultra” where he looks at some graphs and talks and can’t even add/subtract the numbers right. Like, he literally had nothing to say other than “it costs too much” and then makes up straw-man arguments for this and that, and “people can’t afford to upgrade again next year” and just inane, trite filler. I really don’t see how the guy even qualifies to make the video, but there’s like 78k views, so hyperbole, specious conclusions, and general misinformation rules these days.
Sorry, it’s just such a waste, and that kind of stuff bugs me. Anyway, looks like a couple of weeks lead-time, so I’ll be waiting patiently.
@Reco29 go ahead, make fun!!
@Thor.HOG Nah, you go ahead and have fun with this beast loaded up with 512 kb RAM 
You kid, but my first PC had 512k! I had to get an “AST SixPack” add-on card to get 640k. Complete with “Sidekick” resident memory app! I swapped out the 8088 for an NECV20 and had the fastest PC on the block!!
I hear you. You can spend hours weeding through videos and still only garner a sliver of any useful info. Though I suppose a big part of it is the test/review might not apply to your particular workflow.
I went ahead and ordered the M4 Max w/64GB RAM yesterday. More than plenty for my workflow and the unit will probably age-out due to planned obsolescence before I outgrow the power and memory. Will report back on any interesting findings.
It was just seconds before I decided to place an order for my new M4Max Studio when I saw that video on YT.
"Suck! but sus"I thought and I rushed straight here expecting whoever reported such a thing, only to find you guys also wondering what’s going on.
Then I turned to google. I managed to find one guy on the VI-Control forum that have tested M4Max Mac Studio with Cubase 14 and logic. Interestingly his results are even showing the opposite to Barry Johns’ statements (, at first, then he changed something and both DAWs benchmarked as expected) . M4 Mac Studio speculation thread became discussion of the actual | Page 19 | VI-CONTROL
Now I’m confused even more.
I read (what I found to be the relevant parts of) that thread with interest. I’m not trying to convince anyone of anything, but I will make some observations which may provide a metric by which to weigh the relevance of these “tests,” their stated requirements, and resultant “conclusions.” The mod themselves indicated “I run Logic at a 64-sample buffer, except for rare circumstances when I have to raise it to 128. 512 seems excessive to me.” Out of context, that statement doesn’t really mean anything - does he only mean “64 samples” when recording? One can argue that 512 samples (depending on the interface) for playback doesn’t even matter. Yet, that simple setting can have dramatic impact on overall performance and is absolutely relevant to project performance. One could reasonably expect the mod to provide more context, unless, who knows?
There’s also a matter of workflow here… The solution for “I can only put 100 tracks of my fav synth + convolution reverb in my project before I get audio dropouts” may NOT be to compare different OS/DAW/CPUs, but may rather be, “then don’t do that.” Bounce to audio or manage the project better. What I’m missing from literally ALL of these conversations is a real-world reference. I mean (and I think this is the important part), if these guys are contemplating spending all this money on new hardware, then why aren’t they citing existing projects representative of the technical struggles they seek to remedy with the new equipment? Logic dictates that in these discussions the requirement predicates the purchase, yet we see people manufacturing project structures to see if what they already purchased was the right choice, or making up test projects for YouTube in order to pretend they have relevance. It doesn’t make much sense to me.
I’m putting my money where my mouth is, and purchasing the Mac Studio M3 Ultra 32/80 core. I would have done it yesterday but the markets were closed for Good Friday.
If you want the M4 Max, I think that is generally a strong solution. Buy as much RAM as you can afford, and try for the 16-core (12p:4e). My M3 Max is fantastic, and I have no issues at all with it. I have other reasons for buying the Studio M3 Ultra, not the least of which is to cycle out my iMac Pro Xeon which I’ve had for several years now, and am making my determination based on repeated collections of empirical data with real-world projects and software.
Go for it, man.
Ok…It turns out I’m the first one to get an M4Max(12p+4e) Mac Studio. It’s an 128GB 2T one on my hand now. Just arrived today and I’ll test it.
While not an M4 Max Studio, I’ve been running N14 on my MBP M4 Max since N14 was released and it has been very smooth so far.
I had some minor issues with N13 which I didn’t investigate further as I had just gotten the M4 Max MBP in Feb. I basically had crackles when recording a single mic at a buffer size of 128, which I usually used. So had to bump it up to 256.
On N14, I’ve been recording at a buffer size of 64 for 2 full days and have nothing to report.
I also have a scoring session running at 64 with about 30 tracks of Kontakt, EastWest Opus etc. with heavy libraries and the session has been quite smooth so far. On my previous M1 Max, I usually ran sessions at 128 during the initial phase of writing. Had too many crackles at 64 even with just 1 or 2 instruments.
Not sure how the studio M4 Max differs from the MBP.
It take some time to install all my plugins and libraries to open a real project so I just got a quick test on threading behaviour first. It seems Nuendo 12 and plus is working quite well as expected so far.
Test method:
64bit-float engine, ASIO Guard High, 256smpls@44100Hz
Audio track with specific 44.1kHz24bit wave file on it, Record/Monitor off, each track load with one SpecCraft 1.4.1 instance on the insert slot, Oversampling: On, 8x(Slow). Duplicate the track(means that instances are paralleled) to find the maximum track count that can be replayed for 3 minutes without a glitch.
M2: 20 | M2Max: 39 |
M4: 31 | M4Max: 87 |
All cores are nearly maxed out (12 P-cores ~3.9GHz, 4 E-Cores ~2.6GHz) meaning threads in this test are distributed equally. No threading abnormality observed so far.
Notice that this is a very artificial working condition and is based on a certain plugin.
Not until all necessary plugins and libraries are installed can I test scaling a real project.
Exactly , and that’s why most of these you tube copy/paste tracks tests are irrelevant.
It’s alos a reason why DAW bench was started by Vin(Tafkat) many years ago . There’s now going to be a 3rd test which is the bus test which replicates an actual mix type project along with the smaple based Kontact test and the DSP plugin test. Pete at Scan Computers in the UK uses this now as a good petric for performance and I know Pete at Microsoft is heavily involved too.
Unfortunately , not many Mac owbers seem willing to do these tests. I developed the Daw bench Bus test from one of my own real world mix sessions along with Vin and asked for Apple users to run the tests as a comparison to the windows machines, no one… yes no one took up the task… that was a year ago.
This is why there are no Apple results in these tests.
M
I’m totally down for helping. My Studio M3 Ultra should arrive in 2-3 weeks, and I can perform a controlled test between my MBP M3 Max, the new Studio, and my iMac Pro Intel Xeon 8-core (2019 build). I’ll check out that link and ping you for follow-up if that’s OK. I’ve got an iMac i9 as well, but that’s more the “commodity” box. Probably good to use it as well.
Great. PM me when you’re set up. TBH the M3 ultra will be the one to test as it’s the latest and it will be interesting to see how it performs compared to the latest AMD/Intel x86 crowd.
M
We observed many interesting result. To briefly conclude what I and my friends have got these days:
I have not come across any compatibility issue on M4Max (4e+12p) yet.
Cubendo project scaling tests. M4Max is as slightly more than 2× powerful as M4 base(6e+4p) on my Uplifting Trance projects with tones of synths and plugins. Similar to Cinebench R23 scores.
For x86 chips: We have no economy to test Threaddripper or any newer Xeon. So only ‘consumer brand chips’.
For most of those popular synths eg. serum/serum2, pigments, Diva, M4 series is more efficient than any x86 we have tested (intel core 12th 13th ultra1 ultra2, AMD Ryzen 7000s, 9000s, x3d/non x3d). On average M-series chips handles the most polyphony in a single instance; only AMD 7950x(3d)/9950x(3d) sometimes handles more ‘total’ polyphony across multiple instances than M4Max 'coz of more cores and threads. Intel chips perform the worst.
Acustica: M4 p-core ≈ Zen5 > Zen4 >= M2 p-core > Meteor Lake P-core,
Fabfilter plugins Zen5&Zen4 chips a little better than M4 chips. Intel not yet tested.
Kontakt 7 gets the worst efficiency on M series Macs. Maybe because macs limit 4k random read, even when you use internal ssd, or simply because NI stuff are not optimised for Apple silicon.
Different plugins with QUITE DIFFERENT working conditions. Some plugins eats cache/RAM IO so AMD x3d(large L3 cache) and apple chips(large L2 and SLC cache) and lower RAM latencies and higher RAM bandwidth machines get benefits. Some might be sensitive to core to core latencies (for example SpecCraft when you use a lot of them. In fact 7950x3d handles less paralleled SpecCraft than 7800x3d (same on 9950x3d and 9800x3d)…what?). Some are only bottlenecked by float point calculation speed.
Different buffer size also QUITE DIFFERENT results. In general when using low buffer size x3d and M-chips are more efficient. Importance for realtime work: Larger cache, Low RAM Latency > Higher single core performance > Higher RAM bandwidth >> others eg. higher core count. However higher core count benefits way more when you’re dealing with Large projects and you don’t require super low RTL.
[7. & 8.] ==> If someone really wants the absolute ‘best value’ then he have to go and test by himself with his own workflow.
Intel big.LITTLE cpus suffers for most DAWs or plugins. Problematic thread scheduling.
win11 24h2, macOS 15.4.1. We have rule out DPC problems on Desktop Wins.
APFS has a default allocation size of 4096, and as I understand it all SSDs operate at 4k block sizes for random read io - but one shouldn’t interpret that as “Macs limit 4k random read” because that is not accurate. Other factors contribute to actual IOPs like hardware and file system format. Just thought I’d point that out
To follow up. I have had my Mac Studio M4 Max for a couple weeks now and have not had any notable stability issues. Installed perfect and seem to run fine here and I would chalk up that YouTuber’s experience to “operator error.” 
My main hangups with getting the new computer running have largely been with the transition to Apple silicon and also Nuendo no longer supporting VST 2 plugins. A few examples:
It has been a long, long set-up process (fresh install of everything!) and I look forward to really putting Nuendo to the test on this M4 Max soon!