• Hi and welcome to the Studio One User Forum!

    Please note that this is an independent, user-driven forum and is not endorsed by, affiliated with, or maintained by PreSonus. Learn more in the Welcome thread!

M4 Mac Mini Pro performance tests

js1

Member
In other threads, I had mentioned that I have a new M4 Mac Pro, and that getting it up on S1 and all my peripherals and plugins, while not trouble free, went better than expected. I know that there's a lot of conjecture about how it will actually perform, and I really want to see for myself, as well as see if there's any best practices/workflows that will let me squeeze more out of it.

To get a sense of what's going on, I picked up iStat, which is an excellent system monitoring tool (bonus - got it half price for Black Friday). Here's the CPU panel:

istat.png
As a first test, I wanted to quickly push it to its limits, so I grabbed 8 audio tracks, and started loading up T-Racks tape plugins on them. These are notoriously poorly behaved CPU hogs. For reference, on the Mini, a UADx 1176 compressor shows as "1" in the Performance monitor, the T-Racks tape range between 7 and 12!

On my previous machine (Intel 9900k), I could put on 2 of these before everything fell apart. On the Mini I was able to put 71 across the 7 tracks - this is what you're seeing in the screen grab. A few observations:
  • This was at 48Khz, with a buffer size of 4096, process precision single (32 bit) and dropout protection off
  • Got the same result with buffer size of 64 and dropout protection of 2048/1024. With 512, I had to disable three plugins.
  • Changing to double precision did not seem to affect the results
  • Changing the sample rate to 96K dropped the total plugins to 52.
  • It fails in an "all or nothing" way. Audio is perfect, and adding one more plugin causes major dropouts.
  • During playback, you can see the load on the individual cores constantly change. I had read that the M4 had logic to handle core balancing that is much more aggressive than past processors, and that seems to be the case.
  • With Studio One, at full load the loudest the fans got was 47 db. (1 ft away, measured by my Apple Watch). The fan was running at 55%/3100 rpm.
  • Take the load down slightly, the fan drops to 2000 rpm and is inaudible. Real world I don't expect this to be a problem, but I imagine if I was running an application that hit the GPU equally hard, it would be noticeable.
  • This is a new machine, so I don't have lots of other processes running, and certainly nothing that seems to want to use the performance cores.
  • Because I was away from my studio, for this test I was using my Komplete Audio 6. I'll repeat it later on my Quantum.
I've got some other tests in mind, but let me know if there's something you'd like me to look at and I'll see what I can do.

Please keep all Studio One complaints off this thread, including the whole efficiency core debate. I want to see this board grow and thrive, and I'd like to provide useful information that could help someone make their own decision if they're contemplating a purchase.
 
Round 2....

256 audio tracks (no shared tracks). Keep adding Pro EQs until it fails.

To make sure that the Pro EQs are actually doing something, I enabled all the bands, including the zero-phase high pass. This meant that each EQ had a latency of 31.3 ms.

I ultimately added 10 instances to each track (2560 in total), and playback was still smooth. I was getting close to the limit, but I didn't keep going. Other operations in Studio One were getting very sluggish.

That plugin latency did have a major impact on the test (memory use), and I suspect if I had chosen a low latency plugin, the number would have been higher.

After a few other experiments, I'm reasonably sure that all plugins on a track are bound to the same core. This is what I had always believed, but it does seem to bear out in practice. Single core performance still is critical, because if one core overloads, it doesn't matter how many spare cycles the other cores have.

I have to say I'm extremely impressed at how smooth and predictable the audio is with this machine with heavy loads. On my previous machine, I when the Performance meter was sitting at 60-70%, it would occasionally spike to 100%, causing dropouts. I've seen no spiking here.
 
Last edited:
Thank you! Cannot contribute to your measurements, but buying iStat for 50% BF sale was a no-brainer. ;)
 
  • Like
Reactions: js1
Back
Top