Studio One 5.4 M1 Native vs Rosetta - Not What I Expected!

[@MattHepworth]

Thanks, all! I didn't realize Fab Filter was M1 native as well. What other M1 native plugins would everyone like to see? Let me know!

[@Iredidv]

Valhalla and reaper are m1 compatible. Newest ableton also. Reason 12 in a month or 2.

[@MattHepworth]

@@Iredidv Great to know about Valhalla!

[@Gijsknol]

S1 clearly states, if M1 users want to benefit from the optimal performance the 3rd party plugins require to be native M1 as well. As mentioned below 16 samples buffer size will very rarely be a safe and smooth option for tracking. For low latency midi-monitoring its usable though. Very few of my plugins are native m1 yet, could take another 1-2 years and for some plugins it might never happen since some have to be written all over again from the ground up.

[@MattHepworth]

This was a discovery video, so to speak, recorded as I was testing for the first time. I address what you're talking about when I see that it's using Rosetta wrapping plugins in the Activity Monitor. I will do a follow up with using only M1 native plugins, but most plugin companies don't offer that yet, unfortunately. 16 samples is the normal setting for me with Studio One, as it's for tracking and the latency needs to be less than 2ms. Nothing negative was stated in the video, but I feel you may have taken it that way. It was just a first real-world scenario test and I found it quite interesting.

[@davelongenecker649]

I wonder if any of the plugins had a 'nap' during these two tests. Thanks for testing, and all the helpful info!

[@MattHepworth]

I meant to show that. Nap was NOT checked in order to make it the most fair comparison.

[@davelongenecker649]

@@MattHepworth Good call!

[@TheVidman5]

Fabfilter is M1 compatible already. Maybe a test with only FF plugins would be great to compare Rosetta vs native m1 performance.

[@MattHepworth]

Thanks. Looking into other M1 capable plugins as well as FF!

[@jmcbade2960]

How about a test with not Rosetta dependent plugins just for comparison? Obviously we will need our plugins if they are Rosetta dependant or not but to see if that is why would be interesting.

[@MattHepworth]

Hey there! You'll want to check out this one then - only M1 native capable plugins (and no Rosetta) vs the same system under Rosetta - kzbin.info/www/bejne/oGHdd5pni5eEipY

[@adamaayan]

I wonder what the results will be only with VST3 plugins and plugin that are supported with M1 will run. I have a feeling this results will change drastically.

[@MattHepworth]

The fact that it uses a Rosetta wrapper is definitely affecting things still. I used a real tracking session for this video, but I'll come up with something more creative, too. Thanks!

[@steve_barry]

Thanks for this test, not sure if you mentioned it but what is your unified memory for the M1?

[@MattHepworth]

Hi Steve, it's 16GB, which is the max on the mini, I believe.

[@JRandallS]

One strange thing is the M1 in Rosetta mode beats the Mac Mini intel version, even though the intel version had more ram. There are a number of test videos that demonstrate this attribute. What this means is that M1 under Rosetta is more efficient than Intel. So the processing power (or speed) lost in the translation process is more than made up by efficiencies gained in the M1 architecture.

[@MattHepworth]

Great comment. RAM comes into play a lot more if you're using virtual instruments, but I too was surprised how the Intel performed against Rosetta.

[@Newmusicreview]

Cool vid dude..

[@MattHepworth]

Thank you!

[@misanthropolist1808]

I have studio one 3 professional. Will I be able to run that on a M1 Mac? (I’m not that big of a genius with computers)😄

[@MattHepworth]

I don't believe it will, unfortunately. I think you need a newer version (5) to support Big Sur and later OS versions.

[@OneFingerSnap]

The dropout protection isn’t set the same on both modes. On Intel it was at ‘Minimum’, on M1 it was set at ‘Medium’. Maybe this changes something.

[@MattHepworth]

Great catch! It actually very slightly reduces strain on the M1. It's mostly for playback tracks, though.

[@bernigracia3117]

Remember that m1 chips ram and processor work different so it may be saying it's 100% but in reality the processing has been swaped to the hard drive and it's using its resources so it's working beyond the 16gb of ram. At least that what I have seen in my experience. Do the same thing but check the activity monitor and you will se the real consumption of memory used

[@MattHepworth]

indeed! My average RAM usage in these scenarios is about 7GB of the 16GB I have.

[@patrickbear6727]

You showed your Device Block Size (buffer size) but you didn't show your Dropout Protection / Process Block Size. For this test it would be important to make sure the Dropout Protection setting is the same because, unless you are live tracking something, the Device Block Size is not what will determine the actual load on your processor (unless Dropout Protection is to Minimum or Low) because playback tracks will be running at the Process Block Size not the Device Block Size. This is one of the amazing features about Studio One--that you can have essentially different buffer settings for your live tracking monitoring and for your playback tracks. +

[@patrickbear6727]

sorry, I missed an earlier brief discussion on this.

[@MattHepworth]

Yeah, Studio One is brilliantly designed and very effective at helping us balance CPU vs latency.

[@ВасяШевченко-з1э]

Not a bad thing, but it would be interesting to check with native plugins and not through a rosette. Then I think you could see the difference. but it looks like the load has become double.

[@MattHepworth]

I was not able to sufficiently load the processing with the Presonus plugins I had in a workflow I'd use. I tried it at first.

[@thomas-mavian]

Interesting... Would be nice to see the CPU read out from Activity Monitor. And, I *never* use AU in S1, feels a lot more "stable" with only VST3 and VSTs (that could be just my feeling though, have no proof that that's correct).

[@MattHepworth]

Interesting! I don't normally install VST for plugins on this Mac, since it doesn't run Cubase. Just PT, Logic, and S1.

[@embulus75]

Think is a CPU meter delirium caused by M1 native and Rosetta "adapter" working togheter. what says the activity monitor CPU usage? (task manager 4 Windows)... I'm on the Windows side than not sure. You can try to use only bundle plugin like reverb and some maitai instances in Hig quality mode and see the difference. anyway... bye!!!!

[@MattHepworth]

Those were my first thoughts too, but then I did this comparison using only M1 plugins - kzbin.info/www/bejne/bqWkhmiYfbeLgas

[@carlosp.1846]

What is this Hybrid mode ? Where do we set it ?

[@MattHepworth]

When the Green Z is active only record-enabled tracks use the low buffer. The other tracks (playback tracks) use a high, regular buffer.

[@jonnytheshirt]

Same with Logic - if Rosetta run plugins are involved you'll not get the full M1 benefit, and no point in running native in that case unfortunately - a revealing test would be all M1 native plugins v all Rosetta plugins.

[@MattHepworth]

Indeed. Any suggestions for plugins for that comparison? Fab Filter, of course, but looking for a real-world set of plugins to use for it. Thanks!

[@ejmikk]

All CPU meters work differently, they are never comparable between OS's, DAW's or even, as you can see, different processors. There are just too many ways to measure cpu usage and no standards, for example the OS and the DAW always give totally different readings. The only way to ever test how well a given CPU performs is to either add tracks to test multicore processing or add fx to a single track to test single core performance until the glitches and dropouts begin.

[@MattHepworth]

In this case I think the glitches are the best indicator. You're right.

[@MarkusH1966]

16 Samples? That’s not going to work anyway. Test both versions at higher buffersizes and you’ll see a difference. On my M1 it’s obvious.

[@MattHepworth]

I always use 16 samples exclusively, since I only use S1 for tracking. This is true on all the Macs I have, not just the Mini M1.

[@MarkusH1966]

16 Samples? Not gonna work.