Pro Tools 2022.4 on Mac Studio M1 Ultra

[Darryl Ramm]

Originally Posted by Bob Olhsson

My understanding is that multi-core performance doesn't buy you much other than bouncing speed and video performance.

And capacity to run large sessions. The very first thing folks are likely to buy bigger machines for. Pro Tools is pretty well multi-threaded, but yes the multi-core performance will depend on number of the plugin instances and how plugins are chained.

[Carl Lie]

Originally Posted by Eric Lambert

This isn't adding up. How do we account for those who are experiencing great gains in ProTools with Mac Studios? Some are reporting a slew of CPU horsepower with very high track counts. Others seem stuck behind a wall of some sort.

It could be due to comparisons from slower intel macs. If someone has an intel mini and then gets the Ultra they see great gains. Having gone through the various posts it seems many who have 12 core machines aren't seeing that much gain.

Most Ultra responses though have said that they can push the Ultra harder at lower sample rates (ie forget about the PT usage monitors and just keep pushing the machine).

So perhaps if one follows the PT meter usage and doesn't see much change they should see if they can push the Ultra further than they pushed their other system. Maybe the gains come at the edge.

C

[massivekerry]

Originally Posted by Eric Lambert

This isn't adding up. How do we account for those who are experiencing great gains in ProTools with Mac Studios? Some are reporting a slew of CPU horsepower with very high track counts. Others seem stuck behind a wall of some sort.

So, what *really* constitutes the wall, and how is it not an issue for many?

I'd love to see an Activity Monitor report for each of these cases to compare all that's happening under the hood, along with a list of plugins being used. There are so many factors, any one of which could be the sole culprit for unexpectedly bad performance.

I don’t know if Activity Monitor would reveal it. More like how much is audio vs. VIs, what are the session audio settings (sample rate, bit depth, channel format), what plugins, how much elastic audio processing is going on?

Even on my ancient cheese grate, I get results that I can’t believe on some sessions, and on others, I add that one extra instance of FabFilter EQ and Pro Tools says “no.”

[glasswing]

Originally Posted by Selle Sellink

Last week I received my Mac Studio M1 Ultra.

After installing Pro Tools 2022.4, all the plug-ins I use and opening the session I am currently working in, I was a bit disappointed with the GUI Lag. The Mac Studio does not feel significantly faster or snappier than my previous setup. A Mac mini intel based from 2018 with a eGPU Vega 56.

What can I do to make this GUI lag a thing of the past.

I've tried the classics.

Buffer size to 1024 Samples
Reserve a large amount (100GB) of ram for pro tools to use.

I did some trouble shooting and discovered the following.

• I experience Lag in specific actions. Zooming in and playback works fine(ish) but control, option arrow keys up and down is pretty slow. Also using D, F & G to make fades is not snappy.

• When I experience lag in Pro Tools I can easily run Logic next to it working as smooth as butter on a warm day. I used the demo sessions to test this. Because of this it seems to me this is not an issue which has to do with to little compute power.

• When opening a new session and using a single track there is 0 issue.

• When making 1000 empty mono tracks without any plugins their is a clear lag.



I would love for Pro Tools to run as smooth as Logic Pro X does. How to achieve this I would love to know.

Current Mac:
Mac Studio (2022)
Chip M1 Ultra
Memory 128 GB
Built in GPU 48 cores

Previous Mac:
Mac Mini (2018)
Intel 3,2 Ghz 6-core Intel Core i7
Memory 32 GB
eGPU Vega 56 8GB

Displays:
LG Ultra HD — 3840x2160
Samsung S24H85x — 2560x1440
Optoma UHD — 3840x2160

Audio Interface:
Focusrite Clarette

Eucon:
AVID S1
AVID Dock

I'm using 2021.12 on a Mac Studio Ultra (MSU) and haven't noticed what you're experiencing. I haven't tried the 1000 tracks experiment because I've never come close to that number of tracks. 150 sometimes - and those seem to be playing fine on the MSU. I'm still holding off on 2022.4. 2021.12 works so well I don't wanna touch anything!!!! :). Some things are lightning fast - like rendering Elastic Audio - but most are like my old 5.1 machine except that now everything works and doesn't have digital tantrums. When PT goes M1 native, I suspect we'll need seatbelts.

[Selle Sellink]

Originally Posted by glasswing

I'm using 2021.12 on a Mac Studio Ultra (MSU) and haven't noticed what you're experiencing. I haven't tried the 1000 tracks experiment because I've never come close to that number of tracks. 150 sometimes - and those seem to be playing fine on the MSU. I'm still holding off on 2022.4. 2021.12 works so well I don't wanna touch anything!!!! :). Some things are lightning fast - like rendering Elastic Audio - but most are like my old 5.1 machine except that now everything works and doesn't have digital tantrums. When PT goes M1 native, I suspect we'll need seatbelts.

Hi Glasswing,

Would it be possible for you to do the 1000 tracks test? I’m pretty frustrated with this issue. Maybe going back to 2021.12 is the solution.


Verzonden vanaf mijn iPhone met Tapatalk

[685]

Originally Posted by Eric Lambert

This isn't adding up. How do we account for those who are experiencing great gains in ProTools with Mac Studios? Some are reporting a slew of CPU horsepower with very high track counts. Others seem stuck behind a wall of some sort.

So, what *really* constitutes the wall, and how is it not an issue for many?

I'd love to see an Activity Monitor report for each of these cases to compare all that's happening under the hood, along with a list of plugins being used. There are so many factors, any one of which could be the sole culprit for unexpectedly bad performance.

I feel that we don't have a standard way of testing. Testing is all over the place being that there so many different workflows and varieties of hardware setups. Would using the old Dverb benchmark test at the lowest buffer setting help give a better general idea of processing gains from computer to computer(or comparing software versions)?

Overall, my idea of standard testing with be something like Geekbench, Blackmagic Speed Disk, Cinebench, etc. Just something for DAWs were everyone is basically running the same test.

[Dax Productions]

Originally Posted by Selle Sellink

Hi Glasswing,

Would it be possible for you to do the 1000 tracks test? I’m pretty frustrated with this issue. Maybe going back to 2021.12 is the solution.


Verzonden vanaf mijn iPhone met Tapatalk

I'm on a Mac Studio M1 Max. I can't do a 1000 track test since I'm on PT Studio, but I can do a 512 track test.

Does get pretty laggy.

That being said, in my personal workflow, this isn't an issue. For my day to day use, this Mac is a huge step up from my fully loaded 2018 Touch Bar MacBook Pro. I run pretty heavy sessions with usually about 80-100 tracks, and lots of plugins. So far the only thing that seems to bog down the cpu are Acustica plugins. For the waves, plugin alliance, and fabfilter stuff that I use, everything is extremely stable, solid and fast.

I think the best we can hope for is that true m1 support comes sooner than later, and improves the situation for users with extreme needs like yours.

[Eric Lambert]

We can look at benchmarks to show what the computer is capable of, and we've seen plenty of these, but they doesn't tell us much about our particular ProTools rigs, and that's where the confusion enters. It doesn't matter if we're talking about the Studio, or the previous MacPros, or iMacs, etc... There are people who keep a myriad plugins (some of them grossly outdated) and use multiple videos and heavy VIs and lots of other CPU-intensive tasks, and there are those who do basic audio-only editing/mixing. While they might see identical benchmarks, their real-world experiences will be drastically different.

The only way to test your particular machine in your particular workflow is to test your particular machine in your particular workflow. We can all say that the Netflix session is the testable standard, but that's not exactly "real-world."

The best test I've found is a session of mine which is particularly taxing. It's one that requires care (large buffer) just to open and run. It's, essentially, the largest I'd create under normal circumstances, barring new plugins that consume drastic amounts of CPU.

I get the 1,000 track thing, and how that might satisfy a certain curiosity, but I don't find those pressure tests particularly helpful in my day-to-day. I commonly use video, I typically have a lot of VIs, lots of MIDI, lots of audio, and I favor some power-hungry reverbs and synths. The 1,000 tracks don't reveal much about how these sessions will operate.

The only way I'm going to know how capable a Studio is today, prior to Native support, is by getting one and trying it. But I'll be installing side-by-side my current machine, which allows me to decide which to stick with for the near future.

[Darryl Ramm]

Well said Eric.

We are in a time of instability. Adjust your thinking appropriately Changes happening to Pro Tools, macOS, native vs Rosetta, plugins changing, macOS induced iLok changes, interface driver changes as well. All this is going to have significant increase in instability and performance variability even over what fun Pro Tools was to deal with at times before.

Lots of stuff is going to be highly variable, confusing, etc.. I know folks want to look for a single number, aka single figure of merit, that tells about how something will perform for their needs or come up with some hopefully quick test. Single figure of merit is a psychological trap that frequently annoys the hell out of good performance specialists/engineers. And I've spent years around low level CPU benchmarks and TPC server benchmarking etc. All useful stuff but often does not mean what people assume they do. In a more stable environment as folks understand how things are working you might get more use out of other folks running tests/benchmarks. In an unstable highly variable environment SFoM are a lot less useful, or totally useless. And looking at the Pro Tools CPU meters also likely tell you little of value, as has been flogged to death on DUC now for a decade or so.

The benchmark/test that matters the most is what does any system do with your sessions/your workflow for you. If you buy a new Apple silicon Mac please plan on testing test it with your work as soon as possible, you may need to spend time troubleshooting plugins etc. before you decide to return it.

And when things don't perform as you hope, it's next to pointless to just get on DUC and complain, or to leap to an assumption that whatever problem you are seeing is due to one thing, say Rosetta 2. You *need* to troubleshoot as much as you can, especially plugins, and especially running out of memory (I am suspicious that some folks complaining here about performance may be running out of memory).

Complaining about performance without careful troubleshooting is of nearly no value, complaining about issues after careful troubleshooting is of enormous value. Whether that points at core Pro Tools or plugins or macOS or whatever. Like the fantastic work done by users here to identify Aux performance issues on Apple Silicon Macs... betcha when that is fixed it will help many users of large sessions on Apple silicon Macs.

If you are low risk person and just want a system that is going to work then maybe don't buy a Mac now... it's not the right time. There should be no surprises about potential early adopter risks here.

And a reminder that things may (I predict will) destabilize more as Avid comes out with first Apple silicon port and especially as the plugin vendors also ship native plugin ports. Lots of changes there some of which I am sure will cause stability and performance issues. (but hopefully being able to switch back to the intel code running under Rosetta 2 will be a useful work around in the case of some problems).

We are part way into what will be a multi-year transition of technology. We can all see how fast Avid responds to changes in the pasti, it's never impressive, so should not be surprised here how long things will take. And take to settle down/be high-end production quality for a large number of users in a large ecosystem of plugins etc. That's not ever a quick thing to happen.

OK back into my box (where I'll be waiting for the new Mac Pro unveil at WWDC).

[glasswing]

Yeah Darryl and Eric, that's what I figure. There are a trillion variables. And now there are thousands of us with Mac Studios and Ultras - two very different machines. There will be a ton of noise, depending on the skill and needs of the user.
I don't really give a **** about benchmarks etc other than "Does it do what I want?" and "Is it better than what I had?" For me, the Mac Studio Ultra is a dream come true - but I was suffering mightily with my old 5.1's.
I'm not running theoretically insolvable equations or the backbone of the Internet or thousand track sessions. I can't imagine that the upcoming Mac Pro will be that much more useful than what I have. If I was Hans Zimmer or Skywalker Sound maybe - but I'm not. To me it's crucial to figure out what we need, find it and stop looking. Otherwise it's an infinite loop of expensive pursuit. The makers of the equipment we use hope we are on that infinite loop! But I've been there, done that, and spent a fortune doing it. For me it's just music. And I'm no longer taking clients outside a small handful of peeps. For what I do, it's perfect.
If and when it stops working, I'll look again - but I'm 74 - so the math is strongly in favour of the Mac Studio Ultra! :)