When do you think Logic Pro 11 will come out?

[Ploki]

Various VST plugins have already been multicore (like Diva)...

Disabling multicore on Diva leads to better performance. So much for that....

on M1 at least

[syntonica]

Various VST plugins have already been multicore (like Diva)...

Disabling multicore on Diva leads to better performance. So much for that....

Like I said, YMMV. It depends on the OS, the host and the plugin.

Like pdxindy said, CLAP now coordinates the effort between host and plugin. I think plugin says to host, "Thread, please" and the host hands over a thread from its pool to use for a job. I haven't looked at the code yet.

[machinesworking]

I mean it's not like all DAW developers sit around and talk about how they introduce their own multithreading for individual tracks in your DAW, so Diva and other multi core capable plugs can only offer so many choices in this. None of that is surprising, it's more surprising to me that some DAWs have started to multithread a single tracks plug ins into separate cores etc. Which of course means possibly wildly different ways they implement this code wise.

To the point to where when I was messing with this to see, turning off multi core support in Diva was beneficial if you had multiple instances of Diva, for one or two, it was better to have it on.

[lightsfadelow]

...it's more surprising to me that some DAWs have started to multithread a single tracks plug ins into separate cores etc. Which of course means possibly wildly different ways they implement this code wise

Which DAWs are now able to do this?

[medienhexer]

If you‘re seeing poor performance on m1 Pro/Max, that might have to do with a threading issue where, even though plug-ins in Logic have a high scheduling prio, they are processed on the efficiency cores and never moved. Even the real time thread for live-performing instruments might end up on an efficiency core.

I found this while troubleshooting a plugin built with JUCE with the developer. The JUCE framework developers were aware of it and had a solution in mind, but that was probably a workaround for the underlying issue and they are hoping for Apple to solve it propperly.

From what I gather, if you don‘t use third party plugins, there isn‘t an issue. I am able to run hundreds of instrument and drummer tracks and thousands of internal plugins on an m1 Pro. It‘s so much more than anyone could ever use for anything. Like hundreds of Chrome and convolution reverbs, both on tracks and on busses, and thousands of delays, compressors, EQs, etc. basically without any meaningful limit. I think I maxed out at 800 or so tracks with 12 inserts. Some stupid number like that. With 60 or so Drummer tracks playing.

[medienhexer]

Don‘t get me wrong, enabling the GPU could still be very interesting for really heavy number crunching.

And Drummer could use the Machine Learning capabilities to play better drum parts from listening to more than one track, which is the current limit. Provide a track with the basic groove per part. Have it listen to the bass and a percussion track in parallel and then give me automation lanes so I can draw in the changes in intensity and complexity and mark the points for fills and breaks. Just like talking to a drummer.

Base the playing on a huge analysis of the best drummers in every genre and you have just solved a composer problem.

[syntonica]

...I maxed out at 800 or so tracks with 12 inserts. Some stupid number like that. With 60 or so Drummer tracks playing.

I'm just happy I can get more than 10 tracks playing on my MBA M1 in Rosetta. That's with 3rd-party plugins, not builtins. But with 800 tracks, now you can simulate a whole orchestra with phys mod.

[lightsfadelow]

I'm just happy I can get more than 10 tracks playing on my MBA M1 in Rosetta. That's with 3rd-party plugins, not builtins. But with 800 tracks, now you can simulate a whole orchestra with phys mod.

Yikes! Ditch those non-native plugins and go all-native! Are there any non-native plugins still holding you back?

[medienhexer]

10 tracks? That has to be a RAM issue. Do you have less than 16GB RAM?

[machinesworking]

...it's more surprising to me that some DAWs have started to multithread a single tracks plug ins into separate cores etc. Which of course means possibly wildly different ways they implement this code wise

Which DAWs are now able to do this?

One of the Bitwig developers stated as much, and from what I recall years ago testing it with Diva loading it did much much better with multicore off in the Diva instances.

Now the other trade off with Bitwig is like Live it does not have a separate buffer for unarmed tracks etc. so there's no CPU saving tricks it uses for high track counts etc. Logic, DP and Reaper are far more CPU efficient in general than Bitwig because of that. I do not know if these two CPU tricks are mutually exclusive?

[syntonica]

I'm just happy I can get more than 10 tracks playing on my MBA M1 in Rosetta. That's with 3rd-party plugins, not builtins. But with 800 tracks, now you can simulate a whole orchestra with phys mod.

Yikes! Ditch those non-native plugins and go all-native! Are there any non-native plugins still holding you back?

It's not the plugins, it's the DAW (MuLab). I've been patient, but I may have to go with a new DAW. (Reaper, pro-tem. I tried a new menu and skin and aside from the hideous aux windows, it was usable to me.)

[syntonica]

10 tracks? That has to be a RAM issue. Do you have less than 16GB RAM?

That was my old 2014 MacBook Pro.

I got the 8GB model of the Air and don't have any memory problems, per se, although the AAS plugins are total RAM pigs! I don't do anything sample-heavy, so it will be fine.

[Johnnyjohn]

with the lates m2 pro announcement from today they've put more emphasis on their neural engine, really hope this will be made use of with logic in the future

[syntonica]

with the lates m2 pro announcement from today they've put more emphasis on their neural engine, really hope this will be made use of with logic in the future

Jeepers! With all these different processor types all sitting on the same chip, wouldn't be easier to just do 64 or 128 identical RISC cores?

I'm not sure how a neural network can help, unless you want it to learn your style of music and compose it for you. I would find that creatively stifling. I prefer using aleatoric methods when stuck for an idea.

[pdxindy]

I'm not sure how a neural network can help, unless you want it to learn your style of music and compose it for you. I would find that creatively stifling. I prefer using aleatoric methods when stuck for an idea.

Would the neural network be useful for resynthesis of audio (Alchemy), timestretching improvements, Flex Pitch improvements?