Hello everyone, it's great to see you again!

I recently installed Cardinal synth, but I was wondering how to force it to use more cores inside Cubase. I think I read that it uses the threads the DAW assigns to it (something like that)...

How can I assign it more cores?

I was thinking about using a VST host inside Cubase, maybe...?

You're the ones who know, though!

Thank you very much in advance.

 

AFAIK is that not possible. Because one audio stream can't be divided into several streams and processed on multiple cores and then be mixed together again. It doesn't work that way. It' a principle, so there is no trick around it on no computer up to this moment available to mankind. Maybe a quantum-computer can do that ... but you do not want to pay for that. Wait 50 years ... maybe 25 years ...

 

Unify from Pluginguru loads each synth and affects it to one thread or core of a multicore CPU.
But i don't know of any host capable of dispatching cores on only one instance of a VSTi

 

Im not sure if its physically doable atm...or again actually more of a hinderance.
Usually the workflow and processing depends on the program of that particular plugin first, and less to do with DAWs itself.

For example, if you have multiple chains routing to a bus, again and again, the computer would then, use more cores to offload the incoming stream in order to process with minimal latency. But the more cores, you use, the more likely you'll "slow" your CPU, until it comes to scathing halt and crash.

It's better to try to enhance your single-core performance over time, rather than try to use multi-cores as often.

I found this link that may be helpful to you.
This one refers to Cubase more also.

 

that this is not possible is why single core performance is always paramount for cpu selection for studio machines

 

I might be wrong, but when I use U‐he Diva I believe it let's me choose a Multi Core button at the top right of the GUI and it lessens the CPU load on my computer considerably.

 

It can be done if voices can be spread on several cores.
Like 4 polyphony can be spread on 4 cores.

But, as a counter example, Zebra can't do it due to the link between voices who need to be perfectly synced.
So all voices are on the same core.

So yes it can be done, at plugin level. BUT only in some specific situations.
That's why, as a general rule, the answer is NO, dont count on it.

As another example, games are divided in several threads. But some operations can't be divided in threads ...

 

Unless something is designed to utilize several threads (or processes in some rarer cases), then there's no forcing it. Your DAW should be able to run parallel tasks on multiple cores, but it all depends on the audio graph of the DAW and the order and routing of the plugins. Some DAWs also allow you to configure plugin hosting (in DAW process, in a plugin container process, in separate processes), at least Bitwig does.

Of course, some instruments can use multiple cores (u-he as above does this, Kontakt/Reaktor can as well), but it all depends. If you want to balance the CPU usage as well as you can, don't go overboard with FX, sends in particular.

So I don't think there's one single true approach, get some core utilization stats and experiment. There are some tricks to manage process load, but it's not really a good idea with audio software.

 

Audio computing is a very tedious task.
It needs to be low latency (buffers need to be short) to record in (almost) real time AND can't be interrupted without getting glitches.
Short buffers mean more CPU load spikes.

It is not so evident to determine if more cores is better than more frequency.

More cores mean more balance between cores ... if you use less intensive plugins (if you use a SINGLE very intensive plugin, it can ruin the whole balance).
More frequency will help with some heavy plugins ... but will overload with a lot of "less intensive" plugins (like using a compressor/EQ/Whatever on every track).

The best is to have more cores AND more frequency

As a side note, more and more plugins makers provide a setting for real time AND another for project rendering.
Ex : you can use 2x oversampling for real time AND 8x for rendering

And if you don"t record real time audio or very fast MIDI, you can RAISE your audio buffer (ex : 128 instead of 32).
It will help a LOT to avoid CPU spikes.

 

verify if audio settings are the same in both DAWs
Because a lot of plugins are badly programmed (like no 48khz support), and they overload.

 

Last edited: Aug 5, 2024