CPU Overload / Dropout detected with CPU at circa 5%

[MattiasNYC]

@qba55

Cubase chooses it seems to use (up to 100% of ) 2 of the 24 that are available (to be fair, it does seem to use maybe 5-20% of the rest), and it seems it pukes when either of those 2 get maxed out. Is my read correct? I could be wrong but it looks pretty clear.
If this is the case, is it a Cubase issue? Do other DAWs do this better?
This seems on the face of it (caveat: maybe I’m wrong) to be mental, almost a mentally retarded use of CPU resources. If there is no way to fix it (settings/config etc), and another DAW does it better, I’m off.

I think it’s slightly more complicated than what is intuitive. I have the same CPU btw. and am running Nuendo. Here’s a screenshot of my CPU load:

I do post production and generally my load is relatively evenly spread within the mixer. By this I mean that the longest and most processing heavy load is likely going to be for example a dialog track with for example denoising on it, followed by dynamics and some sort of speaker emulation, after which it goes through the rest of the path which contains a bit more processing, and in parallel with that maybe a reverb off of a send going into the same target path (groups and outputs). But that’s not really that heavy relatively speaking, and most of my work is really about more tracks rather than heavier plugins or longer chains.

So basically what I’m saying is that if you have for example one really heavy VSTi with a bunch of processing after it and with processing on your master then that creates a long and heavy signal chain just off of that one source (VSTi). The question is if that processing can be distributed across more than one thread, and I think the answer is that a lot of the time it’s not possible if you’re looking for close to realtime performance. The reason is that before you calculate the reverb after that VSTi you need to calculate the sound output of the VSTi, so things have to happen in order.

(As a side note: The ‘boxes’ you see in Task Manager’s performance view are threads. Each core gets two threads with the Ryzen 5900X.)

Now, in addition to the above what complicates things is if you record enable tracks. I think that changes some things in some situations. And then on top of that you have different options for using ASIO Guard or not, different buffer settings, multicore processing on/off etc.

All of those things make a difference in how Cubase and Windows will work all of that out. So really what you’ll need to do is look at what your project is like in terms of signal chains, and then try out different settings to see if the load on CPU cores changes. For example, I’ve tested it on my setup (same CPU) and it clearly is a very big difference between a parallel load and a serial one. If I have 5 heavy plugins on one audio track and then have that track go to a group, and on that group I add the same 5 plugins, then the load on one core goes up significantly. But if I instead take the one track with 5 heavy plugins and duplicate the track so that the load is now in parallel that won’t happen.

Makes sense?