I’ve always been told that Cubase’s internal data path was 32-bit floating.
Oversampling helps avoid distortion, aliasing and other nasty side effects (apparently). Many new plugs have selectable settings for different levels of oversampling (4x, 8x, etc.).
But I’m running Cubase at 48k, 32-bit floating - so is setting a higher oversampling rate in a VST worth it? A higher oversampling rate hits the CPU harder.
You can select 64 bit word-length internal processing now, as long as you’re on a reasonably recent version.
The word length of Cubase’s internal data representations, though, isn’t related to whether or not to use higher sampling rates.
I think the answer to that question is, it depends!
There are some effects that to my ears (and many other people’s) sound better ruining at 96kHz, particularly reverbs. I did some blind tests with some of my tracks mastered by a third party engineer using 96kHz mixes vs 44.1.kHz mixes and I could reliably tell the difference. That was with him using a mixture of anslogue and digital processors.
But sometimes the effect is not large, and it’s better to run at lower rates to give yourself more power.
I think this video by Dan Worall for FabFilter might be helpful for you:
That’s great! Really clears stuff up. Thanks, Johnny!