Ive some questions about convolution node and its functionality.
When this is enabled cpu% goes down a lot. What are the dis/advantages of enabling multi thread? I assume enabling multithread splits the processing of the IR among several cores. Do I need to consider the end user and their specs when enabling this?
What exactly does oversampling do? Im assuming this increases the resolution of the IR? I cant hear too much of a difference when putting a convo node in an oversampling container.
@JL-LV Multithread moves the heavyweight tail processing into a (single) background thread. The CPU usage doesn't go down (still has to crunch those numbers) but it is not rendered directly anymore.
There's nothing particularly interesting to consider here and it works rather stable (at least from my sparse use of the convolution reverb in various products).
Oversampling is not a guarantee to make sounds better so if you don't hear a difference, then don't use it. There are only a few selected use cases where it makes sense to oversample a signal (the most important one is if you introduce harmonics with a distortion which folds back into the audible frequency spectrum), but I would rather not oversample convolution since it's an extremely heavy process already.
The CPU usage doesn't go down
The floating tile shows a drop in CPU usage though.
@d-healey Yeah but because it only measures the CPU usage of the audio thread (this is how CPU meters usually work in DAWs).
It's basically this: