This ia a good illustration of why I keep ranting on about the need for proper metering in VM.Steve W wrote: ↑Tue Aug 30, 2022 4:57 pm It is odd, though, that I have had my PC running with considerably higher overall CPU usage with all sorts of software running and didn't have the issues I had with this patch. I just tried all different buffer sizes, and all of them have the horrible noise. Oddly enough 480 and 960 samples have a lower rate of glitching.
While overall CPU load is a useful indicator, it's an extremely crude measure of how close we are to buffer underrun. The two things are correlated but only vaguely.
What we need is a meter that shows the time required to fill the buffer divided by the time required to empty the buffer.
When this value exceeds one then the system fails and we get nasty distortion, noise and ultimately periods of silence. That's why we need a meter to see how close to one this value is.
In pseudocode...
t = now()
doProcessing()
display( ( now() - t ) / latency )