I know the point of UniqueMemoryWriterRecordingNames (UMWRN) and it's sibling CancelUnique... are to make the wavetable memory used by a Sound unique and not clash with any other Sound's memory, and so would take up more wavetable memory by design.
But do these *Unique* Sounds have an additional overhead?
I have two copies of a Sound in my Signal Flow that use FeedbackLoops. To stop the memory clashing I thought I'd wrap them both in UMWRN / CUMWRN. If I do that I completely run out of DSP.
If I only wrap the one on the right in a UMWRN / CUMWRN pair, then I'm hardly using DSP. Only wrapping one of them has the same effect as wrapping both in terms of keeping the memory separate. The reason why I want to wrap them both is because that's what would happen if I encapsulated the Sound.
If I wrap the left one, before the Prefixer, then I kill the DSP, even if the right one is not wrapped.
Right side copy of the Sound ("pitchDelay") is wrapped. Lovely jubblies, loads of DSP:
Wrapping them both (or just the left side copy):
... and boom! DSP gone:
Does the UMWRN not like the Prefixer or something?
Here's the Sound.
Now that I've encapsulated this Sound as a class, which includes the UMWRN / CUMWRN wrapping, I'm not getting the DSP problem. I stacked 5 of them up, no problem and also tried using a Prefixer without an issue.
