Hi everyone,
I recently did an Orac port of the very nice Koto Organelle patch.
As you can see in the comments on the patchstorage upload, some encountered performance issues as the cpu of their device couldn’t handle this heavy module in parallel with other modules.
That’s why I thought I could move the synthesis part of the Orac-Koto to a Pd subprocess in order to make use of the multiple cores present on some devices and therefore make it usable for more people.
This second version seemed to work nicely when I tested it, I also have to say that I run Orac on a rpi 4 and so never really could test the case of performance issues.
Nevertheless @alanca still encountered the same issues (crakling noise, more when playing a lot of notes or with other hungry modules) on his Organelle-M. When looking into the output of htop he got this:
You can see how the subprocess running Koto is split away from the Orac process onto another cpu and leaves them at around 50%
Does anybody has any idea why the issues could be present in this case? Could it be something with Pd and not the hardware? They still come up the most when playing polyphonically so it must be the cpu right? 
ps: In this post they show that the original organelle can still play smoothly at much higher cpu % use I think