Hi everybody!

Getting my feet wet with Pure Data on the quest to find an environment that works well for interactive modular synth stuff with MPE compatible controllers.
On first sight I am quite impressed - particularly the "clone" object is really a relevation when it comes to managing several voices (compared to most other eurorack-style mono-voice-module modulars)!

Two questions:
Do we always have to create an abstraction (= a new file) when we want to clone something?
Have some cases where the cloned objects are so highly special that they only make sense within one particular parent patch and even there they occur only once. (E.g. have an "mpein" object for reading mpe midi data that uses a 16 times cloned, quite specialized mpein_channeltovoice" abstraction internally)
Can I somehow use a subpatch instead of an abstraction that is stored together with the main patch file?
(Not a big deal, might just be more convenient when sharing stuff to have one file instead of a bunch of them)

And assuming I have a synth with quite demanding calculations per voice - would it be possible to somehow combine pd~ and clone? E.g. to run 16 instances of an abstraction in parallel on 4 cores with four voices each?
(Also not a big thing if it doesn't work, I can always manually create 4 pd~ instances with "clone myvoiceabstraction 4" inside each. Would be convenient though not having to copy everything N times. Is still ok for 4 cores, but not so much fun for a 32 cores 64 threads CPU .)