because of vvvv’s data structure (the inputs of a node request data at the
outputs of connected nodes/IOboxes, not outputs “pushing” data to the inputs)
it may be useful to switch off inputting IOboxes, so that following nodes don’t
request data anymore.
kalle, i can not really follow you here. gregsn might correct me, but concernig cpu consumption, ioboxes should not make any difference at all. except you might like to switch off UpdateView in the main menu or hide the patch to disable redrawing (which in fact consumes a lot of cpu)
so you mean i need to use lots of switches to cut off the datastream.
i’m gonna try this. i was just wondering if the empty spread has a similar
effect as most of my patch didn’t work when the empty spread was pushed
through.
in general switching off inputs will not improve performance.
the way to go is lowering the spread count. set all spread counts of LInearSpreads etc. to zero. Each slice consumes cpu and the less slices there are, the faster the patch is. The debug view will show you. Apart from that, each node consumes a fixed amount of cpu. So beware of subpatches containing many instances of other subpatches, and try to build subpatches with as less nodes as possible.
Empty spreads in vvvv might be a little nasty to deal with, but they are a fully plausible consequence of vvvv´s architecture, and by no means a bug in the software which needs to be worked around - learn to use them. the basic rule is: as soon as any input of a node has zero slices, the output will have zero slices as well. this may easily generate empty spreads at many places in your patch when having feedback loops, and it may be somehow difficult to learn where the first empty spread appeared, but as soon as you have found the position, enter a ReplaceEmpty or Resample node, and you´re fine.