[LAU] Multiple asynchronous JACK chains resampled?

[Fons Adriaensen]

On Tue, Feb 23, 2016 at 09:32:46PM +0000, Will Godfrey wrote:

> However one must make the distinction between an inactive voice and a
> non-sounding one. In the latter case processes will still be running to check if
> there are any notes to deal with, and any in-line effects will be fully
> operational - reverb being one of the heavyweights.

As an example zita-rev1 takes 2% of a single CPU on my machine.
And since reverb is a linear process you don't need a separate
one for each note. You would have at most one per instrument
in a multitimbral patch. An even a reverb could be bypassed
some time after the last note has ended, as can all effects. 
Checking if there are any notes again doesn't take any
significant time. 

Ciao,

-- 
FA

A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)