On Tue, 2018-01-09 at 13:55 -0600, Chris Caudle wrote:
If the soft-synth MIDI inputs are then both connected to jack MIDI ports
that are coming from the same a2jmidid connection, does that now make them
dependent, and they can't be processed independently?
So back up, assume the soft-synth instances are generating audio
completely independently, e.g. they have their own internal sequencers, if
the outputs are both connected to the same Ardour instance does that then
make them dependent and force onto a single processor?
If the answer to both of the above is yes then I do not know how you could
have independent connection graphs in a realistic system. If the answer
to either or both is no, that does not make the two soft-synth instances
dependent in the connection graph, then maybe you could get the load to
spread out across cores.
I've got a lot to try. But I'm fairly certain I can do very well. Additional latency from localhost networking is going to be tiny beyond tiny, and unless I need yet more hardware, 4GHz should crunch Zita's numbers very nicely. I am currently thinking to run three soft servers thus:
1. Yoshimi 4x connected to non-mixer, and then triple Calf FX chain (Compressor, Reverb, EQ)
2. Calf-wrapped FluidSynth x7 or so connected to non-mixer, and then triple Calf FX chain (Compressor, Reverb, EQ)
3. Linuxsampler connected probably to single or double Calf FX (EQ at least)
and then have all three streams to a hard server connected to the real hardware and doing nothing else. MIDI will be ALSA MIDI initially, I may eventually get brave and try one of the MIDI-over-IP methods.