What '8 MIDI channel bug'? Do you mean that one multi of Play is going to afford you 8 MIDI channels only? I will say that using a whole instance for one articulation strikes me as kind of 20th century usage. Is there some library for Play that lacks that 'Master Keyswitch' (such as I remember from Silver SO)?

I don't use Play extensively but it assigns to whichever of 16 channels here, I'm pretty sure.

163 audio outs to Nuendo, to me is a little insane, though. Unless you absolutely need 163 stems, and that does not strike me as so normal either.
Here I would highly recommend figuring out some type of grouping and learning to mix using VE Pro rather than all of this ported to Nuendo.

Additionally, Nuendo can use VST3 VE Pro, right? (up to 48 MIDI ports, x 16 = 768 channels) I can relate to 9 instances = 9 outs from Play, if they are different instruments. But if you're using 9 instances for 9 articulations of 'vlns 1', 9 separate outs is kind of nutty right there. I don't know what that hardware will bear, but I would expect 163 outs to be a potential bottleneck. 3 machines would probably mitigate that somewhat, of course (albeit Mac Mini not my preference), but again I say you want to consider rethinking this whole modus operandi.

Someone elsewhere mentioned some bug with more then 8 midi channels.. I wanted to know more.  

As for all the outs. I want the flexability to both output as groups which i have made as well as individual instrument articulations if needed for say a revision to soemthing or if they want the separate instrument stems. i am NOT using keyswitched patches. 

I am having multiple slave machines running this. I have midi channels for each articulation.  This lets me easily pick what i want at a whim as I am composing.  I don't need to inturupt the flow at all.  This lets everything be at my fingertips.  

I decided to use and abuse both the FX and Group Channels 

I Made the Group channles be for Strings Brass Percussion and the like.. while the FX channels are each intrument type for example:  Violas, Oboes, French horns.  This also lets me position them where i want them to be.  No i dont' have MIR. 

While it seems complicated to setup .. Once its done it will be very nice to use. The guys over at Thinkspace.co.uk gave me the idea which Seems brilliant. 

I position most everything very particularly myself. You do not need outputs per se to do that unless that positioning has to happen in Nuendo, which I don't see how it does. But I see you are making groups. 163 outs is a lot. Loading an instrument for one articulation adds up. If you can obtain good performance with that much loaded more power to you. I think it's asking a lot of 3 Mac Minis but I could be mistaken.

For me, 81 channels in one VE Pro instance is not a terrific lot. I use 3 or 4 outs with a similar size or larger viframe, though. Not that it sounds complicated, it sounds like a lot for that machine and for LAN.

I am using a mac pro 5.1 at the moment hosting nuendo 6.5.30.  I wanted to do a master template with all of EWQLSO. I decided that the keyswitched intruments were a pain to use in my use case.  So decided to go with the individual articulations. I am using each play plugin instance to run 9 articulations not one play plugin per articulalation.. just to clarify.  vframe 1 has 9 play plugins each loads 9 arcituclations so 81 per instance.  So far on my i7 quad core with 16 gig ram its doing pretty well untill i got over 3 or 4 vframes. Then started getting play resource errors and such. THOGUH memory isn't used up and cpu is not even 1/2 way. SO not sure whats going on there. 

The plan was to load up 3 mac mini 2011 quadcore i7 *8threads* servers with 16 gig ram and one system drive with 128 SSD in the 2nd bay for the streaming samples. Then divide up the vframes to each to load balance thigns.  Yes it requires 3 licenses of gold but then im really flexable as to my load balaning. 

Is that a bit more clear?

@Another User said:

The plan was to load up 3 mac mini 2011 quadcore i7 *8threads* servers with 16 gig ram and one system drive with 128 SSD in the 2nd bay for the streaming samples. Then divide up the vframes to each to load balance

Yeah, I'm a proponent of distributing the load and more machines/more threads rather than expect one machine to distribute everything equally automatically.

My approach is to give the heavy users amongst viframes more threads per instance. If all or most of your Play multis are equal quality users, I'm not real sure how much you're going to benefit by a more expanded setup. I really do suggest to anyone to do as little outputs to the host as feasible.

You say you're not doing MIR Pro, but I would relay one thing I find in this documentation and in tutorials, is that they are presenting scenarios of a single master output. More complexity in mixing really does not need be more outputs to the DAW host, ie., using VE Pro as the mixer and what eg., Nuendo is getting is a submix scenario. In my experience, a project that approaches overload of resources is addressed very nicely by minimizing outputs. I parse things out to like 4 outs from a viframe/instance for reasons of organization and a visual cue to keep track. Before the automation mapping I used many outputs and mixed in Cubase as I was used to doing, a separate level per each output; automation just of the gain of the channel strip in VE Pro obviates the real need for that.

So the idea that i have a one to one corispondance between each articulation in each play instance 9 stereo outs per play instance is insane? i.e.  162 outs per frame back to Nuendo.

So is my issue the outputs racking up my nuendo meters and not just what is in use midi wise? I would have thoguht if the outputs were silent that Nuendo wouldn't be even wasting resources on them. I thought that was the way the settings were indicating in nuendo prefs.  This seems to go to like 50% utilizaion with it idle.  Maybe  shoudl set it to only output to the group channel level.  I.e types of instruments like cellos or trumpets rather then each of the audio outs for each of the 20 artiuclations for say 2 trumpets each on their own chanel. If i wanted to send out separat stems i could use the same group but have others muted i supose. but that gets tricky. and wastes time.

inside of Nuendo i have groups for mixing out puts l.e cellos flutes, oboes  and also have string sesion brass sections etc. then if i want an indidual instrument i select them on the output page as a batch output.

If i don't have those separate returns.. i won't be able to separate them on their outs easily.

I can get maybe 4 frames per machine before they start to protest and that is with a few tracks. If i try and do 1/2 of the string section it really isn't very happy.

 Here is a quick link to a video i did showing how this is setup.. Please comment and see if im going about this the hard and bone headed way or not.  Maybe i can lop off most of the outputs but i don't think so.

<< Template setup movie link >>

@civilization 3 said:

My approach is to give the heavy users amongst viframes more threads per instance. If all or most of your Play multis are equal quality users, I'm not real sure how much you're going to benefit by a more expanded setup. I really do suggest to anyone to do as little outputs to the host as feasible.

You say you're not doing MIR Pro, but I would relay one thing I find in this documentation and in tutorials, is that they are presenting scenarios of a single master output. More complexity in mixing really does not need be more outputs to the DAW host, ie., using VE Pro as the mixer and what eg., Nuendo is getting is a submix scenario. In my experience, a project that approaches overload of resources is addressed very nicely by minimizing outputs. I parse things out to like 4 outs from a viframe/instance for reasons of organization and a visual cue to keep track. Before the automation mapping I used many outputs and mixed in Cubase as I was used to doing, a separate level per each output; automation just of the gain of the channel strip in VE Pro obviates the real need for that.

@Another User said:

inside of Nuendo i have groups for mixing out puts l.e cellos flutes, oboes  and also have string sesion brass sections etc. then if i want an indidual instrument i select them on the output page as a batch output.

If i don't have those separate returns.. i won't be able to separate them on their outs easily.

Anyway, my experience certainly has been, the less I make Cubase receive from VEP in terms of outputs, the more I can load, and work with in real time.

the reason i made it a one to one return for every articulation is for indivdual printing of instruments.

Any idea of a better way to do that? 

I have those instruments feeding instrument groups then those instrument groups feeding sections groups

For example Butter leg. forte for the 18 violins. from vep play 1 midi 1 chanel 1 back to nuendo. 

I can then print that line only out for audio.  then it feeds into the Violins group which then feeds into the strings group.

This gives me a very flexable setup.

But yes its a lead baloon for using up resources. 

I supose i could route intruments to the "instrument group" level where by all violins come back or all cellos etc. to separate inputs then those can be routed to sections inside of Nuendo. but i would be loosing the ability to send out indivual articulations. 

Thoughts?

I'm not sure of what you're saying with the last part. I suppose your reason for printing every articulation is your own and I can't speak to it, but it seems like to me, eg., '1st violins' is going to be mixed as first violins, more or less. Or I'm just not going to get that fancy with it as a realtime scenario. I'm pragmatic and even so it takes a lot of time.