I would like to know why my thinking is wrong around this, but I don’t see why Render/Audio Mixdown export has to be done in real-time for tracks whose destinations are internal, but have external inputs.
As the operation implies, it is a render/export, so the input should not matter.
Would save a lot of time if real-time was reserved for tracks whose destinations only are external.
Of course the input matters – in the sense of “routing”.
Take a hardware synth for example, that is plugged into your audio interface. The synth is the external input.
You have a MIDI Track that sends notes to the synth, and you want to record the returning audio on an Audio Track.
Since the audio signal is analog, it has no other way than playing in real-time, the same as when you play the keyboard manually.
External Instruments and FX configured in Cubase make the rendering to be real-time automatically when you use them.
That scenario is relevant for sure.
I was approaching this from the point of view of the audio already being printed on the track and should have mentioned so. Although I forgot about the above scenario.
My solution is to turn off the external inputs or set the input to an internal bus.
I suppose unnecessary noise floor (from outboard gear) will also be avoided, assuming the audio is already printed.
I don’t really understand what you mean.
The input bus of an Audio Track never get printed in the render unless you have the track in Monitor mode. But even then, it has nothing to do with real-time or not real-time rendering.
If your tracks are monitored in will render the monitored signal instead of the Audio Events that are on the track. If the track isn’t monitored it only renders what’s on it, nothing else.
Real-time rendering isn’t involved in this.
You only need to enable Real-Time when you are using external real-time dependent analog signal path. Your microphone or guitar that is plugged into your audio interface and set as the input bus of your track has nothing to do with that.
What I mean is,
Track has external synth as input, for example
Record audio into track
Then, if I want to export or render the audio with some effects, it will be forced to do it in real-time, as long as the track still has externally set inputs (even though nothing is being captured externally – but Cubase thinks it is due to the inputs).
In summary, I understand why it works this way. The annoyance for me is having to turn off the inputs if I want to avoid real-time render of effects or audio export.
So you are using External Instruments/FX in the project ? You did not say it explicitly. Since the beginning I thought you were talking about physical input bus of audio tracks.
Indeed, in that case you must disable the Instrument Tracks or any Effects that are in the render path. But that kind of breaks the purpose of setting up and using external devices, if you use them in the project it’s usually because you want them to be rendered… Whatever. Sorry if I did not get it at first 
Cheers
Yes I am using external instrument for the track, but I record using traditional recording methods. That is, press record >> record audio into track. Then I must turn off the input in order to render or export while avoiding real-time processing.
Here’s what I don’t get
In other words, if I set an external bus as input, it makes more sense to traditionally record that audio, rather than using Render or audio export. If I set an external bus as the output to some track, then it makes more sense to use Render/audio export.
I think we don’t understand each other 
First, External Instruments and Effects (which is a feature of Cubase Pro only) gives the possibility to set up external hardware gear so they can be used as VST Instruments or VST Insert FX.
External Instruments can only be loaded on Instrument Tracks.
External Effects can be loaded like any other plugin.
When you use this feature, the physical Inputs and Outputs used by the External devices can no longer be used by any other track. They are exclusive to these devices.
For this reason, when using External Instruments, it is not possible to directly record the instrument as audio on an Audio Track.
Since the Instrument Track only contains MIDI, the only way to render the track is to do it in Real-Time, which is the explanation I gave earlier.
You say the following :
But the way you say it means that you are simply using an Audio Track to record your instrument, nothing else involved. Your instrument is let’s say plugged into input 1/2 of your interface, and on the Audio Track you select Input 1/2 as the Input Bus. That’s indeed the traditional recording method, same as when recording a singer with a microphone.
There’s absolutely no real-time outboard signal path involved in your setup. You keep saying that you absolutely need to disable/remove the Input Bus in order to avoid Real-Time rendering, but, could I please know why it is not working for you otherwise ? Do you get unexpected results ? An error message ?
Yes, that’s the way to do it without using External Effects, and in that case you indeed must record or render on the receiver track, in real-time.
Oh Ok I understand the confusion.
For this reason, when using External Instruments, it is not possible to directly record the instrument as audio on an Audio Track.
This is true, to a point. And I understand why what I was writing did not make sense. It’s because you CAN set external instruments/inputs to an audio track.
But there has to be a “middle man”. That is where the Group track comes in.
If you route any external signal to a group within Cubase, then you can use that Group as the input to any track. That’s what I am doing, and that’s what allows me to record (external) audio traditionally on those tracks.
I know we can do that, would have been easier if you explained your exact routing from the start.
That doesn’t change anything to the rendering method, this second “recording” track can be rendered however you want depending on how you use it :
If you have recorded audio on it already, you can simply do a normal render.
If you want to render the group, you draw an empty Part, enable Monitor, and render the Part in Real-Time, since your signal path requires to be processed in real-time. But since you can directly render the MIDI Parts of the instrument track, there’s not much interest of doing that.
Audio tracks that have a Group as the input don’t care about what’s before the group. These tracks are still considered normal tracks and should be handled as such.
So to answer your original concern, as long as the audio has already been recorded on that track, you can obviously do a normal (full speed) render, nothing forces you to use the Real-Time option, and you don’t have to remove the Input Bus. Unless Cubase forces you to use Real-Time ?
That’s what I’m finding - Cubase forces real-time even in the instance where there is audio already on the track.
So as long as it detects any external routing anywhere in the signal path, it will force real-time render/export.
What happens if you mute or disable the instrument track ?
In this use case there’s not much benefit to setting it up as an External Instrument, and some downsides as you’ve discovered. You could just use a plain old Audio Track and record the synth or whatever like you would a mic.
Nothing changes. Render is still forced into real-time processing.
The external instrument set up provides flexibility of auditioning recorded midi without having to set up an audio track.
Another use case here is printing the audio from the return of an outboard compressor, which I’ve set up as an external FX. Although in this case, since there is no midi involved, it might be a better candidate for setting up a plain old audio track set up with the inputs of the return of the outboard compressor.
I see. Looks like there are strong routing dependencies internally. It shouldn’t be doing this with a simple audio track. For some reason when doing this typical setup (Group into Audio Track) it does some really weird things under the hood.
Here’s a recent topic exposing some really weird issue when doing this specific routing : Mix Down to One Audio File incorrectly includes the content of unwanted audio tracks
The Audio Track’s input bus should obviously not seek for the full signal path as long as it is not monitored.
My guess is that fixing all of this would require to rewrite the audio engine from the start (which is planned but no ETA).
Thanks for sticking with me. Your last post summarizes the issue.
I did not explain myself specifically enough from the outset, but I’m glad we’re on the same page.
The Audio Track’s input bus should obviously not seek for the full signal path as long as it is not monitored.
^^ That’s what I’ve been trying to say 
@steve I hope you find the time to go read the above issues; it is summed up by the quote I’ve included here in my post from @Louis_R just above.