Say I have a Footstep Event:
It’s one-shot and short and will be trigger around 4 times a second.
When I want to apply a reverb effect on it, which method would cost less CPU?
I imagine it would be 4=3>2=1.
And if there’re more than one tracks and events involve, 4=3>2>1.
Am I right?
The most optimal would be to use solution 4. The reason for this is if you aren’t using the reverb (eg. you’re not in an area that requires reverb) you can mute the return bus. Muted buses will cause their effects to go idle. In the case of 1, 2, and 3, even if you have the effect set to 100% dry it will still be active as there is signal going into it.
You can use snapshots to control the mixer return’s fader.
Hi Thank you for answering!
How do I mute the bus?
Simply turn the fader down to -∞ dbfs with snapshot?
It would be better to set the bus to -oodB in the normal mixer view, and use snapshots to increase volume. This way it’s muted without the need of a snapshot playing.
In a common game audio setting solution 4 works best also, because then multiple sound events can use and share the same reverb as a send effect. This is usually the most common case when multiple sounds are playing in the same space. And then you’ll only need one DSP instance for all the sound events.
Does muting a return track make any performance difference compared to just sending no audio to it? In our reverb system we have maybe 10 different reverbs on busses. As far as I understand effects are bypassed when they are silent. Is that correct, or would it help if we set the faders on inactive reverbs to to -∞ dbfs?
Yes, muting a return track can improve performance compared to simply stopping audio from being sent to it. In the latter case, effects on the return track may still remain active and continue consuming CPU resources.
Setting the faders on inactive reverbs to -∞ dbfs is recommended, as this will cause the effects to go idle, reducing CPU usage.
Okay that is very good to know! I was under the impression that reverbs would go idle when no audio was fed into them. Will have to think about a good system for this, because I don’t want reverbs from the previous room to immidiately cut off when entering a new room.
Would having a gain effect at -∞ dbfs BEFORE the reverb on the return do the same as turn downing the fader, or does it have to be the fader to make the effect go idle?
Does this also apply to sidechains? So if I have a sidechain on a group in the mixer, but not events are playing in that groupe, will the sidechain still be using more CPU than if the group was turned down?
Inserting a Gain effect before the reverb and setting it to -∞ dB silences the signal entering the reverb. However, the reverb effect remains active in the signal chain and continues to consume CPU resources, even though it’s processing silence. So yes, it has to be the fader to make the effect go idle.
Yes, sidechains can still consume CPU even if no events are playing in the group. To minimize this, you should also set the fader to be -∞db to mute it and save CPU.
However, if you are using FMOD Studio 2.03, setting the Input Gain of a bus or effect to -∞ dB can help optimize CPU usage as well.