The typical error when stressing a DAW too much is dropouts or glitches due to a buffer that empties. There isn’t some sort of gradual degradation of the audio in order to avoid that.
And if you think it’s a static issue due to CPU demands then you’re really talking about limiting the resolution either on one ‘axis’ by limiting dynamic range or the other by limiting bandwidth - i.e. you’d have to program the DAW to lower the sample rate or decrease the bit depth to reduce CPU demands. That’s a ridiculous suggestion. And why those items specifically? Because if you’re talking about the basic audio engine then it’s just very, very simple math we’re dealing with. It’s nothing fancy. It really isn’t. Wanna double something? Multiply by 2. Done. Cut it in half? Multiply by 0.5. That’s it.
So how can you make X * 0.5 any less demanding on the CPU? You can’t change the math because all it is is multiplication. So the only thing that’s possible is making X a smaller number to begin with - i.e. reducing bit depth. Oh, well I suppose the other way of doing it is to do fewer calculations per second - i.e. lowering sample rate.
If you really think that any of the major companies are doing this then you’re truly misled.
Numerous tests have been done on the audio engines of the major DAWs and guess what - they all null to zero or into the dithered noise floor. I.e. they do the math the same way minus pure noise that we can’t hear anyway.
I find it simply unbelievable that people still entertain this silly notion in 2019…