My thinking on this may be without technical merit, it is just my thinking. That is with amp volume at 2/3 to 3/4 ish, the amp has a good portion of it's capacity available for sudden demand - "on tap", such as percussion or symphonic transitions, even though the volume of the DAC is not at a high level, so the actual average listening volume is moderate.
I suppose this is based primarily on hearing the difference between an low powered amplifier that will make sufficient volume with a given set of headphones, and then hearing the same headphones powered by a higher wattage amplifier that makes a big difference in bringing out the potential they have. I'm sure I'm not articulating this well, as a lay person in the technical aspects of DAC and amplifier interaction. At least I am educable
How about looking at it this way. A more simple way I imagine.
Unless you have a DAC that bottlenecks the Dynamic Range of the amp...
DAC for example: 115dB Dynamic Range rating
AMP for example: 125dB Dynamic Range rating
Lets say one or the other will be left at 3/4th ish like you mentioned, while the other is at full output to make this concept easier to understand.
If we leave the Amp at the standard 2V/4V (for this example, this being the max output), and its now capable of 125dB DR, we are going to have the DAC at that 3/4th level, thus producing 85.25dB max Dynamic Range.
If we leave the Dac at the standard 2V/4V (for this example, this also being the DAC's max output), and its' now capable of it's max 115dB DR. We then lower the Amp to 3/4th level instead, and now the amp producing 93.75dB of max Dynamic Range.
So it seems to me, whichever device is cleaner, that's the one I would mess around with volume if I cared about Dynamic Range only for example. But then again...
-----------------------------------------------------------------------------------------------------------------------------
I am also a layman and don't have a full grasp on how many things are calculated (like this whole "on tap" concept.. if that was the case, what is volume control even doing? If someone was to produce a piece of music with massive inter-sample peaks or something, could we theoretically get the amp to output volume as if the knob was turned 4/4th (100%) of the way turned up.
I don't know what the headroom theory is all about outside of context of intersample peaks with respect to potential clipping and such. Likewise how or what determines clipping depending on gain without actually testing the device. So with all I said, I could be literally talking out of my ass. >_<