@Schackmannen, thanks, that's interesting to learn.
You wrote: "if you didn't hear any noise with the volume turned all the way up, then you won't hear any even if you turn the volume way down."
I strongly disagree. It is usually in the quiet passages of music that a listener notices recording and instrumentation noise. Our brain simply can't notice low level noise when there's loud music playing but if you listen to music that has subtle quiet moments (such as classical music for instance) you'll notice background noise pretty easily. The saying that human hearing has 116dB of dynamic range is misleading. There are two ways of looking at a dynamic range, with and without gain control. Will you be able to notice 0dBSPL in a totally quiet environment? probably if your ears are good. Will you be able to withstand 116dBSPL, probably with some pain and ringing. Will you be able to hear somebody whisper while in a rock concert? No, you won't. The "Gain control" in your ear (which uses mechanical means) and the brains abilities won't let you.
You wrote: "I'm not sure why you think that it's "practically unusable with IEMs"."
It's not only me. Please see
@IVX note above: "In case if you bought 9038S to pair with high-sensitive IEM, and you have to listen to my DAC at such crazy low volume, you made a mistake."
You wrote: "The only issue I could see is if the digital volume control doesn't have enough steps so it's hard to adjust the volume to the right level,"
That's exactly the problem.
My claim is that if you implement volume control by attenuating the analog output of the DAC before (buffering and) feeding it to the headphones you'll be able to retain most of the wonderful SNR you worked hard to achieve. This is limited to the attenuator's noise floor of course but it should be possible to have a much lower attenuator noise floor than that of the 32 bit main DAC.
The link to RME's document will probably be a very interesting read. Thank you for that.