Its being used dependend on how "hot" the input signal is with the purpose
- to prevent overloading inputs
- to set the input sensitivity of the converter to be able to operate in an optimal work range according to the strength of the input signal

See KaiS's last recent contributions to this topic in the "ADI-2 Pro & ADI-2 DAC" subforum.

This topic is very relevant to my interests and everything fl just said is totally spot on.

I have only just very recently on my fireface400 started to have the LINE IN setting as Lo Gain instead of the +4dBu I had always been at with my FOCUSRITE ISA ONE because it really puts out a HOT signal.  When I'm at LoGain and have the mic (acoustic guitar) set about right, I'm coming in at -12dbfs peak to -15dbfs peak which is a nice good level.  The +4dBu setting it's just a bit too hot and was -4dbfs -6dbfs in there.  As fl said, this is the balance between preamp and rme coming in.  Lower preamp signal, +4 on rme.  Higher preamp signal LoGain on the RME.  There is also no law or rule that says you cannot use the -10 signal on a "pro" piece of gear but generally that -10 level is for drum machines or similar.  For me, the LoGain works better than the +4dbU.

I am really trying to force myself to record at much LOWER LEVELS like -18dbfs to -12dbfs PEAK (not RMS).  This is for analog stuff of course going in.  For digital and VSTi stuff I use the output of the instrument itself to peak around those same levels on main stereo out of my daw.

I have been guilty of recording WAY TOO HOT and it's a hard habit to break..not seeing such fat waveforms in the project window it "feels" like it's too low but I know the headroom is needed and there is just absolutely zero reason at all to blast the hell out of the analog signal coming in trying to nail down fatty wave forms.  I used to track to the point of JUST BARELY not clipping and that is absolute worst possible thing to do... hard habit to break but I'm getting better!  You track and peak at freaking -1dbFS and as soon as you add a small bump of EQ you are clipping.  Stupid.  I was stupid. 

Not too stupid to learn, by appearances...

For what it's worth, I have a Fireface 800 which I use for location classical music recordings with a Millennia Media eight channel pre-amp. After experimenting, I found I got lower noise by using the Lo Gain setting and making the corresponding increase in the gain settings on the Millennia. A few years ago, I came across a proposal to make -20 dBFS the ideal target level for recording. This seemed like a good idea to me - it's four dB hotter than the North American Broadcast standard for Loudness, and 4 dB less than (what was then) the iTunes standard. In short, it seemed like a happy compromise. After several years of doing this, I have to say that my Fireface converters sound better - sweeter - to my ears, when they are not running with too hot an input level. I rarely if ever encounter clipping on the inputs to my gear. Recording with lower gain settings certainly allows me greater flexibility when it comes to mixing and mastering.

As for outputs, I monitor through a hi-fi amp, so I set the outputs to the -10 dBV setting, and that works well. For CD production, I've had no complaints from any clients when I supply them with mastered CDs and downloadable files which adhere to a -20 LU Loudness standard. Occasionally, I'll need to place a limiter on the output bus of my DAW, in order to maintain the -20 level, but I rarely if ever see gain reduction in excess of 2 or 3 dB with it, and it's usually for big orchestral accents with timpani, bass drum or snare drum, so nothing that lasts longer than an instant. Personally, I don't subscribe to the Loudness Wars idea of making each production louder than anything else - the sacrifice in transient sharpness, particularly for percussion is just not worth it to me just to have a louder playback level. I've got a volume knob, and I'm not afraid to use it! However, as always, YMMV.