Did a bit of experimenting stacking up gainer on top of each other with the max +12dB amplification. The count came up to 57 before the signal went suddenly distorted into nothingness.
I wonder what is the internal precision of calculations in this software?
Wanted to add a screenshot, but it didn’t work, nevermind.
Conversely, the quietest level that has been registered was -590dB, so overall the dynamic range of a track is about 680 + 590 = 1270dB. As one gets 6dB per bit, we have 1270/6 near 212 bits of resolution. It is more dynamic range than one could possibly imagine what to do with.
I find it rather impressive, that is unless I made a misstep in calculations somewhere.
The observations are rather weird though, because last thing I heard was that Intel processors have 80 bit wide float point registers, which would leave one with 80*6 = 480 dB. The only explanation could be that at least three float point numbers are used per calculation but that would be rather unnecessary since 480 dB of dynamic range would still be perfectly sufficient.
Did you turn off the track headroom in the song settings/song playback options?