I thought somebody worked out that DSD64 is roughly the equivalent of 24/117 or so.
Hard to compare directly. DSD's dynamic range or effective bits starts to decline after 24-25Hz because of the increase in quantization noise (from the inaccuracies of converting to 1-bit) Sonic information has a tendency to decrease in amplitude as we go higher up the ultrasonic range (whether we can hear the effects of this has nothing to do with this argument.) What happens is that DSD's quantization noise ends up actually higher than the actual sonic content in the ultrasonic region, effectively masking any real information.
Argument #2, all SACD and DSD software players filter this ultrasonic DSD quantization noise. So there's little past 40kHz, and what remains between 25-40kHz is still noise. Let me refer to the same graph I posted earlier and illustrate.
The three frequency spectrum analyses are not directly comparable because the they were taken from three different tracks, but you get the idea. (There's not huge difference in the frequency analysis in the areas which we want to look at here in this instance.)
So where's the hires in DSD? Almost better off with 24/48.
Technically, DSD is one of the biggest scams pulled off on consumers.