I grabbed a random power supply out of the junk bin and played around with it a bit. It's rated 100-240VAC input, output 5V 30A (150W). PFC circuit is based on the LT1248CS.
This is what I found:
Below approx 50VDC, does not start up or operate correctly, though it does try (no apparent input under voltage lock out). At 60VDC it runs fine, output of PFC regulator sits at about 360VDC which is as designed for this particular power supply. Was able to load it up to 50W with what power resistors I have kicking around (yes I know I really do need a nice big beefy electronic load), and with that load it runs just fine.
However, the efficiency at such a low voltage goes down the toilet: 62% efficiency (ouch!!!) at 60VDC, compared to 71% efficiency at 120VDC. This makes sense considering the input currents are more than double at the lower voltage, leading to increased losses in the NTC, bridge rectifier, PFC switching transistors and inductor. Note that server power supplies made specifically for DC-only operation in so-equipped datacenters lack a bridge rectifier and boost converter entirely, the whole point of which of course is to increase efficiency. Using standard AC power supplies on DC, while it technically works just fine, generally defeats the whole purpose because you still have those additional losses.
Efficiency at 120VAC is also 71%, so there seems to be no difference between AC and DC efficiency at the same RMS voltage, at least without modifications to the power supply. Note that efficiency at 240VAC input is 77%, so clearly the higher the input voltage the more efficient the power supply gets. This is definitely not a very efficient supply, but generally speaking all power supplies will exhibit better efficiency at higher input voltage, just it may not be as much of a huge difference with more efficient units.
Speaking of the bridge rectifier... that is one area to be cautious about, because the total current for the power supply will be passing through one half of the bridge rectifier, so its maximum current rating could potentially be exceeded or it could overheat. If you're going to do this, either get rid of the bridge rectifier (and remember to use correct polarity!) or make sure you're operating it within its ratings and it is adequately cooled.
I did grab a few other random power supplies from the stash and although I did not do as much in-depth testing, all seemed to be perfectly happy with 120VDC. Why 120VDC? That's as much as I could get with all my bench power supplies in series. It also happens that the nominal AC voltage in Japan is 100V, so most power supply manufacturers will design them to operate correctly to at least as low as 90VAC. The peak voltage of a 90VAC sinewave... 127V, Close enough.
To summarize:
-most PFC boost regulators will operate fine with DC input
-keep an eye on current through bridge rectifier diodes
-the higher the input voltage the better, I wouldn't go any less than 120VDC
-if possible, use DC-DC power supplies instead, as they will be more efficient