I used Multi-Deco with GF 100/100 (I don't remember if I used Buhlmann B or C) to compare dive profiles that required no stops to what the Oceanic manual said for NDLs for the same square profiles. The results were within 1 minute or so at all recreational depths.
I understand what you said, but it doesn't seem to refute my thought that diving with a GF High of 85 is more conservative than doing NDL dives on an Oceanic computer with DSAT and no CF set. The OP specified 45m or less, which is another 20' beyond what I looked at, but I'm skeptical that the results would diverge by a huge amount in that bottom 20'.
If your intention was to refute that notion, then I am not knowledgeable to draw the line from what you said to that conclusion. If that is the prooper conclusion, I would be grateful for more explanation to help me understand.
What a computer does for a no stop dive and what it does for a dive with stops are not at all the same thing. The NDL will be reached when the leading compartment gets to the over saturation limit, but if you start doing stops then other, probably slower, compartments will come into play, later those will become the limiting compartments while the fast one will have come down to ambient. You are trying to compare curves derived from a dozen or more equations with a single data point.
Just because one computer has shorter NDL times than another does not mean that when doing stops it will get you out with less deco stress.
To say that computer X is more conservative than computer Y you need to compare those equations. You may find that some compartments allow more oversaturation and some less. Then what? It starts to depend on the dive profile in question. You cannot generalise.
In particular I seem to remember reviews of the DSAT ones saying that it gave much longer stops than the VR Technologies computer using ZHL16C see A sense of algorithm