I sent this out (somewhat paraphrased) in an email to some friends earlier today. I thought it might be worth also posting here, for posterity.
Caveat: Some of this is really only relevant to Rec diving to NDLs, so take away what is useful to you and ignore the rest.
On algorithms:
Here is a link to a post from RonR, who IIRC (I *think*) from other posts is an engineer at Atomic Aquatics. The summary is that different computers can implement their algorithms differently, even if they are the same algorithm. So for example, one computer that implements Buhlmann ZHL-16C might still give you a different dive profile than another computer that also implements Buhlmann ZHL-16C. Needless to say, computers with different algorithms (Buhlmann, RGBM, VPM, Haldanean/DSAT, etc), can give very different dive profiles.
Read RonRs post for the full explanation:
http://www.scubaboard.com/forums/co...-leonardo-too-conservative-4.html#post7342293
I think its also worth noting that many computers use a proprietary implementation of their algorithm. So, attempting to predict what the computer will do by modeling your plan on your desktop computer may or may not produce exactly the same results, for the same reason above. The best way to be assured of the best match between desktop computer and dive computer is by using the desktop software that is provided by the OEM of your dive computer (IF they do in fact provide any). Aftermarket software, for example V-Planner, may use the same algorithm as your dive computer, but it could still be implemented differently, thus giving you a different dive profile.
On conservatism:
The first post in this thread has a simple chart comparing 1st dive NDLs for a variety of different types of computers. One example from the chart: At a depth of 100, USN and NOAA tables allow 25 minutes, PADI RDP allows 20, the most liberal dive computer allows 19, and the most conservative dive computer allows 8 minutes. Of course, pretty much any dive computer will let you adjust it to make it more conservative, so you CAN use the most liberal computer and still have it limit you to not come so close to the PADI 20 minute limit (as one example), if you want.
However, my research convinced me that looking at what the computers do on repetitive dives in the same day is at least as important to look at. For that, Scuba Magazine did a couple of good articles where they put all the computers they were testing into a hyperbaric chamber and replicated doing 4 dives of different profiles, with surface intervals in between. They logged the NDL shown on each computer for various simulated depths.
Their 2014 test results are here:
http://ads.bonniercorp.com/scuba/PDF/ScubaLab-Computer-Test-September-2014-data.pdf
which is from this article:
SCUBALAB 2014: Dive Computer Review - Page 2 | Scuba Diving
As one example, this test shows a 1st dive to 100, then a 60 min surface interval, then a second dive to 70. After EDT (Elapsed Dive Time, on the 2nd dive) of 18 minutes, at 70, the most liberal DC shows an NDL of 20 minutes and the most conservative shows 12 minutes. The differences are even bigger when the diver starts to ascend some and spend time at shallower depths.
The differences get even more dramatic on the 3rd dive, which is after a 120 minute SI, to 80, then gradually working up to 40. After EDT of 36 minutes, at a depth of 40, the most liberal computer allows 56 minutes of remaining time before the NDL, and the most conservative allows 2 minutes.
56 minutes versus 2 minutes is a pretty big difference!
Scuba Magazines test results from 2007 (I think) are here:
http://www.scubadiving.com/files/old/images/pdf/scubalab_200706_10_new_comps.pdf
which is from this article:
dive computers: scuba gear reviews | Scuba Diving
This one is kind of more of the same, but with some different computers, as the first article.
I hope this helps someone as much as finding it all helped me when I was shopping for my computer.