Ever seen a computer malfunction in this way?

[Belzelbub]

There could also be these issues as per the quote below from SubS

OK. Fair enough. That's a good point, and one I hadn't considered. However, in practice, I would expect them to be much closer. I just checked more recent dives on Shearwater Cloud and Subsurface using my Perdix. Perdix was set to 40/85, and Subsurface was the same. At the same point on the same dive, they were both showing the same NDL.

This article is a bit old, but probably still relevant, though I would expect that some of the algorithms may have been tweaked since then. Shows how several different algorithms behaved on the same dives.

A sense of algorithm

Most of the algorithms more or less agreed with eachother, though Oceanic DSAT and Mares RGBM PF1 were significantly more conservative on Dive 2. Mares RGBM PF0 seemed more inline with the others, but this was after a 2:45 SI. PF0 may have been significantly more conservative on Dive 2 with a shorter SI.

 

[Jay]

OK. Fair enough. That's a good point, and one I hadn't considered. However, in practice, I would expect them to be much closer. I just checked more recent dives on Shearwater Cloud and Subsurface using my Perdix. Perdix was set to 40/85, and Subsurface was the same. At the same point on the same dive, they were both showing the same NDL.

This article is a bit old, but probably still relevant, though I would expect that some of the algorithms may have been tweaked since then. Shows how several different algorithms behaved on the same dives.

A sense of algorithm

Most of the algorithms more or less agreed with eachother, though Oceanic DSAT and Mares RGBM PF1 were significantly more conservative on Dive 2. Mares RGBM PF0 seemed more inline with the others, but this was after a 2:45 SI. PF0 may have been significantly more conservative on Dive 2 with a shorter SI.

How close various algo's are is interesting to me as well. For Rec dives, DSAT is on the liberal end of the scale. Pushing it beyond Rec (like in the article you referred to) gives it a meltdown. The various branded RGBM models are on the conservative end for repetitive dives and if (as you mention) shorten SI (to say an hour ...) they become even more conservative. My view is that the RGBM is more or less dead, and RGBM2 (along with Fused / Tec algo's by Suunto) will be the way forward for ‘bubble’ models ... and they are very close what you'll get from Bulhmann on a x/95 setting and DSAT - although SI times are very relevant.

One of my favourite algo difference 'study' is ScubaLab's. They haven't yet published this years spreadsheet. When to sample NDL reading during pot dive is important (at many times of a dive they're not too relevant nor stable). Here's a link to some analysis I did from their 2018 data: Post 8 has 2018's data. How to properly assess the relative conservatism of various algos/computers for Rec divers?

An older article that was interesting for me was Algorithms And The Modern Diver

ScubaLab's 2019 article is here: The Best Dive Computers of 2019, Reviewed by ScubaLab hopefully they'll soon add their data file.