New Decoding Model

[jvde-github]

I've added a new decoding model to AIS-catcher.

On my mini PC it uses about 45% less CPU than the current model and decodes 5–7% more
messages, though this likely depends on the circumstances — I expect the biggest uplift in busy areas. TBC.

If you run the latest, in managed mode you can activate it under Input: unfold Engines and select
AIS Engine v2 Base (experimental):

You can also run both models at the same time and compare them in the Plots tab —
pick one as Source and the other under Compare with:

From the command line it's -m 11, or run the current and the new model in parallel
with -m 2 -m 11.

It's marked experimental for now, but the intention is to make it the default once it has
some more mileage — so feedback on how it behaves on your station is much appreciated.

[doc-nl]

Just enabled it, comparing 1 high with 2 base now. Seem to helps about 4% will let it run throught the night and switch to 1 base and 2 base tommorow

[jvde-github]

Thanks

[doc-nl]

When comparing RTL SDR to an engine gives a much bigger difference than comparing the two engines. What am I comparing in the first case?

[jvde-github]

That looks like 2 different input sources. RTLTCP is typically data you get in via a network connection and not 2 different decoding engines on the same stream.

should be able to see in Input tab where it connects to?

[jvde-github]

One thing that needs fixing is the ppm estimates, seems to be wrong…

[doc-nl]

Ah off course, I was re-routing some data from one node to another, temporarily, that caused the extra input