honestly, I did a load of tests with various kernel parameters, and they seem to make ‘marginal’ differences - there was some, but it was debatable if the effort was worthwhile.
(the key thing was to always make sure the governor was using performance and not ondemand, which is the thing that tripped most of us up, as it tends to get reverted )
anyway , I then I got a rPI3b+ , which has a 1.4 rather than 1.2ghz cpu, and it all seemed to be irrelevant - gives us 200mhz headroom over the CM3 on norns.
also, with 2.0, pushing some of the processing to other cores, will probably have more effect than these kernel optimisations, so might become less relevant.
… on the flipside, this is why I mentioned using the Blokas Patchbox OS, as its using an optimised kernel.
Ive not played with it yet, but I really like the idea of an open source, audio centric, pi distribution.
what makes it ‘next level’ for me, is its built on pi-gen, so we can fork it, and adapt it.
so we could easily make a derivative for norns…
(or my other projects which is really my main focus
)
as an aside, recently I upgraded my norns rPI to the latest stock kernel, (due to soundplane), but not done any tests to see if that’s had any adverse effects. (my suspicion is it’ll be fine)