On 7/14/20, Till Bovermann <lf...@lfsaw.de> wrote: > Dear James, all, > > your hint helped a lot, thanks, James!
I'm glad. > However, after reading this paper > https://hal.archives-ouvertes.fr/hal-02159011/document > > (thanks for the pointer, sletz!) > > I think that computation still happens for every sample, despite the input > value (i.e. previous state) not changing and thus the output also keeping at > a constant value. > > Maybe an expert in faust optimisation could shed some light on this (and > proof me wrong?) I can't check it carefully right now, but yes, from a quick inspection of the code, it seems that in this case f will be recomputed every sample inside demand_1. It may be possible to avoid this by rearranging demand_1 such that f is inside the latch? In this particular case I guess it doesn't matter very much as statemachine is hardly very computationally intensive. I will say that if you know C then a glance at the generated C code can clarify what is being calculated. I have some more thoughts on this that are perhaps worth discussing in a separate thread. Briefly, I feel like a 'dumb' select2, which only ever evaluates one of its branches, would be a useful addition to the language for applications such as this. I am considering trying to implement it myself, and I wonder whether a patch/pull request for such a primitive would be accepted? I'll try to write up my thoughts on this more fully later. Cheers, James _______________________________________________ Faudiostream-users mailing list Faudiostream-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/faudiostream-users
