Very interesting question, but the answer is not easy at all ! The code generated by the Faust compiler has a fixed memory footprint and bounded CPU use. But since the compiler itself has a lot of code generation options (look at what « faust -h » gives), and the C++ compiler used to produce the binary also has a bunch of optimization options, things can become complex quite rapidly. Another complexity is that the Web platform uses another compilation path: Faust DSP => WebAssembly => executable (thanks to the Web browser wasm JIT compiler).
One idea for your use case would be to estimate the ratio between running a Faust DSP on the Web platform, and on a given hardware device (CPU usage), and hope this ratio is « constant enough » across a bunch of different DSP programs (so that knowing the DSP usage on the Web platform could help estimate what would happen when running another DSP on the hardware) We could probably add a « CPU usage at several SR tool » on the Web platform. BTW: in a more general term, you can look at the following pages for infos on optimizations tools: https://faust.grame.fr/community/news/index.html#optimizing-the-dsp-cpu-usage https://github.com/grame-cncm/faust/tree/master-dev/tools/benchmark Stéphane > Le 4 juin 2019 à 15:10, Gary Worsham <gary.wors...@gmail.com> a écrit : > > I had some fun with the seq iterator cascading tons of all pass filters. I > did so many that I locked up the web editor, right at the point it was > sounding REALLY interesting (around 128). > > That's not a huge problem, but supposing (in general) that I am interested in > a standalone hardware device, e.g. Bela or Hoxton OWL or any other thing with > an Arm CPU (for the moment) - how can I predict the most complex single > processing path in my patch that will be "calculable" at any sample rate? > > Thanks, > > GW > > Virus-free. www.avast.com > _______________________________________________ > Faudiostream-users mailing list > Faudiostream-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/faudiostream-users _______________________________________________ Faudiostream-users mailing list Faudiostream-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/faudiostream-users
