To be fair, my testing conditions are not really idea, so I'm not sure how accurate my number is. For example, if I use speex-fixed-10, CPU usage doesn't seem too high, but I get underruns for a few seconds, then the CPU gets lower (to around 10%) and sound is fine for a while. I suppose this is some kind of feedback loop that includes module-rtp-recv. That module uses the resampler to adjust for the clock rate skew; to make things worse, the latency on my network is pretty bad. I could try plugging in a network cable, but I don't know a better way to test the performance. Laurentiu Nicola On Sat, May 10, 2014, at 20:53, Alexander E. Patrakov wrote: > 10.05.2014 23:37, Lauren?iu Nicola wrote: > > Sorry for barging in on this, but is there a way to run a benchmark on > > all the resamplers included in PulseAudio? For me, both speex-float-6 > > and speex-fixed-6 eat around 40% CPU. > > Congratulations for being a perfect mind-reader. I was going to ask the > same question in > http://lists.freedesktop.org/archives/pulseaudio-discuss/2014-May/020624.html > - and you already have an answer. So, based on your benchmark, if others > confirm it, let's just kill speex-float on all platforms. > > -- > Alexander E. Patrakov > _______________________________________________ > pulseaudio-discuss mailing list > pulseaudio-discuss at lists.freedesktop.org > http://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss
