On Mon, 13.04.09 15:19, Tim Redfern (tim at eclectronics.org) wrote: > Hi I'm trying to set up pulseaudio on a small linux box and although its > working, the audio keeps stopping and starting constantly. When the > stream is playing the cpu use of pulseaudio is >50% which I presume is > bad. > > The machine is an NSLU2, 266mhz arm with 32MB ram running debian lenny > (pulseaudio version 9.10.4). - I see in previous discussions on this > list that it should be possible to use pulseaudio on small devices > without an fpu. I'm running pulseaudio from the CLI, the only modules > I'm loading are: > > load-module module-alsa-sink rate=48000 > load-module module-native-protocol-tcp auth-anonymous=1 > > I'm connecting to it with 802.11g - ralink module on the NSLU2 and intel > wifi on the laptop. > > I've tried everything I can see in terms of resampling the stream coming > out of audacious, setting the option --resample-method=speex-fixed-0 ... > > Are there any other options I could try? > Has anyone sucessfully run pulseaudio on a 'slug? Playing around with resampler is always just scratching on the surface of performance problems. PA running well on small devices you won't get for free. You need to sit down with tools like 'oprofile' to figure out where exactly the perfomance is lost. I myself am a desktop developer. While I try to design things in a way that they work fine on embedded systems too I do not do focussed optimizations for those platforms. I never do tests on embedded platforms myself. Hence, YMMV and without actually looking what really is going on this won't get fixed. For example, Nokia recently found out that one of the biggest bottlenecks (or I guess literally, smallest bottlenecks ;-)) on ARM processors is actually PA's use of poll(). Replacing that by epoll() already save quite a bit CPU time. Lennart -- Lennart Poettering Red Hat, Inc. lennart [at] poettering [dot] net ICQ# 11060553 http://0pointer.net/lennart/ GnuPG 0x1A015CC4
