**** BEGIN LOGGING AT Wed Apr 02 02:59:59 2014
Apr 02 08:42:05 <zGrr> moin :)
Apr 02 11:56:32 <Maxdamantus> Hmm .. turns out dmix seems to use a significant amount of CPU :(
Apr 02 12:26:17 * Maxdamantus wonders how pulseaudio uses less CPU.
Apr 02 13:11:19 <DocScrutinizer05> Maxdamantus: never heard of that
Apr 02 13:11:35 <DocScrutinizer05> usually it's PA that hogs CPU
Apr 02 13:12:47 <DocScrutinizer05> Maxdamantus: make sure your ALSA is correctly configured
Apr 02 13:13:46 <DocScrutinizer05> for dmix the IPC is very important
Apr 02 13:14:34 <jaska> on a desktop where cmus is playing mp3s pa usually takes more cpu than cmus
Apr 02 13:15:11 <Maxdamantus> Was trying mplayer and mpd.
Apr 02 13:15:30 <Maxdamantus> mpd was at around 90%
Apr 02 13:15:35 <Maxdamantus> mplayer about 30%
Apr 02 13:25:26 <Maxdamantus> what do you have to do WRT IPC?
Apr 02 13:26:03 <Maxdamantus>         type dmix
Apr 02 13:26:03 <Maxdamantus>         ipc_key 1532
Apr 02 13:28:40 <Maxdamantus> you can just try it in normal Freemantle with mplayer: echo 'pcm.meh { type dmix; ipc_key 1532; pcm.slave "hw:0,0"; }' >>/etc/asound.conf
Apr 02 13:28:52 <Maxdamantus> mplayer foo.mp3 -ao alsa:device=meh
Apr 02 13:29:16 <Maxdamantus> uses 30% of CPU for me, and audio is choppy.
Apr 02 13:30:07 <Maxdamantus> with the default alsa pulse plug thing, about 9% of CPU in pulseaudio and 5% in mplayer.
Apr 02 13:30:52 <Maxdamantus> (the 30% in the first one is in mplayer)
Apr 02 13:31:56 <Maxdamantus> I think the 5% remaining in mplayer is basically just decoding this flac.
Apr 02 13:35:47 <bencoh> flac ? foo.mp3 ?
Apr 02 13:36:08 <Maxdamantus> was an example.
Apr 02 13:36:17 * Maxdamantus doesn't actually have a file called foo.mp3
Apr 02 13:36:25 <bencoh> :)
Apr 02 13:36:41 <Maxdamantus> Well, there's probably a foo.mp3 in /tmp on my main computer, because I don't clear /tmp
Apr 02 13:36:49 <Maxdamantus> or not.
Apr 02 13:37:07 <Maxdamantus> foo.aux foo.c foo.cpp foo.git foo.gitearly foo.hs foo.java foo.log foo.o foo.ogg foo.pdf foo.py foo.tar.gz foo.tex foo.w
Apr 02 13:37:20 <Maxdamantus> ooh, foo.ogg
Apr 02 13:37:21 <bencoh> :))
Apr 02 13:39:32 <Maxdamantus> $ cat foo.w | sed 's/^ */ /' | tr -d \\n; echo foo ::= a @ v:a -| ( b:(a, a) ::= (v, v); return b ); return foo <- 4
Apr 02 13:39:46 <Maxdamantus> er .. foo ::= a @ v:a -| ( b:(a, a) ::= (v, v); return b ); return foo <- 4
Apr 02 13:39:47 <bencoh> obviously.
Apr 02 13:39:57 <Maxdamantus> some language I made a few years ago.
Apr 02 13:53:00 <Maxdamantus> Hm. Significantly better using aplay/dmix rather than mplayer/dmix
Apr 02 13:53:30 <Maxdamantus> Maybe it has to do with buffer sizes used or something.
Apr 02 13:54:30 <Maxdamantus> ah, aplay's not causing it to be resampled.
Apr 02 13:55:15 <Maxdamantus> still far better than mplayer .. even better than pulseaudio.
Apr 02 13:55:23 <Maxdamantus> 7%
Apr 02 13:56:56 <Maxdamantus> aplay doesn't work with pulse. Maybe aplay takes shortcuts that mplayer doesn't.
Apr 02 13:58:22 <DocScrutinizer05> Maxdamantus: hmm, no rate conversion
Apr 02 13:59:17 <Maxdamantus> I hadn't specified a pcm.meh.pcm.rate
Apr 02 14:00:25 <Maxdamantus> I get buffer underflows with aplay/dmix though.
Apr 02 14:01:13 <Maxdamantus> underrun*
Apr 02 14:02:20 <DocScrutinizer05> your pcm.meh doesn't look correct
Apr 02 14:03:14 <DocScrutinizer05> well, it kinda looks syntactically correct, but it lacks a rate inside dmix, and a rate conversion aka pcm type "plug" before the dmix
Apr 02 14:04:22 <Maxdamantus> I added a rate next to the slave afterwards.
Apr 02 14:04:26 <DocScrutinizer05> and I think the hw pcm needs some buffersize etc as well
Apr 02 14:04:37 <Maxdamantus> was under the impression dmix already did resampling.
Apr 02 14:04:46 <DocScrutinizer05> nope, it doesn't
Apr 02 14:05:08 <DocScrutinizer05> it's even worse since it can only have hw slave
Apr 02 14:05:11 <Maxdamantus> It doesn't pick a sensible buffer_size etc by default?
Apr 02 14:05:19 <DocScrutinizer05> and only works with sample rate of that slave
Apr 02 14:05:42 <Maxdamantus> So who's resampling? aplay?
Apr 02 14:05:56 <Maxdamantus> since my wave file is 44.1 kHz
Apr 02 14:06:05 <DocScrutinizer05> not necessarily, the application can define a range of preferred buffer sizes etc, and it will pick one from that
Apr 02 14:06:44 <DocScrutinizer05> I dunno. aplay source is pretty short, and educational
Apr 02 14:07:01 <DocScrutinizer05> check aplay -vv parameter
Apr 02 14:07:19 <DocScrutinizer05> it helps a lot for analyzing what's going on in alsa stack
Apr 02 14:07:58 <Maxdamantus> buffer_size  : 1536; period_size  : 512; period_time  : 11609
Apr 02 14:08:31 <DocScrutinizer05> you also see sample rates
Apr 02 14:08:50 <DocScrutinizer05> of each pcm, in and out
Apr 02 14:09:16 <DocScrutinizer05> alas not for pukeaudio
Apr 02 14:09:27 <DocScrutinizer05> obviously
Apr 02 14:09:49 <Maxdamantus> Direct Stream Mixing PCM: exact rate   : 44100 (44100/1)
Apr 02 14:10:04 <DocScrutinizer05> o.O
Apr 02 14:10:27 <DocScrutinizer05> wasn't N900 audiocard fixed to 48k sampelrate?
Apr 02 14:10:54 <Maxdamantus> afaik, which is why I've been presuming dmix has been resampling.
Apr 02 14:10:59 <Maxdamantus> or something under the dmix device.
Apr 02 14:11:05 <Maxdamantus> rather than the application.
Apr 02 14:11:18 <DocScrutinizer05> obviously nothing does
Apr 02 14:11:34 <DocScrutinizer05> which might explain why you get lots of buffer underruns
Apr 02 14:11:38 <Maxdamantus> Doesn't seem to underrun with period and buffer size 1024 and 8192
Apr 02 14:12:49 <Maxdamantus> They weren't constant before, but it happened every 10s or so.
Apr 02 14:14:50 <Maxdamantus> ah, setting the rate in also actually halved the CPU usage in mplayer.
Apr 02 14:15:34 <Maxdamantus> so it looks like mplayer was spending lots of time doing its own resampling (it listed the alsa output as 48 kHz, but played normally)
Apr 02 14:18:12 <Maxdamantus> wait, that's with 44100 in asound.conf
Apr 02 14:18:18 <Maxdamantus> not 48000
Apr 02 14:18:50 <Maxdamantus> maybe the hardware does do 44100
Apr 02 14:20:39 <Maxdamantus> so mpd is also around 15% like that, which seems acceptable.
Apr 02 14:21:07 * Maxdamantus thinks he might have a few songs in 48 kHz
Apr 02 14:29:09 <Maxdamantus> oops, was using hw:0,0 in mpd.
Apr 02 14:29:51 <Maxdamantus> mk, so still at 15% with mpd/dmix, but it can't play 48 kHz songs.
Apr 02 14:29:58 <Maxdamantus> only have one in my library, actually.
Apr 02 14:30:18 <Maxdamantus> Radiohead, Supercollider
Apr 02 14:38:19 <Maxdamantus> mplayer can almost downsample it without breaks.
Apr 02 14:38:49 <Maxdamantus> AUDIO: 48000 Hz, 2 ch, s32le, 1424.2 kbit/46.36% (ratio: 178022->384000)
Apr 02 14:46:59 <Maxdamantus> Mm, mpd has a cheaper "internal" resampler than the default libsample one.
Apr 02 20:58:34 <Zmr56> What the hell is this?
Apr 02 20:58:41 <Zmr56> Who are you chuckle brothers?
Apr 02 20:59:05 <Zmr56> And why are there so many people here?
Apr 02 20:59:23 <Zmr56> Already ab is my favourite guy here
Apr 02 20:59:27 <Zmr56> You joker!
Apr 02 21:19:23 <hxka> wat
**** ENDING LOGGING AT Thu Apr 03 02:59:59 2014