This is not a stock kernel; it's from f8-based planet-ccrma, 2.6.24.7-1.rt3.2.fc8.ccrmart to wit. Here's /proc/asound/timers: G0: system timer : 1000.000us (10000000 ticks) G1: RTC timer : 976.562us (100000000 ticks) Client sequencer queue 0 : stopped P0-0-0: PCM playback 0-0-0 : SLAVE P0-0-1: PCM capture 0-0-1 : SLAVE and asound/seq/timer: Timer for queue 0 : RTC timer Period time : 0.000976562 Skew : 65536 / 65536 So what constitutes "good, tight midi playback" with alsa these days? Seriously, is it just broken, or should I be trying to get better performance from it? Thanks, Steve On Mon, Jan 12, 2009 at 4:58 PM, Pete Black <pete@xxxxxxxxxxxxxxxxx> wrote: > Good, tight midi playback is pretty much unacheivable in Linux with the > stock kernel due to scheduling jitter. Realtime kernel + 1000Hz timer is > probably best, but you can improve things with a standard kernel by > increasing the timer HZ setting of the kernel to a value around 1000Hz but > that still means a recompile. The 'stock' kernels shipped by many linux > distros are really, noticeably bad for MIDI - they have the timer clock set > as low as 100Hz. > > -Pete > > >> Hi. >> >> I use jackd as follows: >> /usr/bin/jackd -v -R -P70 -dalsa -r48000 -p128 -n2 -D -Chw:0 -Phw:0 -Xraw >> >> Here's my output from alsa-info script: >> http://www.alsa-project.org/db/?f=8b1149c838d952854b79008a7a9b1980bd3ba0ff >> >> I notice that when I record MIDI from my vDrums, the playback is never >> quite right. It doesn't matter what sequencer (e.g. qtractor, >> rosegarden, etc.) I use to play a midi file. >> >> I tried a little experiment: >> 0) record some midi of my drumming >> 1) use qtractor to send MIDI through my Midiman 1x1 USB midi interface >> to the vDrum brain >> 2) use ardour to record the audio output of the vDrum brain >> 3) repeat steps 1 & 2 >> 4) compare the waveforms of the resulting audio tracks >> >> I get notes that vary between the files by 30-40 ms. Incidentally, if >> I bypass alsa and create a WAV file with timidity, it's spot-on >> perfect. >> >> So I don't know if alsa sequencer is just broken or if there's some >> configuration/tuning that I could do to improve this. >> >> Any input appreciated (helpful input preferred :) >> >> Thanks, >> Steve >> >> >> ------------------------------------------------------------------------------ >> This SF.net email is sponsored by: >> SourcForge Community >> SourceForge wants to tell your story. >> http://p.sf.net/sfu/sf-spreadtheword >> _______________________________________________ >> Alsa-user mailing list >> Alsa-user@xxxxxxxxxxxxxxxxxxxxx >> https://lists.sourceforge.net/lists/listinfo/alsa-user >> > > ------------------------------------------------------------------------------ This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword _______________________________________________ Alsa-user mailing list Alsa-user@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/alsa-user
