[LAD] ALSA MIDI latency test results are far away from reality

Ralf Mardorf ralf.mardorf at alice-dsl.net
Wed Jul 14 20:24:31 UTC 2010


On Wed, 2010-07-14 at 22:16 +0200, Ralf Mardorf wrote:
> On Wed, 2010-07-14 at 21:45 +0200, Robin Gareus wrote:
> > On 07/14/2010 03:23 PM, Ralf Mardorf wrote:
> > 
> > [..]
> > > $ su -c "chgrp audio /dev/hpet"
> > Its also writable for the group, right?
> > 
> > > $ su -c "sysctl -w dev.hpet.max-user-freq=64"
> > 
> > Documentation on this value is hard to come by, but I think it's unit is
> > Hz. So you'll want
> 
> Yep :). I've got no idea what would be better a high or a low value, but
> at least 64 when using HR timer was better than using the regular timer.
> Unfortunately not many apps are able to run, when HR timer is used.
> Usage of HR timer can freeze the system, e.g. Rosegarden does this on my
> machine. I also need to start 'qtractor filename'. If I first start
> Qtractor and than try to load a session, here are issues because HR
> timer already is captured.
> 
> I guess one would be more flexible, if not using HR timer. I didn't
> tested if there will be a difference when using the PCI MIDI. I'll do
> this also ASAP. As I mentioned before, for USB MIDI there is a big
> difference.
> 
> No thunder, no storm anymore, but there was a rainbow some minutes
> ago :)
> 
> - Ralf
> 
> >   su -c "sysctl -w dev.hpet.max-user-freq=1024"
> > or even more :)
> > 
> > try:
> > echo 2048 | sudo tee /sys/class/rtc/rtc0/max_user_freq
> > echo 2048 | sudo tee /proc/sys/dev/hpet/max-user-freq
> > 
> > [..]
> > ciao,
> > robin
> > 
> > 
> 

Could it be that a low value is better because it's some kind of 'be
ready for an IRQ every 64 times (of what unit ever)'?

IIRC a long time ago I tested 64 and 1024 and couldn't hear or see (for
the waveforms) a difference.





More information about the Linux-audio-dev mailing list