2009/8/8 Jakub 'Livio' Rusinek
Am Samstag 08 August 2009 02:09:27 schrieb Greg KH:
Yes, PA uses high-resolution timers, which have nothing to do with the HZ value of the kernel.
Lennart Poettering tells otherwise... Tells that each distro but Fedora sucks (because he doesn't even know why PA works best on Fedora, which is awfully funny).
If you are talking about https://tango.0pointer.de/pipermail/pulseaudio-discuss/2009-February/003150....... note that changing HZ from 250 to 1000 is a change from 4ms to 1ms. He is talking about 5ms (Fedora) vs 210ms (openSUSE). He says he obtained the results from alsa-time-test (http://pulseaudio.org/wiki/BrokenSoundDrivers). I suppose these 5ms/210ms values come from clock_gettime(CLOCK_MONOTONIC), the first column. Well, in openSUSE 11.1 I indeed get asserts from that test if I write the output to my main HD. And these asserts come after a sometimes ~200ms clock_gettime(CLOCK_MONOTONIC) jump. But once I use my second HD, that no other process uses, I don't get asserts anymore. So those ~200ms values seems to come from underruns after a *momentary* high CPU/HD load. The normal latency is of ~8ms. If someone runs the alsa-time-test in openSUSE and sees a problem then open a bug report at https://bugzilla.novell.com/, but on my machine I don't see any problem. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org