On Thu, Jul 21, 2011 at 12:38:51AM +0200, Axel Dörfler wrote: > pete.goodeve@xxxxxxxxxxxx wrote: > > wonder if the silence only happens > > on single-core units? Axel -- what's yours? > > I have a quad core where this happens. Well... that scotches that theory then. So if Sean is not getting a hiatus, and you are, what the heck is the difference? > > > I am not able to produce glitches with HDA (as long as the serial output > is turned off). > I guess its buffer size is large enough - it has been developed > primarily on Haiku, anyway. > (... or my thought that multicore might protect from glitches reaching the audio chain...) I took a look at the HDA source, and I see that the buffer size is set so that its period is a little under 1/20 sec (2048 frames at 44100 fps -- adjusted for different frame rates). Experimentally the latency at the mixer (before glitches) is essentially twice the buffer period -- I assume because of the one-buffer delay I remember being mentioned -- so I guess HDA users should see a latency of ~100 msec. [And that's rather heavy for me!] Be nice to get some data points from others... Anybody? -- Pete --