On Thu, Mar 05, 2015 at 02:42:11PM -0300, Henrique Lengler wrote: > > On Thu, Mar 05, 2015 at 01:45:57PM +0100, Alexandre Ratchov wrote: > > Let's first fix system-related audio problems (as programs depend > > on it). Once we're sure audio works well, then we can try to debug > > browsers which are known to not work well yet in all cases. > > > > mplayer is known to work; it has a small lag (iirc, less than > > 500ms, depends on the setup) which is somewhat acceptable. Could > > you confirm at least this works on -current as expected? > > No, still the same thing. mplayer, cmus and html5 video in any browser > (firefox, chromium, webkitbased), are with lag in my system. > Also please remeber that the audio is synchronized, the lag is when I want > to advance the video.
I still don't 100% understand whether you observe an audio subsystem bug (>500ms latency), or you just dislike the default latency. If you don't like the 200-400ms latency (too conservative imo), you could lower sndiod buffer size, possibly patch mplayer, cmus or whatever to use smaller buffers as well, post diffs on ports@ and if this hurts no MP kernel users we'll change the defaults. This would improve everybody's setup. Let me know if you need help for this. Browsers are a different story. They use a huge amount of code with complex interaction that makes simple things complicated to debug. I'd suggest to first make work mplayer, cmus & friends.
