2013/7/12 Alexander Motin <mav_at_freebsd.org>: > On 12.07.2013 23:54, Ulrich Spörlein wrote: >> >> I'm trying to setup XBMC on a -CURRENT box with an IvyBridge CPU and >> GPU. While testing playback via mplayer on a LG TV over HDMI, I noticed >> that sound is lagging video by about 100-200ms or so. When I switch to >> using the jack outputs powered by some Realtek chip, audio is perfectly >> fine. >> >> Is HDMI lag a known problem? Can this be fixed? >> >> root_at_coyote:~# dmesg | egrep vgapci\|pcm >> vgapci0: <VGA-compatible display> port 0x3000-0x303f mem >> 0xe0000000-0xe03fffff,0xc0000000-0xdfffffff irq 16 at device 2.0 on pci0 >> agp0: <IvyBridge mobile GT2 IG> on vgapci0 >> pcm0: <Realtek ALC888 (Rear Analog)> at nid 20 and 24 on hdaa0 >> pcm1: <Realtek ALC888 (Rear Digital)> at nid 30 and 31 on hdaa0 >> pcm2: <Intel Panther Point (HDMI/DP 8ch)> at nid 6 on hdaa1 >> pcm3: <Intel Panther Point (HDMI/DP 8ch)> at nid 7 on hdaa1 >> drmn0: <Intel IvyBridge (M)> on vgapci0 > > > I don't know what to say. I am now using HDMI audio from NVIDIA card to > quite old external 5.1 receiver with XBMC every day, and I haven't noticed > lags. Before that I've also successfully used SPDIF connection for the long > time. Though I've never specially tested it somehow other then watching > movies. :) If you have some good testing methodology -- please, welcome to > share. By the HDA driver HDMI is handled exactly the same way as analog > output from the point of data buffering, so I would not expect there major > differences. > > You may try to experiment with hw.snd.latency sysctl to tune buffering in > kernel to see whether it affect the result. > > Also you may compare delays when doing AC3/DTS pass-through with case of > software decoding and discrete (multichannel) PCM playback. > > The only potentially related effect I have noticed is that my receiver eats > first second or about that of playback stream. It makes short sounds like > GUI event notifications inaudible sometimes. I guess that could be made to > restore audio sync after some unavoidable startup delay, but that is only my > guess. Hmm, weird. I've rebooted the machine (it was up more than 30d before I ever started up Xorg or any sound-using application), and maybe that fixed some contigmalloc issues or something, because I can no longer reproduce the problem. I've had problems like this on an old laptop, where if I didn't start X11 within 24h of rebooting, it wouldn't start up. This system is using ZFS of course, so unless you take memory early on, ZFS will have eaten it :) Anyway, thanks for your suggestions and I can confirm everything's fine. After fixing the /dev/dri/card0 permission problem, even XBMC works like a charm on this system. Cheers, UliReceived on Mon Jul 15 2013 - 06:22:35 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:39 UTC