lol sorry, first time with xbmc set to german, fortgeschritten...experte... you know the drill
EDIT: still too stupid to get it. is it the multi-processor-decoding option?
Read the first post ... you are two days too old.
Hi, I have some problems with audio passthrough.
I connected my linux box to my TV via hdmi. The TV is connected to the audio system via an optical cable.
The ac3 passthrough works well, but dts doesn't.
If I play the same file with VLC dts passthrough works fine, by enabling as output device "HDA ATI HDMI, HDMI 0 HDMI AUDIO OUTPUT".
I followed all the steps described in the first post, plus I installed a basic xfce system, without pulse audio.
Please provide a debug log of xbmc and vlc. Most tvs are too dumb to forward anything despite ac3 correctly.
(2014-01-30, 20:45)unimatrix27 Wrote: [ -> ]until now I had never used anything else but ppa:wsnipex/mesa and I had the bug reported in this youtube video from the start (which was yesterday). exactly same behavior.
I have now tried the test PPA and only see a black screen (to be specific: no HDMI output at all shortly after booting but XBMC runs).
Maybe I did something wrong with the packages, no time tonight to debug...but I conform that the non-testing ppa had the problem - at least for me.
Running Radeon HD 8330 on Zotac Zbox AQ01
I just bought the same box. I do have problem with the BIOS HDMI initialization for my TV. Managed to install using a PC monitor with and hdmi/dvi cable and later connected to the TV back again. The BIOS HDMI initialization still fails, but Linux works after that (as soon as framebuffer driver is loaded). I have already contacted Zotac, but the box is too new: then have not enough problem records for the BIOS. Don't hesitate to report the problem directly including TV model (support page at zotacusa)
For those curious about what works: everything except XBMC vdpau accelerated decoding. But the board is capable to decode blueray material without acceleration fortunately.
First the software setup:
Debian unstable+experimental
Linux 3.13.2
mesa 10.1-rc1
radeon driver 7.3.0-1
glamor: 0.6.0
ffmpeg 2.1.3 (patched to expose a symbol that xbmc uses)
xbmc fernetmenta git master + amd branch last patch
I had no acceleration at the beginning because the radeon driver was compiled disabling glamor that is yet not packaged in Debian. So I had to find a glamor compatible package and rebuild the radeon driver with glamor enabled.
Now vdpauinfo reports what is expected, qvdpautest shows nice figures (still a bit low for H264) but playing with video vdpau acceleration just fails as if the same video sequence was played several time and then a new sequence starts and fails the same way. I have several
CDVDPlayerAudio:: Discontinuity2 - was:70259126.282469, should be:70248911.012554, error:-10215.269915 like message
CDVDPlayerVideo::CalcDropRequirement - hurry: 1.
Will post a full xbmc .log later on.
@EricV: My Kabini works complete following that howto. including hd decoding. qvdpautest I posted lately, was done with that box. Use OpenELEC, it has everything to make it work. The rest is configuration / software issue.
@dongioa:
Quote:main debug: using audio filter module "dtstospdif"
main debug: TIMER module_need() : 0.633 ms - Total 0.633 ms / 1 intvls (Avg 0.633 ms)
main debug: conversion pipeline completed
main debug: looking for audio mixer module: 2 candidates
main debug: no audio mixer module matching "any" could be loaded
main debug: TIMER module_need() : 0.112 ms - Total 0.112 ms / 1 intvls (Avg 0.112 ms)
main debug: input 'dts ' 48000 Hz 3F2R/LFE frame=512 samples/2048 bytes
main debug: filter(s) 'dts '->'dts ' 48000 Hz->48000 Hz 3F2R/LFE->3F2R/LFE
I don't know what this filtere does.
Edit: You don't hear a thing or only noise? It seems that the frames we send, the TV does not like - that's a general problem and not specific to this version, sadly. Try to open a bug on trac.xbmc.org and add Anssi to CC
(2014-02-09, 12:21)fritsch Wrote: [ -> ]@EricV: My Kabini works complete following that howto. including hd decoding. qvdpautest I posted lately, was done with that box. Use OpenELEC, it has everything to make it work. The rest is configuration / software issue.
@
fritsch
Thanks for providing tips. In your above sentence, that box = zotac nano AQ01 or another kabini box? OpenElec is sadly not really an option for me as I use the box for other purpose than solely htpc. Any specific patch I may apply in the meantime (or XBMC git tree I can test).
Exactly that box, yes.
Only things you need (my AMD branch, which is fernets + one patch) - take care there is a bug when stopping videos. Mesa 10.1 + deinterlacing patch. Make sure here, that there are the last two patches done by ckoenig and mareko included concerning flushing. You also need a kernel patch to get more performance (see OpenELEC github in packages/linux/patches/ folder.
I don't know if the "performance fix on the mesa side - the one with the msg" was included in 10.1 wsnipex packages that one separately in our Mesa 10.0.3
That's it - have fun.
(2014-02-09, 12:49)fritsch Wrote: [ -> ]Exactly that box, yes.
Only things you need (my AMD branch, which is fernets + one patch) - take care there is a bug when stopping videos. Mesa 10.1 + deinterlacing patch. Make sure here, that there are the last two patches done by ckoenig and mareko included concerning flushing. You also need a kernel patch to get more performance (see OpenELEC github in packages/linux/patches/ folder.
I don't know if the "performance fix on the mesa side - the one with the msg" was included in 10.1 wsnipex packages that one separately in our Mesa 10.0.3
That's it - have fun.
Is your BIOS handling your TV/monitor correctly? Did you test with several TV set? I hope I do no have a broken board. When I See the figure I have with qvdpautest I bet the mesa performance fix is in (I have also 110 fps...). I did check: the patch is in. I also have the last patch in your XBMC amd branch. Remains mesa and kernel as possible culprit
Will try to figure out where other patches are located and make my shopping. Thanks again for the support.
Had it running on my 40 Zoll Samsung and on my 1080i60 screen - worked fine on both via HDMI.
Edit: in bios of course.
(2014-02-09, 15:15)fritsch Wrote: [ -> ]Had it running on my 40 Zoll Samsung and on my 1080i60 screen - worked fine on both via HDMI.
Edit: in bios of course.
Did you try any TV set at 50Mhz? The screen that works is at 60Mhz. The TV that doesn't is initialized at 1080p50. Same for Linux after BIOS phase except it works there.
On the XBMC side, I did my shopping and applied the two last patches of the deathsimple xbmc branch referred in bug
bug 73191. It now works whith vdpau but
only if I disable vdpau mixer, otherwise I do have exactly the symptoms described in this bug.
Do you have another thing needed for vdpau mixer or is it still unsolved as the bug discussion makes it appears? (and yes I have the amd patch for unmapping the surfaces)
You mean hz I hope? It works in 1080p50 yes - it's my primary use case.
You are testing LiveTV right? <- I also have that stuttering at the begging of LiveTV streams, local files, that don't need to cache, work fine.
(2014-02-09, 16:15)fritsch Wrote: [ -> ]You mean hz I hope? It works in 1080p50 yes - it's my primary use case.
You are testing LiveTV right? <- I also have that stuttering at the begging of LiveTV streams, local files, that don't need to cache, work fine.
Mhz -> Hz of course. For sure 50 Hz is the default in most Europe but that not obviously what the TV declares as it's preferred mode which is usually used by BIOS. I have two samsung TV's at work that BIOS drives at 60 and that need to be forced to 50 for XBMC TV playing (I force it in X11 start to avoid the black screen). I must confess I do not know if these TV set were originally for the French market.
Concerning the video material that expose problem, its not live HDTV 1080i50 material as I haven't yet switched the tuners to the new box. Its 1080p24 blueray content. I have no problem at beginning of playback. I have problem when I turn vdpau mixer on exactly as the PR describes and some other users confirm it. The only difference is the version of mesa 10.1rc1 that apparently did introduce a regression if I read the various PR comments correctly.