2012-02-19, 15:11
Hi there!
First many thanks to FernetMenta for all his work on VAAPI and XvBA.
I successfuly compile XBMC pre11 with VAAPI on OpenSUSE Tumbleweed with LXDE since autumn 2011 on my ASUS E-350 htpc and now tried the XvBA branch. Actually I notice no difference in HD playback. Both ran smooth, except small random jitters that I notice at slow camera moves or the end credits where the letters make tiny jumps every 5 to 20 seconds.
The only difference with XvBA is, that I can activate the option "adjust display refresh rate to match video". This delays the start of the video as expected, but the "jumps" are now much smaller and thus harder to recognize. Yet they are not gone. If I activate this option on vaapi-built then HD-playback-speed is like waves of fast and slow decoding...very bad.
So why do I read here that XvBA playback has better performance - the small jumps are still there, even with the above option. so maybe thats not a performance issue? and maybe my performance is the same for VAAPI because I compile myself?? How do you exactly recognize the difference in performance? Often, 24p is mentioned, but I never had problems with that either. Or is it only for exact 24p, because most vids say 23.976 fps.
Anyone an idea?
Here is my compile string; notice I have to disable pulse for XvBA only, otherwise no digital sound available:
First many thanks to FernetMenta for all his work on VAAPI and XvBA.
I successfuly compile XBMC pre11 with VAAPI on OpenSUSE Tumbleweed with LXDE since autumn 2011 on my ASUS E-350 htpc and now tried the XvBA branch. Actually I notice no difference in HD playback. Both ran smooth, except small random jitters that I notice at slow camera moves or the end credits where the letters make tiny jumps every 5 to 20 seconds.
The only difference with XvBA is, that I can activate the option "adjust display refresh rate to match video". This delays the start of the video as expected, but the "jumps" are now much smaller and thus harder to recognize. Yet they are not gone. If I activate this option on vaapi-built then HD-playback-speed is like waves of fast and slow decoding...very bad.
So why do I read here that XvBA playback has better performance - the small jumps are still there, even with the above option. so maybe thats not a performance issue? and maybe my performance is the same for VAAPI because I compile myself?? How do you exactly recognize the difference in performance? Often, 24p is mentioned, but I never had problems with that either. Or is it only for exact 24p, because most vids say 23.976 fps.
Anyone an idea?
Here is my compile string; notice I have to disable pulse for XvBA only, otherwise no digital sound available:
Code:
git clone https://github.com/FernetMenta/xbmc.git && cd xbmc && git checkout xvba && ./bootstrap && ./configure --enable-xvba --disable-vaapi --prefix=/usr --enable-libbluray --disable-pulse --enable-mid && make -j2 && make install