No multithreaded Video decoding on XBMC INTREPID version (in-depth testing)

xbmc.bin is 0

I found this interesting thread at AVS forum. Where someone succesfully having "killa sample" playing with 70% core utilization with Fedora Core 10 and the new 2.6.27-9 kernel. Maybe there is light...

Unfortunatelly i am totally not familiar with Fedora and rpm package management. Can someone try it out if it's really working there ?

This case i think we can really narrow down the problem.

Link: http://www.avsforum.com/avs-vb/showthrea...st15378996

Fedora 10, with updates, and Nvidia 177.82, and SVN XBMC has exactly the same problem... uname -r reports...

2.6.27.9-159.fc10.i686

Edit: The guy in the thread you mentioned is running an Intel CPU, im still wondering if this is an AMD only thing.

Edit2: Found a workaround. Line 152 in xbmc/cores/dvdplayer/DVDCodecs/Video/DVDVideoCodecFFmpeg.cpp reads like this...


Change it to this...


and you should find it using much more CPU, also I found a fairly significant reduction in dropped frames.

No i am having same issues with intel core 2 duo e8400.

I am really clueless again.

motd2k Wrote:

PHP Code:

int num_threads = std::min(8 /*MAX_THREADS*/, g_cpuInfo.getCPUCount()+2); 


and you should find it using much more CPU, also I found a fairly significant reduction in dropped frames.

Thanks, i will test it tonight.

I've just read on Phoronix, that FFMPEG gets mainline VDPAU support. If this could maybe implemented into XBMC than we won't have to worry about multithreaded decoding at all

http://www.phoronix.com/scan.php?page=ne...&px=Njk3MQ

And a simple search on these forums would have found you a thread dedicated to a discussion on vpdau.

Just search for : vpdau.

Magic

If you want more specifics about the technical difficulties, here you go:

http://forum.xbmc.org/showpost.php?p=264...tcount=168

motd2k Wrote:Edit2: Found a workaround. Line 152 in xbmc/cores/dvdplayer/DVDCodecs/Video/DVDVideoCodecFFmpeg.cpp reads like this...

PHP Code:

int num_threads = std::min(8 /*MAX_THREADS*/, g_cpuInfo.getCPUCount()); 


Change it to this...

PHP Code:

int num_threads = std::min8 /*MAX_THREADS*/, g_cpuInfo.getCPUCount()+2); 


and you should find it using much more CPU, also I found a fairly significant reduction in dropped frames.

Did you just do this to the file in your ~/XBMC directory or did you recompile after that?

Of course you need to recompile. That is XBMC source code.

I'm already using a build compiled from source. I didn't use synaptic to install. Can I jus make + make install on top of it or do I have to remove the old install?