Beta Panplayer Beta


notaz said:
TI only provides 'raw' software, it's still up to the integrator (which is unfortunate enough to be me) to set things up correctly. And if something goes wrong, something won't work, we are on our own because this is too small project (not enough parts sold) for TI to provide support.
Sounds unfair. :( I guess, Big Companies even get full access to Driver sourcecode so they can make things working easily. I once thought TI likes the Pandora Project, so a little bit more support from them shouldn't hurt anyone.
 
Last edited by a moderator:
This is brilliant thankyou. Prior players couldn´t run downloaded youtube flv files or more importantly, seemed to struggle with standard DVD vob files. This plays them nicely with no need to re-encode my collection. Kept the kids nicely amused with disney films on my holiday car journeys. Priceless!
 
Thanks for all the work on this, it looks very promising so far.

I'm encountering AV sync issues, the same issue that I had with the mplayer from the community codec pack. The video lags behind the audio and seems to get worse and worse over time.

The video is xvid, encoded with mencoder svn-r30200. It plays fine with VLC on the pandora, and mplayer on my laptop. Let me know if you would like me to create a sample.

I did not notice AV sync issues with h264 encoded videos that I tried.


Second issue I noticed was with picklelauncher. For some reason it did not display any subdirectories in my movie directory. I also noticed that picklelauncher was using about 47% CPU just sitting there. My movie directory was mounted over sshfs if that makes any difference.
 
pder said:
Second issue I noticed was with picklelauncher. For some reason it did not display any subdirectories in my movie directory. I also noticed that picklelauncher was using about 47% CPU just sitting there. My movie directory was mounted over sshfs if that makes any difference.

directories should show up, im going to guess the sshfs had something to do with it.
ive confirmed the cpu usage and I will look into it to reduce it.

Update: Ive improved the frame handling. I added a proper 60 fps frame limiter with frame skip. Whats different from normal is that the screen will not be updated if nothing has changed First tests show with nothing changing my laptop cpu usage is around 10 %. User activity or screen movement will do around 40%

Update 2: Did one better and disabled any blits if no activity is detected, now i measure 1% cpu :)
 
Last edited by a moderator:
It seems fast... but when I play my m4v movie the video is too fast... but the audio is fine. Any ideas?
 
I also have some troubles playing .mkv files i think the files might be o big , but when I play them there is no video just audio . Can't we just have it where it automaticly sizes the video to fit screen no matter the size?
 
I use this player on a daily basis, and it is for me one of the greatest additions to the Pandora (I really wanted to use the Pandora as a video player). I was wondering if it was using SDL at all? If so, don't you think it might benefit from switching to Notaz's SDL version? It may help with tearing for example?
 
panplayer is a plain (rather recent) mplayer compiled with some Pandora specific flags. The "pan" part is just that pickelauncher is used for starting mplayer with the correct parameters.
 
notaz said:
torpor said:
i) Regardless of this issue, isn't it good to move to the new kernel?
Sure but the new kernel has some missing stuff and several regressions, and there is no one wanting to work on these. Properly maintaining a kernel is almost a full time job really, and it's a bit too much for me with unrelated day job. Yeah maybe I could somehow manage it (like I did with current .27 kernel over the years) but would have to kill all my 'fun' projects for this. Still I'll be trying to get something for upcoming major firmware update (Y*).

fusion_power said:
notaz said:
I'm tired repeating this, but doing it once more: the real problem is not the old kernel but the DSP framework requirements to reserve RAM (so that Linux can't use it) and no one capable doing such work wanting to get the damn thing working.
Sorry, I just don't understand much of this technical stuff, so I had to ask. ^^"
And actualy it should be the Job of TI/SGX and/or Angstrom devs to implement such basic stuff, not yours. ;) Nobody should release a Chipset without full implemented and working Feature Support in Software. At least they have to give others the Source Code if they are not able to do it right themselves.
This is the point that makes me sad and not your very valuable work onto fixing all the stuff, others made wrong. ^^
TI only provides 'raw' software, it's still up to the integrator (which is unfortunate enough to be me) to set things up correctly. And if something goes wrong, something won't work, we are on our own because this is too small project (not enough parts sold) for TI to provide support.

I'll see what I can dig up on this. I did this long time ago, it required telling Linux that it had max physical memory minus memory used for dsplink for total memory. Then the DSP code on the host and DSP used physical memory addresses to access the memory carved out. We used a custom software package on top of dsplink on both the host and DSP instead of the codec engine that TI uses.
 
Last edited by a moderator:
Back
Top