New beagleboard demo image, now with 100% more DSP

I got the DSP to decode a h.264 move using gstreamer in the demo image but it is very slow and blocky. Any Ideas why?

bootargs: mem=88M omapfb.video_mode=1024x768MR-16@60

modprobe cmemk phys_start=0x85800000 phys_end=0x86800000
pools=20x4096,8x131072,4x829440,1x5250000,1x1429440,1x256000

modprobe dsplinkk
modprobe lpm_omap3530
export DISPLAY=:0

gst-launch -v filesrc location=/home/root/
elephantsdream-480-h264-st-aac.mov !
qtdemux name=demux.video_00 ! TIViddec2 ! xvimagesink

Also I tried updating with opkg install gstreamer-ti but it fails with a opkg fork error.

Thanks,
Kevin

I got a different video to decode, must have been something weird with the other file.
Decoding without the audio is very stable (file is on a USB stick), but as soon as you add audio the whole system grinds to a halt in a matter of seconds. Would having a revC board help this because of more ram or is this a problem with something else? I’ve tried both alsasink and osssink and they both work for a little bit then the system becomes locked up requiring a restart.

Another question I have is do I need to allocate more memory for the cmemk driver if I’m doing audio or wil the standard call work:
modprobe cmemk phys_start=0x85800000 phys_end=0x86800000
pools=20x4096,8x131072,4x829440,1x5250000,1x1429440,1x256000

Thanks for your help

Kevin

Kevin,

This memory map looks okay for playing D1_NTSC video and audio.
Elements (audio and video) allocates CMEM circular buffer to copy the
data from upstream to contiguous memory location. If your cmem pool
is not a big enough then element will report ERROR. If you do not see
any error that means you are fine with your cmem args. I think your
issue is more of related to available free memory.

RevC will definatly help but for that you need to rebuild codec-combo
and other components , this page
http://wiki.davincidsp.com/index.php?title=Changing_the_DVEVM_memory_map
talks about some details for rebuilding combo's and other tools of
dm6446 evm and pretty much the same holds true on OMAP3. Note that
the default combo's are hard-coded for mem=88M option. Look at
loadmodule.sh script in my previous post. This will give you default
memory map.

Checkout this video to get familiar with TI plugins [1]
Also refer [2] and [3] for more info on element properties and some
sample pipeline. If you have any technical questions then i would
suggest posting on gstreamer.ti.com forums.

[1] https://gstreamer.ti.com/gf/project/gstreamer_ti/wiki/?pagename=OverviewPresentation

[2] http://wiki.davincidsp.com/index.php?title=GStreamer

[3] http://wiki.davincidsp.com/index.php?title=GstTIPlugin_Elements

Thanks
Brijesh

Hi all,
I’ve followed these step:

  1. set boot args mem=88M video=omapfb:vram:10M omapfb.video_mode=1024x768MR-16@60

  2. modprobe cmemk phys_start=0x85800000 phys_end=0x86800000 pools=20x4096,8x131072,4x829440,1x5250000,1x1429440,1x256000

  3. modprobe dsplinkk

  4. modprobe lpm_omap3530

  5. export DISPLAY=:0

  6. gst-launch-0.10 filesrc location="/media/mmcblk0p3/movie.avi" ! avidemux ! omx_mpeg4dec ! omapfbsink

Then I got these message:

Setting pipeline to PAUSED …
ERROR: Pipeline doesn’t want to pause.
Setting pipeline to NULL …
FREEING pipeline …

when I use --gst-debug=3 --gst-debug-no-color, I see that :

Setting pipeline to PAUSED …
0:00:00.334045424 1798 0x15050 INFO GST_STATES gstelement.c:2434:gst_element_change_state: have FAILURE change_state return
0:00:00.334716811 1798 0x15050 INFO GST_STATES gstelement.c:2044:gst_element_abort_state: aborting state from NULL to READY
0:00:00.335449233 1798 0x15050 INFO GST_STATES gstbin.c:2369:gst_bin_change_state_func: child ‘omapfbsink0’ failed to go to state 2(READY)
0:00:00.336059584 1798 0x15050 INFO GST_STATES gstelement.c:2434:gst_element_change_state: have FAILURE change_state return
0:00:00.336700453 1798 0x15050 INFO GST_STATES gstelement.c:2044:gst_element_abort_state: aborting state from NULL to PAUSED
ERROR: Pipeline doesn’t want to pause.

Any help please?
Regards,
Giang Nguyen

ps: I attached the debug file.

debug.txt (13.2 KB)