Difference between revisions of "Talk:Romit2 VPU"

From MyLabWiki
Jump to: navigation, search
(Created page with " == Logbook == === 2011.12.05 === Re-assembled the Gumstix VPU using the old parts (including used cable). Capture.elf worked right away, but gstreamer still gives the v4l2src...")
 
Line 1: Line 1:
 
 
== Logbook ==
 
== Logbook ==
  
Line 16: Line 15:
 
Retried the startup procedure I discovered yesterday, namely to start the GST pipeline with GST_DEBUG=v4l2src:2 – it still woks! It didn't work before enabling the debug. There must be some weird timing issue and the reason why it works must be that the pipeline is slower to start with debug enabled.
 
Retried the startup procedure I discovered yesterday, namely to start the GST pipeline with GST_DEBUG=v4l2src:2 – it still woks! It didn't work before enabling the debug. There must be some weird timing issue and the reason why it works must be that the pipeline is slower to start with debug enabled.
  
I have let the Gumstix stream for 5 hours using default encoder parameters:
+
I have let the Gumstix stream for 6 hours using default encoder parameters:
 
* CPU load steady at 22-24 %
 
* CPU load steady at 22-24 %
 
* Memory consumption steady at 66 MB
 
* Memory consumption steady at 66 MB

Revision as of 23:01, 6 December 2011

Logbook

2011.12.05

Re-assembled the Gumstix VPU using the old parts (including used cable). Capture.elf worked right away, but gstreamer still gives the v4l2src error. Then I exported the GST_DEBUG=v4l2src:4 ran the pipeline again and noticed that it continuously printed DEBUG and INFO messages. Killed it, set the debug level to 2 and executed the pipeline again – It was streaming!!!

While it streaming I secured the Overo and camera using 2 screws so that was not the problem

Note that when capture.elf exits, gstreamer stops streaming. I could, however, restart it every time.


2011.12.06

Retried the startup procedure I discovered yesterday, namely to start the GST pipeline with GST_DEBUG=v4l2src:2 – it still woks! It didn't work before enabling the debug. There must be some weird timing issue and the reason why it works must be that the pipeline is slower to start with debug enabled.

I have let the Gumstix stream for 6 hours using default encoder parameters:

  • CPU load steady at 22-24 %
  • Memory consumption steady at 66 MB