Bug 16287 - Avoid use of pcm_route ALSA plugin for Fab4
: Avoid use of pcm_route ALSA plugin for Fab4
Status: CLOSED WORKSFORME
Product: SB Touch
Classification: Unclassified
Component: Audio
: 7.5.0
: PC Debian Linux
: P2 major (vote)
: 7.6.0
Assigned To: Spies Steven
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2010-06-07 05:54 UTC by Alan Young
Modified: 2011-05-09 09:41 UTC (History)
4 users (show)

See Also:
Category: Task


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Alan Young 2010-06-07 05:54:33 UTC
Testing had revealed that use of the ALSA pcm_route plugin adds some significant CPU overhead in the output path.  The existing "hifi" configuration is based of the examples from the ALSA documentation. Experimentation suggests that a simpler plugin chain will work just as well for our purposes.
Comment 1 SVN Bot 2010-06-07 05:56:03 UTC
 == Auto-comment from SVN commit #8844 to the jive repo by ayoung ==
 == http://svn.slimdevices.com/jive?view=revision&revision=8844 ==

bug 16287: Avoid use of pcm_route ALSA plugin for Fab4 
Change default ALSA configuration to use a setup that avoids including the pcm_route plugin in the chain.
Comment 2 Alan Young 2010-06-07 06:00:42 UTC
Assign to QA for verification testing. The main thing to check is bit-perfect output from the digital outputs.

John, would you be prepared to look at this?
Comment 3 John Swenson 2010-06-07 09:30:18 UTC
Sure, I'd be glad to try it out. I tried to look at the new file but svn was running excruciatingly slow, I'll have to try it tonight.

John S.
Comment 4 John Swenson 2010-06-09 12:31:25 UTC
I tried the new file and it does work, it plays music fine. I tried to run my usual bit perfectness tests but something strange seems to be happening. This records the S/PDIF stream with another device and then compares the files. It does a good job of aligning the files. Normally this works very well and can find single bit differences. With this output not only did it say there are differences but they are major. This is probably a problem in the test setup rather than any problem with the ALSA configuration. I'll have to do some more experimenting with this. 

One thing I can try is running the tests  with the original alsa file and with the new one. If THAT comes out the same then there is a good probability the new config is bit perfect. 

John S.
Comment 5 John Swenson 2010-06-11 13:22:28 UTC
I finally got the test methodology working correctly. The new ALSA config is definitely bit perfect. 

So far it works fine, I haven't found any issues with it.

John S.