[RFC] add mux/demux documentation

[juimonen]

Signed-off-by: Jaska Uimonen jaska.uimonen@intel.com

[lgirdwood]

I think this will do in the meantime until the enum kcontrols are done. Do you have an ETA for enums yet ?

[juimonen]

@lgirdwood updated.

• removed emacs backup file
• removed white space
• fixed a typo in header

This is now depending on:

• [RFC] Demux blob sof#2594
• muxdemux: Optimize and cleanup mux/demux internal and external interface. sof#2600.

I also talk about python blob generator: thesofproject/sof#2536, which people seem not to like, but it is the only "generic" tool we would have to generate blobs... But I can remove this if no traction. So the python tool could be used to generate dc blocker, mux/demux, or even simple eq blobs.

And now when I'm on it: we should really do the optimization for demux @Keyon was talking about in thesofproject/sof#2533.

Then the demux would be in pretty good shape, so this here looks like mux todo list :)

[juimonen]

@ranj063 thanks for comments! updated now, hopefully addressed all comments...

[libinyang]

@juimonen I'm a newer to mux/demux. And I find it's a little hard for me to understand the picture. Why do we need set column 3, 5, 7 to 1? Shouldn't it be 0s although it doesn't take efforts here. Or we must set 1s in column 3, 5, 7?

[juimonen]

@libinyang just think as the row as input channel, and 0/1 as if this row (channel) is put to column output channel. so for example first row (0,1,0,1,0,1,0,1) would mean that copy input channel 1 to output channels 2, 4, 6, 8. And so on for other channels. There's always 1 matrix per stream-to-stream mapping.

[keyonjie]

@lgirdwood updated.

• removed emacs backup file
• removed white space
• fixed a typo in header

This is now depending on:

• thesofproject/sof#2594
• thesofproject/sof#2600.

I also talk about python blob generator: thesofproject/sof#2536, which people seem not to like, but it is the only "generic" tool we would have to generate blobs... But I can remove this if no traction. So the python tool could be used to generate dc blocker, mux/demux, or even simple eq blobs.

And now when I'm on it: we should really do the optimization for demux @Keyon was talking about in thesofproject/sof#2533.

Then the demux would be in pretty good shape, so this here looks like mux todo list :)

Hi @juimonen, thanks for the documentation! Only one suggestion that to remove the complicate matrix part if we are aligned to remove the "channel mixing" feature in the future. The good point is that nobody is using that mixing feature until now, so don't mention that could be wise choice, to make easier life not only for people reading this documentation, but also for people who might need to maintain/update the documentation later.

[lgirdwood]

@deb-intel technical approved.

[deb-intel]

@lgirdwood Thanks!

[deb-intel]

Nice job. I will make grammatical edits in a separate PR.

[marc-hb]

This introduced a number of warnings, at least these:

sof-docs/algos/demux/demux.rst: WARNING: document isn't included in any toctree
sof-docs/algos/demux/demux.rst:62: WARNING: Pygments lexer name 'm4' is not known
sof-docs/algos/demux/demux.rst:119: WARNING: Pygments lexer name 'm4' is not known