Re: [PATCH] ASoC: rsnd: adg :: AUDIO-CLKOUTn can synchronizes with L/R clock.

[Mark Brown <broonie@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 1367 bytes --]

On Wed, Jun 08, 2016 at 12:27:48AM +0000, Kuninori Morimoto wrote:

> > Why would a user not want these clocks to be synchronous?  A lot of
> > CODECs will at least have better performance if their master clock is
> > synchronous to the audio clocks so it'd be a better default, is there an
> > advantage to not doing it?

> I'm now confusing. We can set system clock on audio card, for example
> simple-card case, it is called as "system-clock-frequency".
> In my understanding, this "system clock" and above "master clock" are same clock.
> but "system clock" is fixed rate (= not related to audio clock in many cases).
> Because of this, some codec doesn't request synchronous between
> master clock <-> audio clock, but, some codec requests synchronous it.
> Am I wrong ??

A lot of CODECs can to varying degrees tolerate this but it will tend to
show up in performance numbers, it's often not immediately obvious just
from a listening test.  Usually those that don't mind have a FLL or PLL
they're using to generate the actual audio master clock at a useful rate
for dividing down, or sometimes fancy digital logic to match things
up.

It's not a problem to have this configurable, I'm just thinking it might
be better to have the default be the other way around so that we default
to synchronous but can turn that off if it's required.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 473 bytes --]