Allow better AudioRendererMixer reuse by ignoring irrelevant params.

Allow better AudioRendererMixer reuse by ignoring irrelevant params.

Several AudioParameters fields are irrelevant when converting from
one type to another via AudioRendererMixer.  Not ignoring these
values results in expensive extra renderer side resampling load and
extra physical stream creation. Physical stream creation has a long
tail of additional overhead (multiple threads, shared memory, etc).

AudioRendererMixerManager will now ignore these irrelevant params
when determining whether it should reuse an existing mixer.

BUG=none
TEST=new unittest.

Committed: https://crrev.com/cee23788475df08e55797adfb4222192cff882ff
Cr-Commit-Position: refs/heads/master@{#296304}

[DaleCurtis, 2014-09-23 20:17:11]

dalecurtis@chromium.org changed reviewers:
+ xhwang@chromium.org

[DaleCurtis, 2014-09-23 20:17:12]

xhwang: You're looking into the many complexities of our rendering pipeline, so
here's a CL which tweaks some of audio output portion of that pipeline :)

Here's a rough sketch of how this connects to WebMediaPlayer and ultimately the
AudioRenderer:

RenderFrameImpl::createMediaPlayer() requests an AudioRendererSink from
AudioRendererMixerManager which creates an AudioRendererMixerInput which knows
how to get an AudioRendererMixer instance from the ARMM. This ARMI is passed
through WebMediaPlayer all the way to AudioRendererImpl.  If more media players
are created which share the same parameters they'll all use the same ARM which
mixes them all out into a single physical stream.

Let me know if you want to know more, but that's the gist of it.

[xhwang, 2014-09-23 21:14:04]

lgtm % nits

https://codereview.chromium.org/596073002/diff/1/content/renderer/media/audio...
File content/renderer/media/audio_renderer_mixer_manager.h (right):

https://codereview.chromium.org/596073002/diff/1/content/renderer/media/audio...
content/renderer/media/audio_renderer_mixer_manager.h:72: typedef std::pair<int,
media::AudioParameters> MixerKey;
nit: comment that the "int" is for render view ID.

https://codereview.chromium.org/596073002/diff/1/content/renderer/media/audio...
content/renderer/media/audio_renderer_mixer_manager.h:75: // mixers where
irrelevant keys mismatch (i.e., effects, bits per sample).
s/where/where only/

[DaleCurtis, 2014-09-23 21:17:45]

Thanks for review!

https://codereview.chromium.org/596073002/diff/1/content/renderer/media/audio...
File content/renderer/media/audio_renderer_mixer_manager.h (right):

https://codereview.chromium.org/596073002/diff/1/content/renderer/media/audio...
content/renderer/media/audio_renderer_mixer_manager.h:72: typedef std::pair<int,
media::AudioParameters> MixerKey;
On 2014/09/23 21:14:04, xhwang wrote:
> nit: comment that the "int" is for render view ID.

Done.

https://codereview.chromium.org/596073002/diff/1/content/renderer/media/audio...
content/renderer/media/audio_renderer_mixer_manager.h:75: // mixers where
irrelevant keys mismatch (i.e., effects, bits per sample).
On 2014/09/23 21:14:04, xhwang wrote:
> s/where/where only/

Done.

[DaleCurtis, 2014-09-23 21:18:27]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2014-09-23 21:22:01]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/596073002/20001

[commit-bot: I haz the power, 2014-09-23 23:26:12]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-09-23 23:26:14]

Exceeded time limit waiting for builds to trigger.

[DaleCurtis, 2014-09-23 23:28:00]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2014-09-23 23:29:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/596073002/20001

[commit-bot: I haz the power, 2014-09-23 23:46:27]

Committed patchset #2 (id:20001) as f41cc4addc3349d0858841bc8bb8472a7802f732

[commit-bot: I haz the power, 2014-09-23 23:47:05]

Patchset 2 (id:??) landed as
https://crrev.com/cee23788475df08e55797adfb4222192cff882ff
Cr-Commit-Position: refs/heads/master@{#296304}