Use larger buffer sizes for lower power on Linux.

Use larger buffer sizes for lower power on Linux.

HTML5 playback on Linux ALSA, Pulse, and ChromeOS will now use a
buffer size of 1024+ for media playback.

On my Z620 this takes pulseaudio daemon CPU usage from a solid 3-4%
down to 1%.  Likely there are savings in the Chrome process as well,
but those are harder to measure against process noise.

I suspect we'll see greater savings on Chromebooks.

BUG=362261
TEST=Audio sounds the same always.  CPU usage down.
NOTRY=true

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=264443

[dgreid, 2014-04-11 21:17:34]

lgtm,

The only issue from chromeos might be when a low-latency stream is started while
a high-latency stream is playing.  The high latency strem will be interrupted as
the system is re-configured to do faster wake ups.

If this is a problem we can fix it on the platform side similar to how
pulseaudio handles it by rewinding samples back out of the hardware buffer and
re-mixing.

[DaleCurtis, 2014-04-11 21:41:13]

Given the existing code, Flash would have already had that problem, so I guess
the answer to your question depends on whether you've seen issues in the field
around that :)

[dgreid, 2014-04-11 21:43:13]

On 2014/04/11 21:41:13, DaleCurtis wrote:
> Given the existing code, Flash would have already had that problem, so I guess
> the answer to your question depends on whether you've seen issues in the field
> around that :)

No complaints about that so I think we're good to go.

[henrika (OOO until Aug 14), 2014-04-14 08:03:23]

Drive-by comment.

https://codereview.chromium.org/235723003/diff/1/media/base/audio_hardware_co...
File media/base/audio_hardware_config.cc (right):

https://codereview.chromium.org/235723003/diff/1/media/base/audio_hardware_co...
media/base/audio_hardware_config.cc:115: // values are unused.  OSX may have a
value as low as 128.  Windows is device
Should comments about OSX and Windows be within defined(OS_LINUX)?

[no longer working on chromium, 2014-04-14 08:51:50]

https://codereview.chromium.org/235723003/diff/1/media/base/audio_hardware_co...
File media/base/audio_hardware_config.cc (right):

https://codereview.chromium.org/235723003/diff/1/media/base/audio_hardware_co...
media/base/audio_hardware_config.cc:104: // to 20 ms worth of samples.  For a
given sample rate, this works out to:
curiously, 20ms does not sound "high" latency, why do you want to use that? 
why not simply a large hard coded buffer size like 2048?

https://codereview.chromium.org/235723003/diff/1/media/filters/audio_renderer...
File media/filters/audio_renderer_impl.cc (right):

https://codereview.chromium.org/235723003/diff/1/media/filters/audio_renderer...
media/filters/audio_renderer_impl.cc:281: const AudioParameters& hw_params =
hardware_config_->GetOutputConfig();
why not add a GetHighLatencyConfigue() method to hardware config?

[DaleCurtis, 2014-04-14 18:08:18]

https://codereview.chromium.org/235723003/diff/1/media/base/audio_hardware_co...
File media/base/audio_hardware_config.cc (right):

https://codereview.chromium.org/235723003/diff/1/media/base/audio_hardware_co...
media/base/audio_hardware_config.cc:104: // to 20 ms worth of samples.  For a
given sample rate, this works out to:
On 2014/04/14 08:51:50, xians1 wrote:
> curiously, 20ms does not sound "high" latency, why do you want to use that? 
> why not simply a large hard coded buffer size like 2048?

As discussed in the meeting this morning, I choose 20ms because for a couple
reasons:
- It results in the same buffer size as Flash for 44.1kHz and 48kHz (our most
common hw sample rates).
- It's not so long that pages still using <audio> to play effects (instead of
WebAudio) will suffer during looping or when listening for the ended event.

https://codereview.chromium.org/235723003/diff/1/media/base/audio_hardware_co...
media/base/audio_hardware_config.cc:115: // values are unused.  OSX may have a
value as low as 128.  Windows is device
On 2014/04/14 08:03:24, henrika wrote:
> Should comments about OSX and Windows be within defined(OS_LINUX)?

I hope eventually this section will have all OS, seems unnecessary to update the
comment every time :)

[DaleCurtis, 2014-04-16 00:39:46]

xians, henrika: Please review.

https://codereview.chromium.org/235723003/diff/1/media/filters/audio_renderer...
File media/filters/audio_renderer_impl.cc (right):

https://codereview.chromium.org/235723003/diff/1/media/filters/audio_renderer...
media/filters/audio_renderer_impl.cc:281: const AudioParameters& hw_params =
hardware_config_->GetOutputConfig();
On 2014/04/14 08:51:50, xians1 wrote:
> why not add a GetHighLatencyConfigue() method to hardware config?

I wanted it to be clear that this is only adjusting the buffer size and no other
parameters.

[no longer working on chromium, 2014-04-16 08:26:08]

lgtm

[henrika (OOO until Aug 14), 2014-04-16 08:40:15]

LGTM. Nice work Dale.

[DaleCurtis, 2014-04-16 18:18:22]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2014-04-16 18:19:02]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dalecurtis@chromium.org/235723003/20001

[commit-bot: I haz the power, 2014-04-16 18:26:08]

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

[commit-bot: I haz the power, 2014-04-16 18:26:08]

Try jobs failed on following builders:
  tryserver.chromium on linux_chromium_clang_dbg

[DaleCurtis, 2014-04-16 22:19:22]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2014-04-16 22:19:58]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dalecurtis@chromium.org/235723003/40001

[commit-bot: I haz the power, 2014-04-16 23:30:18]

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

[commit-bot: I haz the power, 2014-04-16 23:30:18]

Try jobs failed on following builders:
  tryserver.chromium on linux_chromium_chromeos_rel
  tryserver.chromium on linux_chromium_rel

[DaleCurtis, 2014-04-17 01:06:51]

Hmm, pepper was requesting crazy 32k buffer sizes, so I've added a maxmium
buffer size check and set it to 8k for Pulse and CRAS. Presumably this still
lgty, otherwise let me know.

[DaleCurtis, 2014-04-17 01:06:58]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2014-04-17 01:07:17]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dalecurtis@chromium.org/235723003/60001

[commit-bot: I haz the power, 2014-04-17 02:41:06]

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

[commit-bot: I haz the power, 2014-04-17 02:41:07]

Try jobs failed on following builders:
  tryserver.chromium on win_chromium_rel

[DaleCurtis, 2014-04-17 06:03:40]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2014-04-17 06:04:09]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dalecurtis@chromium.org/235723003/80001

[commit-bot: I haz the power, 2014-04-17 06:06:50]

Change committed as 264443