Add histogram for percentage of sent frames that are limited in resolution due to bandwidth.

Add histogram for percentage of sent frames that are limited in resolution due to bandwidth:
- "WebRTC.Video.BandwidthLimitedResolutionInPercent"

If the frame is bandwidth limited, the average number of disabled resolutions is logged:
- "WebRTC.Video.BandwidthLimitedResolutionsDisabled"

BUG=

Committed: https://crrev.com/da535c405597864b8396b2029dec70ab9fb76e8b
Cr-Commit-Position: refs/heads/master@{#10333}

[åsapersson, 2015-09-10 07:45:28]

Patchset #1 (id:1) has been deleted

[åsapersson, 2015-09-10 07:45:33]

Patchset #1 (id:20001) has been deleted

[åsapersson, 2015-09-10 07:45:40]

Patchset #1 (id:40001) has been deleted

[åsapersson, 2015-09-10 07:45:46]

Patchset #1 (id:60001) has been deleted

[åsapersson, 2015-09-10 07:46:06]

Patchset #1 (id:80001) has been deleted

[åsapersson, 2015-09-11 08:49:09]

asapersson@webrtc.org changed reviewers:
+ pbos@webrtc.org, stefan@webrtc.org

[pbos-webrtc, 2015-09-11 11:28:24]

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
File webrtc/video/send_statistics_proxy.cc (right):

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
webrtc/video/send_statistics_proxy.cc:68:
RTC_HISTOGRAM_PERCENTAGE("WebRTC.Video.BandwidthLimitedResolutionInPercent",
Can we report this per SSRC/simulcast layer? I think the following stats would
be useful:

WebRTC.Video.Simulcast_{0,1,2}.BandwidthLimited...

Because then we can see how often HD is limited, and how often VGA is limited
rather than how often any of them are limited.

[stefan-webrtc, 2015-09-16 07:49:27]

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/modules/video_cod...
File webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc (right):

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/modules/video_cod...
webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc:140: return true;
I think you could replace this with return std::find(streams.begin(),
streams.end(), false) == streams.end();

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/modules/video_cod...
webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc:958: bw_limited_resolution =
!SendingAllStreams(send_stream_);
Same reasoning as in the quality CL. Would it be useful to track how many
resolutions are disabled here?

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
File webrtc/video/send_statistics_proxy.cc (right):

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
webrtc/video/send_statistics_proxy.cc:68:
RTC_HISTOGRAM_PERCENTAGE("WebRTC.Video.BandwidthLimitedResolutionInPercent",
On 2015/09/11 11:28:23, pbos-webrtc wrote:
> Can we report this per SSRC/simulcast layer? I think the following stats would
> be useful:
> 
> WebRTC.Video.Simulcast_{0,1,2}.BandwidthLimited...
> 
> Because then we can see how often HD is limited, and how often VGA is limited
> rather than how often any of them are limited.

It doesn't scale very well, and as we make simulcast more general this may
become a problem, but it could be worth ignoring that issue for now if we think
tracking per layer is important.

I was thinking that we could track the number of resolutions we disable. Not the
same thing though.

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
File webrtc/video/send_statistics_proxy.h (right):

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
webrtc/video/send_statistics_proxy.h:105: struct BoolSampleCounter {
class

[åsapersson, 2015-09-17 14:12:03]

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/modules/video_cod...
File webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc (right):

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/modules/video_cod...
webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc:958: bw_limited_resolution =
!SendingAllStreams(send_stream_);
On 2015/09/16 07:49:27, stefan-webrtc (holmer) wrote:
> Same reasoning as in the quality CL. Would it be useful to track how many
> resolutions are disabled here?

Done.

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
File webrtc/video/send_statistics_proxy.cc (right):

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
webrtc/video/send_statistics_proxy.cc:68:
RTC_HISTOGRAM_PERCENTAGE("WebRTC.Video.BandwidthLimitedResolutionInPercent",
On 2015/09/16 07:49:27, stefan-webrtc (holmer) wrote:
> On 2015/09/11 11:28:23, pbos-webrtc wrote:
> > Can we report this per SSRC/simulcast layer? I think the following stats
would
> > be useful:
> > 
> > WebRTC.Video.Simulcast_{0,1,2}.BandwidthLimited...
> > 
> > Because then we can see how often HD is limited, and how often VGA is
limited
> > rather than how often any of them are limited.
> 
> It doesn't scale very well, and as we make simulcast more general this may
> become a problem, but it could be worth ignoring that issue for now if we
think
> tracking per layer is important.
> 
> I was thinking that we could track the number of resolutions we disable. Not
the
> same thing though.

Added the average number of resolutions disabled when bw limited.

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
File webrtc/video/send_statistics_proxy.h (right):

https://codereview.webrtc.org/1311533012/diff/100001/webrtc/video/send_statis...
webrtc/video/send_statistics_proxy.h:105: struct BoolSampleCounter {
On 2015/09/16 07:49:27, stefan-webrtc (holmer) wrote:
> class

Done.

[åsapersson, 2015-09-18 10:59:26]

asapersson@webrtc.org changed reviewers:
+ mflodman@webrtc.org

[stefan-webrtc, 2015-09-23 15:10:35]

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/video/send_statis...
File webrtc/video/send_statistics_proxy.cc (right):

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/video/send_statis...
webrtc/video/send_statistics_proxy.cc:71: int num_disabled =
bw_resolutions_disabled_counter_.Avg(kMinRequiredSamples);
Same as in the other CL, is this average per frame or per second?

[åsapersson, 2015-09-24 08:23:45]

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/video/send_statis...
File webrtc/video/send_statistics_proxy.cc (right):

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/video/send_statis...
webrtc/video/send_statistics_proxy.cc:71: int num_disabled =
bw_resolutions_disabled_counter_.Avg(kMinRequiredSamples);
On 2015/09/23 15:10:34, stefan-webrtc (holmer) wrote:
> Same as in the other CL, is this average per frame or per second?

Per frame.
If a frame has resolutions disabled, |bw_resolutions_disabled_counter_| is
updated with the number of disabled resolutions for that frame. The average is
recorded.

[stefan-webrtc, 2015-09-28 13:45:11]

lgtm

[pbos-webrtc, 2015-09-28 15:34:38]

lgtm, please file a bug to implement this in SimulcastEncoderAdapter so we can
get stats on Android/ChromeOS/platforms with HW encoders.

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/modules/video_cod...
File webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc (right):

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/modules/video_cod...
webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc:958: if (bw_available)
int bw_resolutions_disabled = bw_available ? NumStreamsDisabled(send_stream_) :
0;

If you want to.

[åsapersson, 2015-09-28 15:47:39]

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/modules/video_cod...
File webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc (right):

https://codereview.webrtc.org/1311533012/diff/140001/webrtc/modules/video_cod...
webrtc/modules/video_coding/codecs/vp8/vp8_impl.cc:958: if (bw_available)
On 2015/09/28 15:34:38, pbos-webrtc wrote:
> int bw_resolutions_disabled = bw_available ? NumStreamsDisabled(send_stream_)
:
> 0;
> 
> If you want to.

Done.

[åsapersson, 2015-10-16 11:08:45]

Magnus, can you take a look at webrtc/video_frame.h?

[mflodman, 2015-10-16 11:55:54]

video_frame.h LGTM

[åsapersson, 2015-10-19 11:00:19]

The CQ bit was checked by asapersson@webrtc.org

[åsapersson, 2015-10-19 11:00:19]

The patchset sent to the CQ was uploaded after l-g-t-m from pbos@webrtc.org,
stefan@webrtc.org, mflodman@webrtc.org
Link to the patchset: https://codereview.webrtc.org/1311533012/#ps240001 (title:
"rebase")

[commit-bot: I haz the power, 2015-10-19 11:00:24]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1311533012/240001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1311533012/240001

[commit-bot: I haz the power, 2015-10-19 13:00:38]

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

[commit-bot: I haz the power, 2015-10-19 13:00:41]

Try jobs failed on following builders:
  android_arm64_rel on tryserver.webrtc (JOB_TIMED_OUT, no build URL)
  android_dbg on tryserver.webrtc (JOB_TIMED_OUT, no build URL)
  android_rel on tryserver.webrtc (JOB_TIMED_OUT, no build URL)

[åsapersson, 2015-10-20 06:31:27]

The CQ bit was checked by asapersson@webrtc.org

[commit-bot: I haz the power, 2015-10-20 06:31:37]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1311533012/240001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1311533012/240001

[commit-bot: I haz the power, 2015-10-20 06:32:45]

Committed patchset #8 (id:240001)

[commit-bot: I haz the power, 2015-10-20 06:32:58]

Patchset 8 (id:??) landed as
https://crrev.com/da535c405597864b8396b2029dec70ab9fb76e8b
Cr-Commit-Position: refs/heads/master@{#10333}