All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=656019

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg1ZnA6wgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgtfKd-AoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgtfDfhQgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg9ayUsQoM


Bot(s) for this bug's original alert(s):

chromium-webrtc-rel-7
chromium-webrtc-rel-linux
chromium-webrtc-rel-mac
chromium-webrtc-trunk-tot-rel-linux

This patch was reverted in https://codereview.chromium.org/2403363003, and it seems that this revert has fixed this regression.

I will keep this bug active for re-landing the patch in question.

It looks like the expanded samples are present only in the beginning of the test:
Before: [0.1068115234375,0,0,0,0,0,0,0,0,0]
After: [0.14984130859375,0,0,0,0,0,0,0,0,0]

Number of expanded samples stays approximately the same, it is the length of the first reporting interval which has gone down:
Before:
expanded_speech_samples = 520 expanded_noise_samples = 3480 timestamps_since_last_report = 37440 expand_rate = 1750
After:
expanded_speech_samples = 1084 expanded_noise_samples = 3160 timestamps_since_last_report = 28320 expand_rate = 2455

Is it a problem with the metric?

+hlundin,

Hi Henrik,

 altimin@ is now trying to relanding the CL and he figured the reason why expand_rate is affected.

Would you help judge if this is acceptable?

Hi, Henrik,

Would be good to take a look at this.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf

commit d8bd26c6124f76be5a0c7cfacc39c3996dd29adf
Author: altimin <altimin@chromium.org>
Date: Fri Nov 04 11:44:54 2016

[scheduler] Teach scheduler about audio state

This patch wires information about audio state to renderer scheduler
and stops all throttling (background and offscreen) while
audio is playing and for a short period after audio is stopped.

BUG= 642321 , 616519 ,  656019 
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/a7a651d546b76499821b4ba47b8f017d4b8becac
Review-Url: https://codereview.chromium.org/2383473002
Cr-Original-Commit-Position: refs/heads/master@{#423405}
Cr-Commit-Position: refs/heads/master@{#429861}

[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/cc/test/ordered_simple_task_runner.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/content/browser/media/audio_stream_monitor.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/content/browser/web_contents/web_contents_impl.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/content/browser/web_contents/web_contents_impl.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/content/common/page_messages.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/content/public/browser/web_contents.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/content/renderer/render_view_impl.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/content/renderer/render_view_impl.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/platform/scheduler/renderer/renderer_scheduler_impl.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/platform/scheduler/renderer/renderer_scheduler_impl.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/platform/scheduler/renderer/task_queue_throttler.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/platform/scheduler/renderer/task_queue_throttler.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/platform/scheduler/renderer/task_queue_throttler_unittest.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/platform/scheduler/renderer/web_view_scheduler_impl.cc
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/platform/scheduler/renderer/web_view_scheduler_impl.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/web/WebViewImpl.cpp
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/Source/web/WebViewImpl.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/public/platform/WebViewScheduler.h
[modify] https://crrev.com/d8bd26c6124f76be5a0c7cfacc39c3996dd29adf/third_party/WebKit/public/web/WebView.h

The graphs regressed again, and it seems to correlate with https://chromium.googlesource.com/chromium/src/+/afe6872b2448af4355b7792d006e0cee83791cc5.

altimin@, is this similar to your original and reverted CL?

Yes, it's the same thing chopped into pieces and relanded. I still stand by my claim that it's not a real regression but an artifact of measurement process.

I agree with you. Thanks for helping us!