All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=635886

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICglpHStQoM


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

webrtc-android-tests-nexus72

aleloi: Could this be related to your CL https://codereview.webrtc.org/2206223002 ?

I can't think of why that would be the case. The CL just routes calls from media_engine to voice_engine along a different path. I could be mistaken of course.

I'm currently running the test locally before and after my change. Is this the right test? (Never used the perf tests yet).

./out/Debug/webrtc_perf_tests --gtest_filter=FullStackTest.ForemanCif500kbps

The test printed this:
[after] RESULT media_bitrate: foreman_cif_500kbps = {472848.355932, 46091.035785} bps
[before] RESULT media_bitrate: foreman_cif_500kbps = {483912.440678, 51826.326657} bps

The numbers were higher before. Is the difference large enough for it to be a regression?

More data from 10 runs before and after my commit:

[after]
RESULT media_bitrate: foreman_cif_500kbps = {482117.118644, 52575.638891} bps
RESULT media_bitrate: foreman_cif_500kbps = {467676.864407, 50595.862686} bps
RESULT media_bitrate: foreman_cif_500kbps = {474071.033898, 45396.520069} bps
RESULT media_bitrate: foreman_cif_500kbps = {455254.322034, 45942.820188} bps
RESULT media_bitrate: foreman_cif_500kbps = {473287.508475, 46354.195718} bps
RESULT media_bitrate: foreman_cif_500kbps = {482059.728814, 52115.174141} bps
RESULT media_bitrate: foreman_cif_500kbps = {460246.067797, 48671.677389} bps
RESULT media_bitrate: foreman_cif_500kbps = {453636.847458, 45919.465630} bps
RESULT media_bitrate: foreman_cif_500kbps = {480199.000000, 48824.984509} bps
RESULT media_bitrate: foreman_cif_500kbps = {467532.525424, 57975.279313} bps


[before]

RESULT media_bitrate: foreman_cif_500kbps = {460843.898305, 49189.237499} bps
RESULT media_bitrate: foreman_cif_500kbps = {470074.322034, 50331.156245} bps
RESULT media_bitrate: foreman_cif_500kbps = {474943.813559, 50938.498567} bps
RESULT media_bitrate: foreman_cif_500kbps = {470452.000000, 50528.778188} bps
RESULT media_bitrate: foreman_cif_500kbps = {459460.677966, 47288.986803} bps
RESULT media_bitrate: foreman_cif_500kbps = {484727.440678, 51458.606900} bps
RESULT media_bitrate: foreman_cif_500kbps = {480594.406780, 50636.787068} bps
RESULT media_bitrate: foreman_cif_500kbps = {457423.881356, 54395.768977} bps
RESULT media_bitrate: foreman_cif_500kbps = {472221.169492, 50318.283210} bps
RESULT media_bitrate: foreman_cif_500kbps = {479274.118644, 51903.356893} bps


I will look at it more closely tomorrow.

 Issue 635885  has been merged into this issue.

It does not look as if your CL was the culprit for this. I'll take the issue and see if I can find another reason.

holmer@: This seems related to bit-rate selection. But I cannot find anything in the list of CLs 
https://chromium.googlesource.com/external/webrtc/+log/bfc7a30c663dbc7ecad145b8e834ebb47b3f75ec..90860af8811ccb524516c9003cdade85d2305f0e)
and
https://chromium.googlesource.com/chromium/src/+log/33f87688e4..f65883a9c0
that is related to this.

Do you know of anything that could have caused this?
Also, do you have a feeling for how bad that regressions are?

holmer@: Could this be caused by your CL https://codereview.webrtc.org/2201093006 ?

Yes, probably. I think the detection was a bit off in this case, so that it detected it earlier than it should.