All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=643682

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgocnU9QsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoe_stwkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4eSTrgsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgod_vvgoM


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

webrtc-android-tests-nexus5
webrtc-android-tests-nexus72

Hi Henrik, it looks like there was a recent regression in neteq perf stats, but it's unclear which CL is causing it due to the very gradual deterioration. Could you have a look and investigate?

I tried to reproduce this locally on a Nexus 5 in a number of different ways (including git bisect). Eventually I built and tested #13987 and #14009 in release builds on the 0_pl_0_drift test. My results were as follows:

#13987: 8992, 8823, 8874, 8817, 8871, 8872, 8903, 8878, 8761, 8826, 8811
average: 8857

#14009: 8710, 8721, 8741, 8736, 8707, 8709, 8744, 8725, 8732, 8740, 8700
average: 8724

Could it be related to build changes? (I used only GN).

Maybe build related, but nevertheless I think we can close this. You've done local testing and could not repro.