All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=907743

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=7cb1366f9ece57f97db8e5115a22e4fb3aaf76aaa3975ede7e5ba46622fb8bd7


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

webrtc-android-tests-nexus4-lollipop
webrtc-android-tests-nexus72
webrtc-linux-large-tests
webrtc-mac-large-tests
webrtc-win-large-tests

marpan: can you please have a look and see if the changes are related to the libvpx roll.

We did make some changes in libvpx that could have caused this. Most of the changes in these tests are positive (lower number of dropped frame, lower CPU, lower encode_times), there are some small regressions in psnr, but overall we'll take the change.

Surprised to see decrease in encode time for ARM, not sure we did anything for that in last libvpx roll. Jerome can you take a look?

Marking as won't fix, since most of the changes are positive. Though worth looking into what caused the decrease in encode_time for nexus tests