All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=591021

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDA7_T9qwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDA77mIrwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDA79Dp5QoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr-vlqgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr4LJtwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDA75DLrwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr-3KrgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr86XuAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr-ybrgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr5OTpgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr8DOsQoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr-qJtAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr86krgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr-qWpQoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr4mCsAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAz9exuAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr-rCuAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr5PKswoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDAr5O0ogoM


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

webrtc-android-tests-nexus5
webrtc-android-tests-nexus72
webrtc-mac-large-tests
webrtc-win-large-tests

maybe related to https://chromium.googlesource.com/chromium/src/+/f6bbfe118c462885ca8cfd16a581c8ab6ad3a0e2

but I'd assign to a video expert to verify.

Erik, could you look at this one?

The roll mentioned in #2 contains a change that reduces encoding time for VP9
(about ~5-7% speedup), so that probably accounts for at least some of these improvements.

As per #5, closing as wai