All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=847506

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=5271f6ba0b5e6278d677a6c1acba665a061a17891468552545efe5b53dd76ec2


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

android-one

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1688d766240000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/1688d766240000

Stop getting the user-agent string in renderer by yhirano@chromium.org
https://chromium.googlesource.com/chromium/src/+/b0465f7f1ab77ae73cf2dd4c0950fee51f7528d4
0.4249 → 0.4306 (+0.00575)

Roll AFDO from 68.0.3432.0_rc-r2 to 68.0.3433.0_rc-r1 by afdo-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/9263cc55b019c080fc4f0fd62b5152b4aa1066bb
0.4303 → 0.4367 (+0.006375)

Assigning to sheriff gbiv@chromium.org because "Roll AFDO from 68.0.3432.0_rc-r2 to 68.0.3433.0_rc-r1" is a roll.

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

The tick down at '560966 - 561030' appears to have another AFDO roll in it, and it looks like we're roughly stable at pre-regression levels after that point (though given that this is a 2% regression, it's a bit hard to say for sure).

I'm trying to address AFDO instability in general at the moment; will look into this as a part of that, but I suspect it's already been 'fixed', so I won't be rushing to stare at it. :)