All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=810504

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=2b3300cb78ca15e40beb430ebc6e53399e064f90d8f172ea33611efc2b5f7bae


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

win-high-dpi

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1660a839840000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/1660a839840000

Reland "Add HTTP-Bad Phase 3 experiment to field trial testing config" by estark@google.com
chromium @ e1b3c560ca1f4dfb3ce74911b6cdc7e32254cc23

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

estark, can you take a look? It looks like your CL 'Reland "Add HTTP-Bad Phase 3 experiment to field trial testing config"' caused a slight regression to IMDB on Windows.

Please see https://bugs.chromium.org/p/chromium/issues/detail?id=809490#c5 -- we've gotten a number of these regressions assigned for drawing a security chip in the omnibox on page load, and it's not clear why that makes any difference to the benchmark.

+tdresser, kouhei: can one of you help estark and elawrence understand why drawing the security chip in the omnibox affects this benchmark? see  bug 809490  and  764344 

None of the breakdown metrics show the regression, which isn't a good sign.

The closest are:
https://chromeperf.appspot.com/report?sid=6dfd48455fa1d64533b665916b437238cae4eff5b607736381e4e3fb4d01985a&start_rev=529182&end_rev=539411

I don't understand the little histogram at the side of the pinpoint run - the traces only show a single FCP per trace, but the histogram has a ton of values.

There's enough noise that the traces aren't very useful.

Unassigning myself as I'm not sure how to address this without help from the benchmark owner

Since the feature is Finchd, I think we should monitor the UMA dashboard results.
I checked it right now, but it didn't have statistically significant amount of samples yet.