All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=841849

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=aef6a395a816648f7a75d982cc979ce00035edecbedf0675a59786f5d8a7d63a


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

android-nexus5

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/150b97d7c40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/150b97d7c40000

Disable error page isolation temporarily. by nasko@chromium.org
https://chromium.googlesource.com/chromium/src/+/d515cab0d9f6913d8a139dc0484a451f6d23524a

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

Wontfix. This is just a regression of a previous improvement. Sorry about the noise.

Here is the CL which I believe improved the metric:

https://chromium.googlesource.com/chromium/src/+/d83b571dc700010f144b68fe3d8628923c008773

The fact that error page isolation has effect on memory metrics during perf runs suggests that we are hitting errors in navigations in our metrics. If that is correct, this means that our metrics might not be measuring what we believe we are measuring. I would suggest investigating this by the benchmark owners.

Re #5: Filed bug 842323 on the error page in the benchmark.