All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=830517

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=3f95a3e470134b5d24ccbad3e6d2ee4f865f9a0d238a5d9b008dde90a7cd3a01


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

chromium-rel-win10

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1393abccc40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/1393abccc40000

Update Windows order files by hans@chromium.org
https://chromium.googlesource.com/chromium/src/+/8a45d2a9c56fd9a9de3bbdf188f1a3ed1f9c65b3

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

This doesn't make sense.. it seems the graph just reset to about the level it was a few days before.

How reliable is this metric? For example, the perf report for my change (https://chromeperf.appspot.com/group_report?rev=548389) indicates it reduced memory usage of media.desktop for another benchmark by 50%. On Mac :-p

Seems flaky to me.

https://cs.chromium.org/chromium/src/tools/perf/benchmark.csv?type=cs&q=memory.desktop&sq=package:chromium&l=34 suggests memory.desktop is owned by erikchen. erikchen, is that metric known to be super flaky? Is there a bug for that? (See previous comment)

The chart linked in this regression is for "memory:chrome:all_processes:reported_by_chrome:v8:effective_size".

I personally don't think we should alert on this because it's flaky, see the bug I filed: https://bugs.chromium.org/p/chromium/issues/detail?id=736506

The v8 team requested that we keep this alert: https://bugs.chromium.org/p/chromium/issues/detail?id=736506#c8

The V8 metric in the linked graph https://chromeperf.appspot.com/group_report?bug_id=830517 varies by +/-200KB on 8MB (i.e. about 3%).
I have a pending change that considerably reduces then noise of V8 memory metrics: https://chromium-review.googlesource.com/c/catapult/+/983775

The metric in comment #4 that changed by 50% is the compositor memory metric (memory:chrome:all_processes:reported_by_chrome:cc), which is unrelated to V8 metric.