All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=830828

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=e7e7f9b54eab0634e006bc7e81118f746efe6889b764dc139653529d5c2513b0


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

android-nexus5

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14a41cf4c40000

Speed>Bisection: Pinpoint job done (no repro), but never posted back.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/16dac60ac40000

^^ new job, wider range

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/16dac60ac40000

Add fieldtrial testing config for video SurfaceLayer. by mlamouri@chromium.org
https://chromium.googlesource.com/chromium/src/+/2725fa428722a5c26d55d78cfc299069521ce78d

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

The regression is on memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg

And it seems to randomly regress (roughly every other tests).

+enne@ is this something we should expect given the change?

No, this is not something that I would expect.  My expectation is that surfaces for video might change the timing of frames (with video frames being more regular than they would have been and the compositor maybe being delayed in some cases).  However, the amount of textures that are in flight should be the same.

You can do a memory trace in about:tracing and look at a breakdown of where that memory is going.  Maybe that would help illuminate what's going on.  Because this is flaky, I suspect maybe some sort of timing issue where the renderer compositor has more textures in flight possibly because it missed a frame, or something.

It's interesting that this is the only test that regresses too.  All of the others are neutral (or better?) in that time frame.  Is this only on Android too?

Assigning to self as  https://chromium.googlesource.com/chromium/src.git/+/74be4bc85264530ca32901400e3a50f9bf60dae3 could have fixed this. The test is a bit flaky and goes up and down all the time so I would rather wait a couple of days.

The NextAction date has arrived: 2018-06-13