All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=596469

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgkMjSsgoM


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

android-nexus6

This is likely due to me flipping on the variations for the unified media pipeline: 
https://chromeperf.appspot.com/group_report?rev=382165

This actually looks like an improvement per the graph, not a regression?

Oh I see, the improvement is some revisions before (unfortunately for me :)

In either case, this wasn't entirely unexpected, we're now allocating memory that was previously uncounted by Chrome (since it was listed in Android system MediaServer process).

Are we looking at the same thing?  I'm seeing a nice 10% improvement followed by a possible 1.4% regression.  The latter is what I'm looking into :)

Yeah I got too excited and thought this was pointing at the improvement :)

OK if this was expected, feel free to close :) 

Will wait for bisect to come back and then relabel and close as wontfix if it points to my change.

This is within expectations, so marking as WontFix.