All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=599571

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgiLCFuQoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg8J75rwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg8PfhoAoM


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

chromium-webrtc-trunk-tot-rel-mac

CL range:
http://test-results.appspot.com/revision_range?start=383743&end=383783

That range includes phoglund's test filtering logic change (https://codereview.chromium.org/1839653003), but I doubt that is the culprit here.

In that range, I see a WebRTC roll:
https://codereview.chromium.org/1840133002

Changes in that roll:
https://chromium.googlesource.com/external/webrtc/trunk/webrtc.git/+log/df760cd..ce55976

Most CLs in that roll seem related to audio, and of the non-audio ones, nothing immediately stands out. :(

mflodman@, can you have someone take a look?

I think I know what this is, that machine was upgraded from 10.9 to 10.10 at March 29, which fits with the regression. I'll leave it to mflodman@ to close the bug though (in case there may be something else to look at).

I did a few local test runs on my mac, with builds at r383743 and r383783 respectively. Can't say I see any real differences, both of them are in line with the behavior after the regression.

I'm thinking this is related to https://bugs.chromium.org/p/chromium/issues/detail?id=600508

Kjellander, could this be caused by the OS upgrade you mention there?

Right, there were several other regressions caused by that upgrade. I just didn't want to close it blindly because of that, but have someone else at least have a quick look first.

Hm, just to be sure I backed down a fair bit, to r383400. I looks like results are now more consistent with how it looked before the regression, so maybe there is something here after all...
I'll continue bisecting.

I retract my previous statement. After a very slow and painful attempt to bisect this, it seems that when I first tried 383400, the stars aligned a just happened to produce 5 nice runs in a row. I no longer see any difference there, so I'm gonna chalk this down to the OS upgrade after all.
I'll close this as won't fix. :(