All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=596079

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoPGwoAkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoPPoqQkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4LiupQkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoK22vgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoK71vwkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4MSsrAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoK7cpgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoI2OqQsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoIvyqwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoK34tgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoO7OpQoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgoJDVtwsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4J6NuQoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4LrMpAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4KCYuQoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4MDPqAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgwOeptgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg4MSwtQoM


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

webrtc-linux-large-tests
webrtc-mac-large-tests
webrtc-win-large-tests

Hi Alex, these performance regressions look like they're caused by your CL: https://codereview.webrtc.org/1773173002 , could you have a look to see if that's expected? Thanks!

Well, it definitively is expected that the complexity increases since we don't downsample in the reverse stream. But the increase seems unreasonably high. I am going to have a look next week.

aluebs: will you have time to look into this bug this week?

Yes, I will look into it this week.

Here I have a CL up for review that should at least get rid of the crazy high values: https://codereview.webrtc.org/1844583003/

But some increase in complexity in the reverse stream is expected, since the splitting filter is used instead of the downsampler.

The CL landed, but I am not sure I can see the impact it had on this regression.

My CL improved many of the metrics, but also surfaced a threading issue, that is tracked here: https://bugs.chromium.org/p/chromium/issues/detail?id=599944
Closing this bug favor of that one, but feel free to re-open if you have any concerns.