All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=682640

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4Nb-6AgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4Pb3qQoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4LvtsgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgoJPApgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4PCJtAsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgoKPFuQsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4JDVmAsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4PagpgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4OjIpgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4JzurwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4NKesgkM


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

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

Hello,

these perf graph shows lots of changes, which could perhaps be explained as a result of the improved bandwidth estimate from your cl https://codereview.webrtc.org/2613543003

Can you have a look and say if these changes are expected and reasonable?

Who should this be assigned to?

It was intended for sergeyu. Reassigning.

I don't think there is anything actionable.
On Mac foreman_cif_delay_50_0_plr_5_ulpfec gets higher BW estimate as result more time is spent in the encoder.
Same in screenshare_slides_vp9_2sl: the sender gets a bit higher BW esimates that changes the number we get in that test.
TransportSequenceNumberSimulcastRedRtx seems rather noisy and is affected even by debug logging. 10packets is the same value it gets on Linux before my change

Thanks for looking into this. 

Stefan, do you agree no further action is needed regarding the ramp-up graphs? Packets sent increased from 6 to 10, and I guess most of the other graphs (byte counters and delays) follow from that.

yeah, seems ok.

For the record, the culprit CL was reverted in https://codereview.webrtc.org/2626473004/ (15979), and relanded in 16127.