All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=608644

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICglK3QowoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICglPrArwoM


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

webrtc-mac-large-tests

 Issue 608643  has been merged into this issue.

Minyue, it looks like your CL (https://codereview.webrtc.org/1901633002) is causing a regression in several NetEQ perf stats. Could you have a look at it?

Problem is found. The reason is that the performance of sync buffer in NetEq is affected by its length. A quick fix is at hand. It may not get the value down to what it was, though.

The graphs confirm that the values dropped to a lower level. Do we need to look into this any further, or do you think we can close the issue?

We can keep this bug open. I can try to drag it down a bit more.

Perf sheriff ping - looks like we're making progress here though.

We may likely see a significant improvements in the following days, when this takes its effect 

https://codereview.webrtc.org/1948483002/

look good so far. I think we can close this issue.

Nice. So most bots actually show an improvement comparing with the original code. Good work!

Nice work Minyue!