VR CC memory regression |
||||||||
Issue descriptionhttps://chromium.googlesource.com/chromium/src/+/58c7f9d66e579c353b59e1154912b53be2061523 caused the memory regressions seen at https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICQxNSWrwoM. This is a pretty substantial increase of ~700%/15MB, but looking at the history of results, that used to be the norm until around December 15th. There, it dropped dramatically and has randomly been spiking back to its old value every now and then. The offending CL just causes it to occur 100% of the time again. My guess is that it's the same as Issue 801376 , and may be fixed with Issue 803174.
,
Feb 22 2018
,
Feb 28 2018
,
Apr 6 2018
,
Apr 16 2018
Actually, if you look at the latest graph, it looks like it was fixed for a short period of time and then regressed again. I did a bisect and it looks like Ian's CL which disable his optimization is the culprit. Let's keep watch the graph once the optimization is enabled again.
,
Apr 16 2018
I mean, this issue is clearly a race that's not directly related to any of our CLs landing now.
,
Apr 18 2018
,
May 10 2018
Issue 841921 has been merged into this issue.
,
May 10 2018
My guess is the variable regression here probably has something to do with what cc thinks the control container size is when VSync is paused.
,
May 15 2018
,
May 29 2018
Graphs have dropped back down to their original values for a few weeks now.
,
Jul 4
|
||||||||
►
Sign in to add a comment |
||||||||
Comment 1 by mthiesse@chromium.org
, Jan 19 2018Owner: ----
Status: Available (was: Assigned)