Closed Bug 1208444 Opened 6 years ago Closed 6 years ago

2-4% regression in TART seen on fx-team (v.44) September 24 from revision e5afaf840b1b

Categories

(Firefox :: Theme, defect)

defect
Not set
normal

Tracking

()

RESOLVED FIXED
Firefox 44

People

(Reporter: jmaher, Assigned: dao)

References

Details

(Keywords: perf, regression, Whiteboard: [talos_regression])

We have a series of alerts coming in regarding a tart regression.  I am still collecting data, but I wanted to get this in.

I did some retriggers initially for linux64 e10s tart and we see the values go from ~5.35 -> ~5.45 when we landed bug 1208037:
https://hg.mozilla.org/integration/fx-team/rev/e5afaf840b1b

a compare chooser view (e10s):
https://treeherder.allizom.org/perf.html#/compare?originalProject=fx-team&originalRevision=7b145ea2fbde&newProject=fx-team&newRevision=e5afaf840b1b&e10s=1

and the non e10s view:
https://treeherder.allizom.org/perf.html#/compare?originalProject=fx-team&originalRevision=7b145ea2fbde&newProject=fx-team&newRevision=e5afaf840b1b

I am awaiting all the data before listing the entire set of regressions.  I have seen alerts for almost all platforms, so there is a good chance this isn't specific to linux64 e10s.
Isn't this just the reverse of the gain that we got when we introduced the fade-out, a couple of days earlier?
(In reply to :Gijs Kruitbosch from comment #1)
> Isn't this just the reverse of the gain that we got when we introduced the
> fade-out, a couple of days earlier?

Maybe... hard to keep track of it at this point. I'll do a try run with a full backout to see if there's any meaningful net loss or gain.
Assignee: nobody → dao
thanks, this did the trick!
https://hg.mozilla.org/mozilla-central/rev/095a97e854f0
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → FIXED
Target Milestone: --- → Firefox 44
You need to log in before you can comment on or make changes to this bug.