Closed Bug 990183 Opened 10 years ago Closed 10 years ago

3..44% tart regression on windows xp after the central -> aurora merge

Categories

(Core :: JavaScript Engine, defect)

x86
Windows XP
defect
Not set
normal

Tracking

()

RESOLVED WONTFIX

People

(Reporter: jmaher, Assigned: jandem)

References

Details

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

tart regressed on windows xp when we merged central to aurora 2 weeks ago:
http://graphs.mozilla.org/graph.html#tests=[[293,52,37],[293,1,37],[293,132,37]]&sel=1393701595288,1396293595288&displayrange=30&datatype=running

it is hard to determine what really happened, but it looks like this merge from b2g-inbound might be the root cause:
http://hg.mozilla.org/mozilla-central/pushloghtml?changeset=d2dac18d0562

of course I don't see a bug from early march to call this out.  I know there were 2 weeks where I wasn't watching the tree as much and that would be right in the middle of that.

I wanted to get this on the books.  avi, did we rewrite tart and expect a regression?
(In reply to Joel Maher (:jmaher) from comment #0)
> I wanted to get this on the books.  avi, did we rewrite tart and expect a
> regression?

No and no.
did some retriggers, we don't run these tests on b2g-inbound, so I need to work of m-c for now:
https://tbpl.mozilla.org/?fromchange=32a41942974c&tochange=ff96e428da76&jobname=Windows%20XP%2032-bit%20mozilla-central%20talos%20svgr
bbouvier, this hasn't been getting much attention, Aurora will be going to beta in the near future.  Can you take a look at this?
Flags: needinfo?(benj)
I mixed up my need info, actually jdemooij, can you comment on this bug?
Flags: needinfo?(benj) → needinfo?(jdemooij)
Assignee: nobody → jdemooij
30 is our Beta now, if this is an unexpected performance change then please report in the bug about the options available for addressing it: backout offending patch, forward/speculative fix (would have to be ready to uplift in the next 2 weeks), or if there's nothing that can be done.
Sorry for the delay. This is basically the same as bug 987900 comment 8; hard to fix without adding more complexity or regressing perf elsewhere.

Joel, is this WinXP only?
Flags: needinfo?(jdemooij) → needinfo?(jmaher)
Sounds to me like a great candidate for wontfix.
as far as I can tell this is winxp only.

Looking on datazilla:
https://datazilla.mozilla.org/?start=1397730697&stop=1399816204&product=Firefox&repository=Mozilla-Beta&os=win&os_version=5.1.2600&test=tart&graph_search=c95ac01e332e&tr_id=5303331&graph=icon-close-DPI1.error.TART&x86_64=false&project=talos

we have these subtests regressing:
icon-close-DPI1.error.TART
icon-close-DPI2.error.TART
simple-close-DPI1.error.TART

Looking at the overall picture (not worrying about a this specific issue), the uplift to beta didn't show any regressions, only what could appear to be slight improvements.
Status: NEW → RESOLVED
Closed: 10 years ago
Flags: needinfo?(jmaher)
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.