1174771 - Win[XP/7/8] and Linux[32/64] talos regression allyr with e10s enabled (5%-40%).

Status: RESOLVED FIXED

(Testing :: Talos, defect)

Description

[Mike Conley (:mconley) (:⚙️)]

Initial measurements on mozilla-central changeset 0093691d3715 comparing e10s to non-e10s shows the following data for Windows after several retriggers:

Win XP:

:( a11yr                175.8 +/-  4% (6#)  [ +13.7%]     199.9 +/-  3% (6#)

Win 7:

:( a11yr                178.0 +/-  1% (6#)  [  +7.7%]     191.6 +/-  2% (6#)

Win 8:

:( a11yr                170.2 +/-  1% (6#)  [  +5.2%]     179.0 +/-  0% (1#)


Somehow, I think this is probably expected, as I suspect we're exercising the sync a11y APIs. Not sure if it can be optimized - we might just have to eat this regression, but I figure we should at least get it on file.

Comment 1

[Mike Conley (:mconley) (:⚙️)]

We're seeing similar regressions on Linux 32 and 64 as well:

Linux 32:

:( a11yr                247.6 +/-  1% (6#)  [ +36.7%]     338.3 +/-  1% (6#)

Linux 64:

:( a11yr                205.0 +/-  0% (6#)  [ +40.4%]     287.8 +/-  1% (6#)

Comment 2

[Joel Maher ( :jmaher ) (UTC -8)]

from June 29th:
Linux:
:( a11yr                250.8 +/-  1% (7#)  [ +36.1%]     341.3 +/-  1% (7#)
Linux64:
:( a11yr                212.6 +/-  0% (7#)  [ +37.4%]     292.2 +/-  1% (7#)
Win7:
:( a11yr                177.8 +/-  1% (7#)  [  +8.6%]     193.0 +/-  1% (7#)
WinXP:
:( a11yr                174.7 +/-  4% (7#)  [ +12.5%]     196.5 +/-  2% (7#)
Win8:
:( a11yr                173.1 +/-  1% (7#)  [  +6.8%]     184.8 +/-  2% (7#)
OSX 10.10:
:( a11yr                263.6 +/-  9% (7#)  [ +25.0%]     329.3 +/-  6% (7#)

note, we are adding osx 10.10 in here, otherwise linux/windows are the same.

Comment 3

[Brad Lassey [:blassey] (use needinfo?)]

This is now a win across the board except for linux 64, which is a 2% regression. Closing as fixed