1268302 - 4.05 - 38.55% glterrain (linux64, windowsxp) regression on push 18d290be0df2e10e8904451d16f2b9aaf936974e (Wed Apr 27 2016)

Status: RESOLVED FIXED

(Core :: Graphics: CanvasWebGL, defect, P2)

Description

[William Lachance (:wlach)]

Talos has detected a Firefox performance regression from push 18d290be0df2e10e8904451d16f2b9aaf936974e. As author of one of the patches included in that push, we need your help to address this regression.

This is a list of all known regressions and improvements related to the push:

https://treeherder.mozilla.org/perf.html#/alerts?id=1036

On the page above you can see an alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the Talos jobs in a pushlog format.

To learn more about the regressing test(s), please see:

https://wiki.mozilla.org/Buildbot/Talos/Tests#glterrain

Reproducing and debugging the regression:

If you would like to re-run this Talos test on a potential fix, use try with the following syntax:

try: -b o -p linux64,win32 -u none -t g1[Windows XP],g1-e10s[Windows XP] --rebuild 5  # add "mozharness: --spsProfile" to generate profile data

(we suggest --rebuild 5 to be more confident in the results)

To run the test locally and do a more in-depth investigation, first set up a local Talos environment:

https://wiki.mozilla.lorg/Buildbot/Talos/Running#Running_locally_-_Source_Code

Then run the following command from the directory where you set up Talos:

talos --develop -e [path]/firefox -a glterrain

(add --e10s to run tests in e10s mode)

Making a decision:

As the patch author we need your feedback to help us handle this regression.
*** Please let us know your plans by Monday, or the offending patch(es) will be backed out! ***

Our wiki page outlines the common responses and expectations:

https://wiki.mozilla.org/Buildbot/Talos/RegressionBugsHandling

Comment 1

[Kelsey Gilbert [:jgilbert]]

This is only a regression for non-accel layers, so let's leave it in and fix it.

Comment 2

[Ethan Lin[:ethlin]]

Okay, I will check this issue.

Comment 3

[Ethan Lin[:ethlin]]

Attachment: Fix performance regression.

In bug 1250954, I added CloneSurface when surfaceClient has a mismatched forwarder. I should skip the surfaceClient without forwarder since the surfaceClient is just created and it will do InitIPDL correctly.
I also use AsCompositableForwarder to do the forwarder comparison. It should be a better way.

Comment 4

[Ethan Lin[:ethlin]]

try server:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=efe26a40b2e0

Comment 5

[Ethan Lin[:ethlin]]

The talos test on try server still got similar scores. 
*With this bug's patch: 
https://treeherder.mozilla.org/#/jobs?repo=try&revision=efe26a40b2e0&selectedJob=20534029&filter-searchStr=talos
The glterrain test on Linux x64 opt is 18.53. 

So I removed the patch in bug 1250954 and try again.
*Remove all patches:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=efe26a40b2e0&selectedJob=20534029&filter-searchStr=talos
The glterrain test on Linux x64 opt is 18.44. 

The two scores are almost the same and I think before regression this test's score should be about 24. Is the regression windows wrong?

Comment 6

[Avi Halachmi (:avih)]

18 is without the regression. It then regressed to 24. Also, your two links are identical.

Comment 7

[Ethan Lin[:ethlin]]

(In reply to Avi Halachmi (:avih) from comment #6)
> 18 is without the regression. It then regressed to 24. Also, your two links
> are identical.

I see. I misunderstanded the number. Thanks.

Comment 8

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/4f85ccb14c7a

Comment 9

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/4f85ccb14c7a

Comment 10

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

some winxp wins here:
https://treeherder.mozilla.org/perf.html#/alerts?id=1133

the linux stuff has odd patterns, waiting for the trees to open to collect more data and figure out what is going on.

Comment 11

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

linux stuff is fixed, the pattern appears to be infrastructure related, but I can decipher it now that I have more data.