1151039 - 3% Win7 tp5o regression on Mozilla-Inbound on April 02, 2015 from push 0f1095c9567f

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

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

Talos has detected a Firefox performance regression from your commit 0f1095c9567f in bug 1144317.  We need you to address this regression.

This is a list of all known regressions and improvements related to your bug:
http://alertmanager.allizom.org:8080/alerts.html?rev=40cbc1892b3a&showAll=1

On the page above you can see Talos 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, please see: https://wiki.mozilla.org/Buildbot/Talos/Tests#tp5

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 win32 -u none -t tp5  # add "mozharness: --spsProfile" to generate profile data

To run the test locally and do a more in-depth investigation, first set up a local Talos environment:
https://wiki.mozilla.org/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 tp5

Making a decision:
As the patch author we need your feedback to help us handle this regression.
*** Please let us know your plans by Wednesday, or the offending patch will be backed out! ***

Our wiki page oulines the common responses and expectations:
https://wiki.mozilla.org/Buildbot/Talos/RegressionBugsHandling

Comment 1

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

note- this was tracked down to mozilla-inbound, I see this on pgo:
https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&filter-searchStr=chromez windows pgo&fromcahnge=8c968097ff98&tochange=78a291fd62b3

Comment 2

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

and the graph is here for non-pgo which shows the problem as well:

http://graphs.mozilla.org/graph.html#tests=[[255,131,25]]&sel=1427573160000.2295,1428086406064.8525&displayrange=30&datatype=geo

Comment 3

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

sorry, the treeherder url above is incorrect, that is for another bug I am investigating, this is the tp5 one for win7:
https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&filter-searchStr=Windows 7 32-bit mozilla-inbound talos tp5o&fromchange=171bbcec7d18&tochange=8762f2f5da16

Comment 4

[Mason Chang [Inactive] [:mchang]]

All the session restore improvements are mine as well. Can we please not back out bug 1144317. It greatly improves smoothness, so this regression might have to be taken. Overall, the patch improves Firefox's performance.

Comment 5

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

got it!  thanks for weighing in here.  If you want to look at the specific pages which regressed, this view helps:
https://treeherder.mozilla.org/perf.html#/graphs?timerange=1209600&series=%257B%2522project%2522%253A%2522mozilla-inbound%2522%252C%2522signature%2522%253A%25224f4c985f96640473553241c0335ebad898d5f8fd%2522%252C%2522visible%2522%253Atrue%257D&highlightedRevision=0f1095c9567f&zoom=%7B%7D (wait for it to load and scroll down to see individual pages, google.com is the second worse near the bottom)

Mason, when could you make a decision to accept this, or work on it?

Comment 6

[Mason Chang [Inactive] [:mchang]]

This is what we had from compare talos before bug 1144317 landed - http://compare-talos.mattn.ca/dev/?oldBranch=Try&oldRevs=c0ae486a7c8f&newBranch=Try&newRev=9f89885a1e84&submit=true

No tp50 regression, so something changed.

I'll take a look at this after a few other blockers. But overall, this is a very nice performance win for Firefox.

Comment 7

[Mason Chang [Inactive] [:mchang]]

(In reply to Joel Maher (:jmaher) from comment #5)
> got it!  thanks for weighing in here.  If you want to look at the specific
> pages which regressed, this view helps:
> https://treeherder.mozilla.org/perf.html#/
> graphs?timerange=1209600&series=%257B%2522project%2522%253A%2522mozilla-
> inbound%2522%252C%2522signature%2522%253A%25224f4c985f96640473553241c0335ebad
> 898d5f8fd%2522%252C%2522visible%2522%253Atrue%257D&highlightedRevision=0f1095
> c9567f&zoom=%7B%7D (wait for it to load and scroll down to see individual
> pages, google.com is the second worse near the bottom)
> 
> Mason, when could you make a decision to accept this, or work on it?

It looks like it came back down Today, April 3rd though?

Comment 8

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

it is pretty close to back down, maybe a couple ms off the total summarized number.  If it isn't obvious I wouldn't spend a lot of time on it.

Comment 9

[Vaibhav (:vaibhav1994)]

We have got a couple of more regressions for this bug: http://alertmanager.allizom.org:8080/alerts.html?rev=d2f72af9fc50&showAll=1&testIndex=0&platIndex=0

It causes 5.88% regression on "SVG, Opacity Row Major" test and 2.7% regression on "Tp5 Optimized" test on Mozilla-inbound for windows.

Comment 10

[Avi Halachmi (:avih)]

(In reply to Vaibhav (:vaibhav1994) from comment #9)
> We have got a couple of more regressions for this bug:
> http://alertmanager.allizom.org:8080/alerts.
> html?rev=d2f72af9fc50&showAll=1&testIndex=0&platIndex=0
> 
> It causes 5.88% regression on "SVG, Opacity Row Major" test and 2.7%
> regression on "Tp5 Optimized" test on Mozilla-inbound for windows.

These are the symptoms of reader mode more than vsync IMO.

Comment 11

[Mason Chang [Inactive] [:mchang]]

(In reply to Avi Halachmi (:avih) from comment #10)
> (In reply to Vaibhav (:vaibhav1994) from comment #9)
> > We have got a couple of more regressions for this bug:
> > http://alertmanager.allizom.org:8080/alerts.
> > html?rev=d2f72af9fc50&showAll=1&testIndex=0&platIndex=0
> > 
> > It causes 5.88% regression on "SVG, Opacity Row Major" test and 2.7%
> > regression on "Tp5 Optimized" test on Mozilla-inbound for windows.
> 
> These are the symptoms of reader mode more than vsync IMO.

What is reader mode?

Comment 12

[Avi Halachmi (:avih)]

(In reply to Mason Chang [:mchang] from comment #11)
> What is reader mode?

Bug 1150656.

Comment 13

[Mason Chang [Inactive] [:mchang]]

(In reply to Vaibhav (:vaibhav1994) from comment #9)
> We have got a couple of more regressions for this bug:
> http://alertmanager.allizom.org:8080/alerts.
> html?rev=d2f72af9fc50&showAll=1&testIndex=0&platIndex=0
> 
> It causes 5.88% regression on "SVG, Opacity Row Major" test and 2.7%
> regression on "Tp5 Optimized" test on Mozilla-inbound for windows.

See bug 	1151603

Comment 14

[Mason Chang [Inactive] [:mchang]]

From comment 8, resolving as WFM.