New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 759671 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner: ----
Closed: Oct 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

202.7% regression in loading.desktop at 494361:494538

Project Member Reported by kraynov@chromium.org, Aug 28 2017

Issue description

See the link to graphs below.
 
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Aug 28 2017

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=759671

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=ffa609be62a072b7934f9695229715d063e883dc77e5a3009886c8ed8e270b9d


Bot(s) for this bug's original alert(s):

chromium-rel-mac12-mini-8gb
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Aug 28 2017


=== BISECT JOB RESULTS ===
NO Perf regression found

Bisect Details
  Configuration: mac_10_12_mini_8gb_perf_bisect
  Benchmark    : loading.desktop
  Metric       : cpuTimeToFirstMeaningfulPaint_avg/pcv1-cold/ja.wikipedia

Revision             Result                  N
chromium@494360      2619.34 +- 333.862      21      good
chromium@494538      2668.0 +- 379.991       19      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=ja.wikipedia loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970000411624063584


For feedback, file a bug with component Speed>Bisection
Kicked off a few more bisects, some on wider range. Looks like there may be multiple regressions.
Cc: kerrnel@chromium.org
Owner: kerrnel@chromium.org
Status: Assigned (was: Untriaged)

=== Auto-CCing suspected CL author kerrnel@chromium.org ===

Hi kerrnel@chromium.org, the bisect results pointed to your CL, please take a look at the
results.


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Greg Kerr
  Commit : 9705b7aaabf4c81d4e9cb13174338df9bde61cc4
  Date   : Tue Aug 15 19:42:42 2017
  Subject: Enable Finch Trial Test for MacV2Sandbox.

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : loading.desktop
  Metric       : cpuTimeToFirstMeaningfulPaint_avg/pcv1-warm/FDA
  Change       : 14.41% | 281.420833333 -> 321.974833333

Revision             Result                  N
chromium@494399      281.421 +- 2.82812      6      good
chromium@494465      282.45 +- 7.14277       6      good
chromium@494498      282.956 +- 4.36996      6      good
chromium@494500      282.822 +- 8.82162      6      good
chromium@494501      285.129 +- 6.97171      6      good
chromium@494502      320.493 +- 7.44149      6      bad       <--
chromium@494506      319.543 +- 4.47796      6      bad
chromium@494514      320.822 +- 8.16364      6      bad
chromium@494530      317.518 +- 9.27686      6      bad
chromium@494660      321.975 +- 6.21749      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=FDA loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8966189902716904080


For feedback, file a bug with component Speed>Bisection
Cc: benhenry@chromium.org
9705b7aaabf4c81d4e9cb13174338df9bde61cc4 was reverted, so if this regression still exists, something else is the culprit.
Owner: ----
Status: Available (was: Assigned)
Do you know so many of the perf bisect jobs pin the cause on CLs that were already reverted?
Project Member

Comment 10 by 42576172...@developer.gserviceaccount.com, Oct 10 2017


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Greg Kerr
  Commit : 9705b7aaabf4c81d4e9cb13174338df9bde61cc4
  Date   : Tue Aug 15 19:42:42 2017
  Subject: Enable Finch Trial Test for MacV2Sandbox.

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstMeaningfulPaint_avg/pcv1-cold/AirBnB
  Change       : 5.75% | 1449.65222222 -> 1532.99811111

Revision             Result                  N
chromium@494122      1449.65 +- 118.767      9       good
chromium@494456      1464.08 +- 53.076       9       good
chromium@494498      1464.08 +- 107.009      14      good
chromium@494501      1453.43 +- 61.039       9       good
chromium@494502      1542.89 +- 92.0982      6       bad       <--
chromium@494503      1534.8 +- 183.863       14      bad
chromium@494504      1530.63 +- 76.236       9       bad
chromium@494509      1544.8 +- 32.788        6       bad
chromium@494519      1502.15 +- 150.012      14      bad
chromium@494540      1526.1 +- 97.0282       9       bad
chromium@494623      1531.0 +- 54.7498       6       bad
chromium@494790      1548.57 +- 44.8259      6       bad
chromium@495457      1533.0 +- 78.664        9       bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=AirBnB loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8966189918177581088


For feedback, file a bug with component Speed>Bisection

Comment 11 by dtu@chromium.org, Oct 10 2017

Let me try one more with a different range.
Project Member

Comment 13 by 42576172...@developer.gserviceaccount.com, Oct 10 2017

Cc: pmeenan@chromium.org
 Issue 760144  has been merged into this issue.
Project Member

Comment 14 by 42576172...@developer.gserviceaccount.com, Oct 10 2017


=== BISECT JOB RESULTS ===
NO Perf regression found

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstMeaningfulPaint_avg/pcv1-cold/AirBnB

Revision             Result                  N
chromium@494277      1433.44 +- 134.24       21      good
chromium@494399      1449.59 +- 184.409      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=AirBnB loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8966158133356456736


For feedback, file a bug with component Speed>Bisection
Status: Fixed (was: Available)
kerrnel: sorry for all the noise. Yes, perf bisects do assign a culprit even if the CL has already been reverted. There are a lot of ups and downs in these graphs so it wasn't possible to tell whether the regression was related to Mac V2 sandbox or something else. I split off some unrelated alerts into  bug 773263  and closing this one as fixed by the revert.

Sign in to add a comment