All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=783622

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=77d03c99bda15fd21352ff99f8e8b00293b9dcb5f99703737e036edaaa77e6ae


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

android-nexus5X
android-nexus6
android-one

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8963323624347189200

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

Hi ulan@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 : Ulan Degenbaev
  Commit : 6bb1d47e6e82f75ead36cdab209afc5dafa8329c
  Date   : Tue Nov 07 13:11:10 2017
  Subject: [heap] Re-enable concurrent marking.

Bisect Details
  Configuration: android_one_perf_bisect
  Benchmark    : thread_times.simple_mobile_sites
  Metric       : thread_other_cpu_time_per_frame/http___www.nyc.gov
  Change       : 251.18% | 0.234203082222 -> 0.822471404386

Revision                           Result                    N
chromium@514613                    0.234203 +- 0.138476      6       good
chromium@514629                    0.201566 +- 0.143248      8       good
chromium@514637                    0.216675 +- 0.138877      8       good
chromium@514641                    0.194763 +- 0.121816      7       good
chromium@514641,v8@67d543ec31      0.183108 +- 0.106815      8       good
chromium@514641,v8@6bb1d47e6e      0.758455 +- 0.306448      4       bad       <--
chromium@514641,v8@ac0661b358      0.889834 +- 0.330264      4       bad
chromium@514641,v8@80cb266c75      0.791116 +- 0.311281      5       bad
chromium@514642                    0.832264 +- 0.077432      5       bad
chromium@514643                    0.788318 +- 0.837674      11      bad
chromium@514644                    0.806473 +- 0.353438      8       bad
chromium@514675                    0.856388 +- 0.241344      5       bad
chromium@514737                    0.822471 +- 0.103692      6       bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=http...www.nyc.gov thread_times.simple_mobile_sites

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8963323624347189200


For feedback, file a bug with component Speed>Bisection

 Issue 783627  has been merged into this issue.

 Issue 783677  has been merged into this issue.

 Issue 783691  has been merged into this issue.

 Issue 783681  has been merged into this issue.

 Issue 783686  has been merged into this issue.

 Issue 783675  has been merged into this issue.

 Issue 783636  has been merged into this issue.

 Issue 783637  has been merged into this issue.

 Issue 783678  has been merged into this issue.

 Issue 783770  has been merged into this issue.

 Issue 783687  has been merged into this issue.

 Issue 783693  has been merged into this issue.

 Issue 783640  has been merged into this issue.

 Issue 783683  has been merged into this issue.

 Issue 783628  has been merged into this issue.

 Issue 783631  has been merged into this issue.

The regressions are expected since we are using more threads to make faster progress in garbage collection.

In addition to thread_times metrics system_health.common_desktop and blink_perf.canvas also regressed.  Maybe this is an acceptable trade off.

+junov@ and +charliea@ as FYI

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8963033696366404720

The blink_perf.canvas regression is not expected. This CL re-enables a feature that was enabled in September and was recently disabled only for a week.

The blink_perf.canvas graph does not show any impact on the initial enabling and disabling of concurrent marker:
https://chromeperf.appspot.com/report?sid=ef24889183e15962dfe73d5ec262b08d99e383deb02bac064e7f5631493d72e0&start_rev=499423&end_rev=515921
Maybe there was a CL in the meantime that made blink_perf.canvas sensitive to concurrent marking. I'll try to reproduce.

The system_health.common_desktop renderer_eqt_max looks too noisy for the load_chrome_blank page. My CL actually improves this metric (bisect seems confused here):

Bisect Details
  Configuration: win_x64_perf_bisect
  Benchmark    : system_health.common_desktop
  Metric       : total:500ms_window:renderer_eqt_max/load_chrome/load_chrome_blank
  Change       : 51.54% | 86.3194536591 -> 41.8330192458

Revision                           Result                     N
chromium@514531                    86.3195 +- 754.455         14      good
chromium@514597                    9.42487 +- 78.9623         9       good
chromium@514630                    58.0561 +- 491.638         9       good
chromium@514639                    5.03225 +- 41.7015         9       good
chromium@514641                    7.77774 +- 64.9653         9       good
chromium@514641,v8@67d543ec31      13.6998 +- 115.248         9       good
chromium@514641,v8@6bb1d47e6e      0.752041 +- 5.19631        9       bad       <--
chromium@514641,v8@ac0661b358      28.7254 +- 242.572         9       bad
chromium@514641,v8@80cb266c75      32.2633 +- 272.568         9       bad
chromium@514642                    0.141876 +- 0.0171712      9       bad
chromium@514643                    0.141426 +- 0.0292614      9       bad
chromium@514647                    0.13826 +- 0.0128199       9       bad
chromium@514663                    41.833 +- 305.582          14      bad

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

Suspected Commit
  Author : Ulan Degenbaev
  Commit : 6bb1d47e6e82f75ead36cdab209afc5dafa8329c
  Date   : Tue Nov 07 13:11:10 2017
  Subject: [heap] Re-enable concurrent marking.

Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : blink_perf.canvas
  Metric       : getImageData/getImageData
  Change       : 9.30% | 90.2517336806 -> 81.8582134781

Revision                           Result                   N
chromium@514608                    90.2517 +- 0.77908       6      good
chromium@514631                    90.2636 +- 1.32809       6      good
chromium@514637                    88.1371 +- 0.801082      6      good
chromium@514640                    88.8326 +- 1.62346       6      good
chromium@514641                    87.9701 +- 1.95184       6      good
chromium@514641,v8@67d543ec31      88.3932 +- 1.63973       6      good
chromium@514641,v8@6bb1d47e6e      80.5681 +- 1.3339        6      bad       <--
chromium@514641,v8@ac0661b358      80.2101 +- 0.890418      6      bad
chromium@514641,v8@80cb266c75      80.2863 +- 1.40881       6      bad
chromium@514642                    80.7391 +- 1.53681       6      bad
chromium@514653                    81.9496 +- 2.35205       6      bad
chromium@514697                    81.8582 +- 1.63779       6      bad

Please refer to the following doc on diagnosing blink_perf regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/speed/benchmark_harnesses/blink_perf.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.canvas

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8963033696366404720


For feedback, file a bug with component Speed>Bisection

 Issue 787432  has been merged into this issue.

 Issue 787437  has been merged into this issue.