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

Issue 775864 link

Starred by 1 user

Issue metadata

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



Sign in to add a comment

86.3%-136.4% regression in v8.runtimestats.browsing_desktop at 509091:509194

Project Member Reported by jgruber@chromium.org, Oct 18 2017

Issue description

See the link to graphs below.
 
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Oct 18 2017

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=775864

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=d2d438a3f148c2c80c28dbcaf912a38321024ab47ca5962cc983de11a7d4c32b


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

chromium-rel-win7-gpu-ati
chromium-rel-win7-gpu-intel
chromium-rel-win7-x64-dual
chromium-rel-win8-dual
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Oct 18 2017

Cc: mlippautz@chromium.org
Owner: mlippautz@chromium.org
Status: Assigned (was: Untriaged)

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

Hi mlippautz@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 : Michael Lippautz
  Commit : 4b42656dd6fca982c60185c004e2648093d70b51
  Date   : Mon Oct 16 10:00:08 2017
  Subject: [heap] Unify incremental and main marking visitor

Bisect Details
  Configuration: winx64intel_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : v8-gc-incremental-step_max/browse_media/browse_media_youtube
  Change       : 159.13% | 3.795 -> 9.83383333333

Revision                           Result                  N
chromium@509091                    3.795 +- 1.25348        6      good
chromium@509143                    3.94917 +- 2.19726      6      good
chromium@509169                    4.4245 +- 2.06205       6      good
chromium@509182                    4.29883 +- 2.21052      6      good
chromium@509185                    4.4665 +- 2.46505       6      good
chromium@509185,v8@4b42656dd6      9.49183 +- 1.03343      6      bad       <--
chromium@509185,v8@089dd7d244      9.21383 +- 1.52744      6      bad
chromium@509185,v8@5820041884      9.348 +- 2.43702        6      bad
chromium@509185,v8@e659f45610      9.054 +- 2.00339        6      bad
chromium@509185,v8@af49af006f      10.0705 +- 2.66056      6      bad
chromium@509186                    8.99917 +- 1.54169      6      bad
chromium@509187                    10.2245 +- 3.12185      6      bad
chromium@509188                    9.966 +- 1.67223        6      bad
chromium@509194                    9.83383 +- 1.48719      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=browse.media.youtube v8.runtimestats.browsing_desktop

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8965404304501294528


For feedback, file a bug with component Speed>Bisection
Cc: -mlippautz@chromium.org
Components: Blink>JavaScript>GC
Status: Started (was: Assigned)
Project Member

Comment 5 by 42576172...@developer.gserviceaccount.com, Oct 19 2017

 Issue 776296  has been merged into this issue.
Project Member

Comment 6 by bugdroid1@chromium.org, Oct 19 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/v8/v8.git/+/9b51519e52b1280a62a0f337d634773456d89079

commit 9b51519e52b1280a62a0f337d634773456d89079
Author: Michael Lippautz <mlippautz@chromium.org>
Date: Thu Oct 19 16:20:15 2017

[heap] IncrementalMarking: Avoid using atomcis for live byte count

The regression sneaked in when unifying the marking visitors in
4b42656dd6fca982c60185c004e2648093d70b51. The concurrent marker keeps a
local live byte count per page so it is safe to rely on non-atomic
writes from the main thread.

CQ_INCLUDE_TRYBOTS=master.tryserver.v8:v8_linux64_tsan_rel;master.tryserver.v8:v8_linux64_tsan_concurrent_marking_rel_ng;master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

Bug:  chromium:775864 
Change-Id: If67d2accd1d7953aa30d815da8bf41b0039ad2dd
Reviewed-on: https://chromium-review.googlesource.com/728239
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#48758}
[modify] https://crrev.com/9b51519e52b1280a62a0f337d634773456d89079/src/heap/heap.h
[modify] https://crrev.com/9b51519e52b1280a62a0f337d634773456d89079/src/heap/incremental-marking.cc
[modify] https://crrev.com/9b51519e52b1280a62a0f337d634773456d89079/src/heap/mark-compact-inl.h
[modify] https://crrev.com/9b51519e52b1280a62a0f337d634773456d89079/src/heap/mark-compact.cc
[modify] https://crrev.com/9b51519e52b1280a62a0f337d634773456d89079/src/heap/mark-compact.h

Project Member

Comment 7 by 42576172...@developer.gserviceaccount.com, Oct 20 2017

 Issue 776298  has been merged into this issue.
Regression is not yet recovered. Taking another look.
Status: WontFix (was: Started)
Graphs recovered without interfering. We can close it as it is back to the baseline.

Started another bisect to retry on the original culprit.
Project Member

Comment 11 by 42576172...@developer.gserviceaccount.com, Oct 25 2017


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

Suspected Commit
  Author : Michael Lippautz
  Commit : 4b42656dd6fca982c60185c004e2648093d70b51
  Date   : Mon Oct 16 10:00:08 2017
  Subject: [heap] Unify incremental and main marking visitor

Bisect Details
  Configuration: win_8_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : v8-gc-incremental-step_max/browse_media/browse_media_pinterest
  Change       : 25.90% | 5.79466666667 -> 7.29538095238

Revision                           Result                  N
chromium@509080                    5.79467 +- 4.61311      21      good
chromium@509140                    5.57921 +- 3.53991      14      good
chromium@509169                    5.93676 +- 5.07338      21      good
chromium@509184                    5.64114 +- 3.3515       14      good
chromium@509185                    5.76356 +- 2.41248      9       good
chromium@509185,v8@4b42656dd6      7.79767 +- 3.72916      9       bad       <--
chromium@509185,v8@089dd7d244      7.1865 +- 4.05165       14      bad
chromium@509185,v8@5820041884      7.263 +- 2.41047        9       bad
chromium@509185,v8@e659f45610      6.935 +- 3.43999        14      bad
chromium@509185,v8@af49af006f      7.98056 +- 5.82718      9       bad
chromium@509186                    7.18781 +- 6.46181      21      bad
chromium@509188                    7.99886 +- 6.75348      14      bad
chromium@509192                    7.09067 +- 6.4724       21      bad
chromium@509199                    7.29538 +- 6.36879      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=browse.media.pinterest v8.runtimestats.browsing_desktop

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8964870311658036256


For feedback, file a bug with component Speed>Bisection

Sign in to add a comment