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

Issue 748118 link

Starred by 1 user
Status: Fixed
Owner:
rsesek@chromium.org
Closed: Aug 2017
Cc:
█ michaeln@google.com
jrumm...@chromium.org
mark@chromium.org
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

17.6% regression in media_perftests at 488677:488777

Project Member Reported by jrumm...@chromium.org, Jul 24 2017 
It improved not that long ago (486361:486451), and is now back to the prior values. Running a bisect to make sure the improvement made previously (if there is one) hasn't been reverted.
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Jul 24 2017 

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=748118

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=1c977542e4aa957fe28e5ba448e8679d0f72b1f73b8ca50414f4eae888659ed6


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

chromium-rel-mac11
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Jul 24 2017

Cc: michaeln@google.com
Owner: michaeln@google.com

=== Auto-CCing suspected CL author michaeln@google.com ===

Hi michaeln@google.com, 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 Nordman
  Commit : 20b379ba1a5ce5ae093386c7ffe764d63246b66f
  Date   : Fri Jul 21 18:08:09 2017
  Subject: Revert "Reland "Reimplement base::WaitableEvent with a kqueue on Mac.""

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : media_perftests
  Metric       : demuxer_bench/sfx_s16le.wav
  Change       : 17.36% | 2624.99003119 -> 2169.41442181

Revision             Result                  N
chromium@488676      2624.99 +- 192.928      6      good
chromium@488702      2630.04 +- 90.2774      6      good
chromium@488704      2655.61 +- 56.1691      6      good
chromium@488705      2642.07 +- 99.8238      6      good
chromium@488706      2177.23 +- 75.4582      6      bad       <--
chromium@488709      2228.07 +- 133.945      6      bad
chromium@488715      2176.49 +- 40.8837      6      bad
chromium@488727      2172.58 +- 104.276      6      bad
chromium@488777      2169.41 +- 80.1908      6      bad

To Run This Test
  ./src/out/Release/media_perftests --single-process-tests

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8973165386288394800


For feedback, file a bug with component Speed>Bisection

Comment 4 by jrumm...@chromium.org, Jul 24 2017 

 Issue 748092  has been merged into this issue.
 Issue 748094  has been merged into this issue.
 Issue 748095  has been merged into this issue.
Project Member

Comment 5 by 42576172...@developer.gserviceaccount.com, Jul 25 2017 

 Issue 748278  has been merged into this issue.

Comment 6 by michaeln@chromium.org, Jul 26 2017

Cc: mark@chromium.org
Owner: rsesek@chromium.org
Bouncing to rsesek@, i reverted the WaitableEvent change because it was somehow responsible for an increase in crashes (see crbug/744550).

Comment 7 by rsesek@chromium.org, Aug 11 2017

Status: Fixed (was: Untriaged)
Went back down after d54c0c1eccbddf90ea69bd14ca1290d721eee9b9
►

Sign in to add a comment