New issue
Advanced search Search tips

Issue 732988 link

Starred by 1 user
Status: Duplicate
Merged: issue 732289
Owner:
thakis@chromium.org
Closed: Jun 2017
Cc:
dalecur...@chromium.org
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 2
Type: Bug-Regression



Sign in to add a comment

12.7% regression in media.android.tough_video_cases at 478517:478521

Project Member Reported by dalecur...@chromium.org, Jun 13 2017 
Seems android specific, so guessing this has to do with the gcc -> clang switch over which is in the range. I'd guess we're losing some ffmpeg specific decoder optimizations somehow.

Initially => thakis since he handled the roll, but probably media/ team will have to dig into see why ffmpeg performance changed so much. The time_to_play, and dropped frames metrics look a little noisy, but the buffering_time one is a clear delta.

https://chromeperf.appspot.com/report?sid=8295dac679e36ee2bd10fff79695d4e63f5bdaa66f48510cfa2579e1c6270675&rev=478520

Comment 1 by dalecur...@chromium.org, Jun 13 2017 

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=732988

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg1unk_QoM


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

android-nexus7v2

Comment 3 by dalecur...@chromium.org, Jun 13 2017

Labels: OS-Android
Project Member

Comment 4 by 42576172...@developer.gserviceaccount.com, Jun 14 2017

Mergedinto: 732289
Status: Duplicate (was: Assigned)

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

Suspected Commit
  Author : thakis
  Commit : f823f843f06636feb082521d76c282a16fe2d577
  Date   : Sat Jun 10 17:33:46 2017
  Subject: android: switch from gcc to clang.

Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : media.android.tough_video_cases
  Metric       : time_to_play/video.html?src_crowd1080_vp9.webm
  Change       : 12.51% | 643.807222222 -> 724.328333333

Revision             Result                  N
chromium@478516      643.807 +- 124.732      9      good
chromium@478517      729.826 +- 53.1128      6      bad       <--
chromium@478518      725.413 +- 27.2981      6      bad
chromium@478519      719.565 +- 86.5066      6      bad
chromium@478521      724.328 +- 54.5203      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=video.html.src.crowd1080.vp9.webm media.android.tough_video_cases

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8976863774373848368

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4561149686710272


| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Speed>Bisection.  Thank you!
►

Sign in to add a comment