Issue metadata
Sign in to add a comment
|
1%-40% regression in media.tough_video_cases at 459388:459639 |
||||||||||||||||||||
Issue descriptionBuffering 2 seconds shows a mixture of improvements and regressions, this is a tracking bug for all of them to decide if there's anything actionable.
,
Mar 27 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8983944436074237760
,
Mar 27 2017
TL;DR: a surprisingly small increase in memory (1-2 kb, if the units in the dashboard are correct.) idle wakeups go down by a good 15-20% seek time for webm go up by 10-20% The seek time seems bad, and there doesn't seem to be much data to back up any improvements. Maybe I should revert it, add some tests that would show improvements and then try submit it again?
,
Mar 27 2017
Hmm, did you click show improvements at the top right? I saw quite a few.
,
Mar 27 2017
Nope, missed that. Doesn't seem to do anything though. (I follow the "All graphs for this bug" link, and then click that button.) The "Original alerts at time of bug-filing" link doesn't work, I guess it's too long.
,
Mar 27 2017
,
Mar 27 2017
Interesting, summarizing this link: Major improvements for "buffering_time". (20-650%) (Why is this negative??) clockless video playback improves (2-13 %) idle_wakup_renderer improves (30-60%) time to play improves across the board (10-30%) seek improves in more cases than it regresses (-20 - +40%)
,
Mar 28 2017
=== Auto-CCing suspected CL author hubbe@chromium.org === Hi hubbe@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 : hubbe Commit : 83bbe324e42ebe5b1205d6caeb1ea4149bad31f1 Date : Fri Mar 24 21:48:44 2017 Subject: Buffer 2 seconds of data in the ffmpeg demuxer. Bisect Details Configuration: android_nexus7_perf_bisect Benchmark : media.android.tough_video_cases Metric : vm_private_dirty_delta/crowd.wav_renderer Change : 26.98% | 2260.66666667 -> 2870.66666667 Revision Result N chromium@459484 2260.67 +- 66.3124 6 good chromium@459529 2264.67 +- 25.5604 6 good chromium@459551 2249.33 +- 58.697 6 good chromium@459562 2263.33 +- 49.1257 6 good chromium@459564 2267.33 +- 61.0355 6 good chromium@459565 2876.0 +- 70.6541 6 bad <-- chromium@459568 2865.33 +- 57.3178 6 bad chromium@459573 2870.67 +- 59.2396 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 media.android.tough_video_cases Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8983944436074237760 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=5687071785615360 | 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!
,
Apr 11 2017
,
Jul 21 2017
,
Oct 4 2017
|
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by dalecurtis@google.com
, Mar 27 2017