Issue metadata
Sign in to add a comment
|
10.1% regression in blink_perf.paint at 455819:455885 |
||||||||||||||||||||||||
Issue descriptionSee the link to graphs below.
,
Mar 13 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8985217302780849568
,
Mar 14 2017
=== Auto-CCing suspected CL author chrishtr@chromium.org === Hi chrishtr@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 : chrishtr Commit : 890521e5189ff65fb2c16f67c837cc09bcf503c9 Date : Thu Mar 09 20:22:24 2017 Subject: Reland: Improve performance of GeometryMapper cache. Bisect Details Configuration: android_nexus5_perf_bisect Benchmark : blink_perf.paint Metric : paint-offset-changes/paint-offset-changes Change : 6.45% | 2176.35083333 -> 2316.76758333 Revision Result N chromium@455818 2176.35 +- 19.9332 6 good chromium@455835 2220.55 +- 48.1508 6 good chromium@455844 2305.7 +- 82.5892 6 good chromium@455848 2314.99 +- 32.921 6 good chromium@455849 2296.64 +- 92.4975 6 good chromium@455850 2386.25 +- 41.994 6 bad <-- chromium@455852 2397.5 +- 22.6628 6 bad chromium@455885 2316.77 +- 90.2297 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 blink_perf.paint Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8985217302780849568 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=4932625631805440 | 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!
,
Mar 16 2017
Original land of my CL was at 455571.
,
Mar 16 2017
I don't think my patch was the cause of this bug, because the graph shows no regression associated with 455571, and also no positive change when it was reverted.
,
Mar 16 2017
Also, this graph improved quite a bit at revision 453070, presumably due to https://codereview.chromium.org/2707063003.
,
Mar 16 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8984928478939718272
,
Mar 17 2017
,
Mar 17 2017
=== BISECT JOB RESULTS === Perf regression found with culprit Suspected Commit Author : pdr Commit : ce1b788de6382f8c3725efe1fb14e9d4c639b5ab Date : Thu Mar 09 04:36:32 2017 Subject: Heap allocate PrePaintTreeWalk context object Bisect Details Configuration: android_nexus5_perf_bisect Benchmark : blink_perf.paint Metric : paint-offset-changes/paint-offset-changes Change : 14.11% | 2017.57 -> 2302.15641667 Revision Result N chromium@455589 2017.57 +- 33.4543 6 good chromium@455641 2032.84 +- 26.5504 6 good chromium@455667 2020.55 +- 48.5291 6 good chromium@455668 2199.12 +- 25.6653 6 bad <-- chromium@455669 2196.72 +- 21.794 6 bad chromium@455671 2188.55 +- 24.2253 6 bad chromium@455675 2184.71 +- 37.4305 6 bad chromium@455680 2190.33 +- 35.6031 6 bad chromium@455692 2243.22 +- 56.9988 6 bad chromium@455795 2215.39 +- 66.9694 6 bad chromium@456000 2302.16 +- 50.6021 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 blink_perf.paint Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8984928478939718272 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=6355033303547904 | 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 |
|||||||||||||||||||||||||
Comment 1 by hjd@chromium.org
, Mar 13 2017