Issue metadata
Sign in to add a comment
|
181.7% regression in thread_times.simple_mobile_sites at 481163:481215 |
||||||||||||||||||||
Issue descriptionSee the link to graphs below.
,
Jun 23 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8976006458571388592
,
Jun 23 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8976006451913713632
,
Jun 23 2017
=== Auto-CCing suspected CL author dtapuska@chromium.org === Hi dtapuska@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 : Dave Tapuska Commit : 4cef659a8009c5cb4f5708336ae37654d23653b3 Date : Wed Jun 21 15:16:45 2017 Subject: Enable scrollTopLeftInterop by default. Bisect Details Configuration: android_one_perf_bisect Benchmark : thread_times.simple_mobile_sites Metric : thread_raster_cpu_time_per_frame/thread_raster_cpu_time_per_frame Change : 181.77% | 1.13792328389 -> 3.20628471763 Revision Result N chromium@481162 1.13792 +- 0.103006 6 good chromium@481189 1.43585 +- 0.822091 6 good chromium@481202 1.26899 +- 0.694108 5 good chromium@481206 1.14445 +- 0.0280497 6 good chromium@481207 1.15484 +- 0.0397196 6 good chromium@481208 3.24605 +- 0.438329 6 bad <-- chromium@481209 3.60326 +- 0.970508 6 bad chromium@481215 3.20628 +- 0.106803 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 thread_times.simple_mobile_sites Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8976006451913713632 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=5325038057684992 | 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!
,
Jun 23 2017
=== BISECT JOB RESULTS === Perf regression found with culprit Suspected Commit Author : Dave Tapuska Commit : 4cef659a8009c5cb4f5708336ae37654d23653b3 Date : Wed Jun 21 15:16:45 2017 Subject: Enable scrollTopLeftInterop by default. Bisect Details Configuration: android_one_perf_bisect Benchmark : thread_times.simple_mobile_sites Metric : thread_raster_cpu_time_per_frame/thread_raster_cpu_time_per_frame Change : 175.18% | 1.11392461448 -> 3.06531289294 Revision Result N chromium@481162 1.11392 +- 0.0196522 6 good chromium@481189 1.11612 +- 0.0362004 6 good chromium@481202 1.10759 +- 0.125939 6 good chromium@481206 1.11063 +- 0.065558 6 good chromium@481207 1.35673 +- 0.862452 6 good chromium@481208 3.34355 +- 0.970331 6 bad <-- chromium@481209 3.38808 +- 0.543608 6 bad chromium@481215 3.06531 +- 0.370308 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 thread_times.simple_mobile_sites Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8976006458571388592 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=5449898796777472 | 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!
,
Jun 23 2017
Issue 736277 has been merged into this issue.
,
Jun 23 2017
Issue 736277 has been merged into this issue.
,
Jun 26 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/2906e9a3ebf242d7c4723377cedeb5557de949bc commit 2906e9a3ebf242d7c4723377cedeb5557de949bc Author: Dave Tapuska <dtapuska@chromium.org> Date: Mon Jun 26 14:03:21 2017 Remove http://www.apple.com/mac from simple_mobile_sites The page is behaving weird with the ScrollLeftTopInterOp. An analysis of the WPR snapshot shows it is using scrollLeft/scrollTop in a different fashion than the current version. The increased rasterization on low end Android hardware and decreased raster times on mid range Android hardware indicate increased software rasterization is causing the regression. It isn't clear what the cause from the javascript is. There appears to be no visual changes when running the tests. There are exceptions thrown when the page is loaded in Nexus5X devtools console. I believe this page needs to be re-recorded. BUG= 736272 Change-Id: I8c9b9e2bb723acb6686c9b2ad82bb2a0878fda04 Reviewed-on: https://chromium-review.googlesource.com/546297 Reviewed-by: Ned Nguyen <nednguyen@google.com> Commit-Queue: Dave Tapuska <dtapuska@chromium.org> Cr-Commit-Position: refs/heads/master@{#482261} [modify] https://crrev.com/2906e9a3ebf242d7c4723377cedeb5557de949bc/tools/perf/page_sets/simple_mobile_sites.py
,
Jul 12 2017
|
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by kraynov@chromium.org
, Jun 23 2017