Issue metadata
Sign in to add a comment
|
1.1%-4.7% regression in system_health.memory_mobile at 449944:449992 |
||||||||||||||||||||||||
Issue descriptionSee the link to graphs below.
,
Feb 14 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8987660713847356752
,
Feb 14 2017
=== Auto-CCing suspected CL author fdoray@chromium.org === Hi fdoray@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 : fdoray Commit : b83be4ca0e00b50b17619adf7f7de275455e9852 Date : Mon Feb 13 16:39:47 2017 Subject: Use TaskScheduler instead of WorkerPool in v8_platform.cc. Bisect Details Configuration: android_webview_arm64_aosp_perf_bisect Benchmark : memory.top_10_mobile Metric : memory:webview:all_processes:reported_by_chrome:malloc:effective_size_avg/background/after_https_www_google_co_uk_hl_en_q_science Change : 3.30% | 34661116.0 -> 35805824.0 Revision Result N chromium@449959 34661116 +- 564034 6 good chromium@449969 34501232 +- 1212175 6 good chromium@449974 34480167 +- 592920 6 good chromium@449975 34912848 +- 1233545 6 good chromium@449976 36041641 +- 586473 6 bad <-- chromium@449978 35805824 +- 606964 6 bad Please refer to the following doc on diagnosing memory regressions: https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md To Run This Test src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests memory.top_10_mobile Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8987660713847356752 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=5850907700363264 | 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!
,
Feb 16 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8987495667332242384
,
Feb 16 2017
=== BISECT JOB RESULTS === Perf regression found with culprit Suspected Commit Author : fdoray Commit : b83be4ca0e00b50b17619adf7f7de275455e9852 Date : Mon Feb 13 16:39:47 2017 Subject: Use TaskScheduler instead of WorkerPool in v8_platform.cc. Bisect Details Configuration: android_nexus5X_perf_bisect Benchmark : memory.top_10_mobile Metric : memory:chrome:all_processes:reported_by_os:system_memory:native_heap:proportional_resident_size_avg/background/after_http_www_amazon_com_gp_aw_s_k_nexus Change : 2.59% | 54625644.4444 -> 56072890.2857 Revision Result N chromium@449949 54625644 +- 2364191 9 good chromium@449965 54227080 +- 3173550 6 good chromium@449973 54635560 +- 1899904 14 good chromium@449975 54150048 +- 2538330 9 good chromium@449976 56184627 +- 1325709 6 bad <-- chromium@449980 56072890 +- 1731211 14 bad Please refer to the following doc on diagnosing memory regressions: https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md 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 memory.top_10_mobile Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8987495667332242384 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=5619227769176064 | 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 2 2017
This is a regression we care about and need to respond to - either fix the regression or justify - as it is blocking our ability to release Chrome 58 to Android.
,
Mar 2 2017
The CL was reverted 4df731b9954e2973ae8e206185f3de4556a806be r452058. These graphs recovered in the revert range: https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg6M-WvQgM https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg2Kaa6QgM (recovery is bigger than regression so there is probably an independent memory improvement involved) https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg2MT_7wgM This graph clearly didn't recover: https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguIG7owkM Could there be another regression in the 449921 - 449966 http://test-results.appspot.com/revision_range?start=449921&end=449966 range? In other graphs, the recovery seems to be hidden by noise.
,
Mar 2 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8986195071413798528
,
Mar 3 2017
=== BISECT JOB RESULTS === Perf regression found with culprit Suspected Commit Author : karlo Commit : 934becac5daa91ea979fb66e4ae21761ca11ebc9 Date : Mon Feb 13 13:52:07 2017 Subject: Support subpixel layout of borders. Bisect Details Configuration: win_perf_bisect Benchmark : blink_perf.layout Metric : large-table-with-collapsed-borders-and-no-colspans/large-table-with-collapsed-borders-and-no-colspans Change : 20.06% | 248.107166667 -> 297.875166667 Revision Result N chromium@449920 248.107 +- 3.08346 6 good chromium@449932 259.389 +- 1.7328 6 good chromium@449938 251.259 +- 3.98985 6 good chromium@449941 250.432 +- 3.90422 6 good chromium@449942 251.006 +- 3.22649 6 good chromium@449943 296.563 +- 1.87949 6 bad <-- chromium@449966 297.875 +- 4.72946 6 bad To Run This Test src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.layout Debug Info https://chromeperf.appspot.com/buildbucket_job_status/8986195071413798528 Is this bisect wrong? https://chromeperf.appspot.com/bad_bisect?try_job_id=5894256025141248 | 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 11 2017
|
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by kouhei@google.com
, Feb 14 2017