New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 670391 link

Starred by 1 user

Issue metadata

Status: Duplicate
Merged: issue 670352
Owner: ----
Closed: Dec 2016
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression

Blocked on:
issue 664760



Sign in to add a comment

21.5%-31.9% regression in thread_times.key_idle_power_cases at 434896:434955

Project Member Reported by lanwei@chromium.org, Dec 1 2016

Issue description

See the link to graphs below.
 

===== BISECT JOB RESULTS =====
Status: completed


===== TESTED REVISIONS =====
Revision         Mean      Std Dev     N   Good?
chromium@434902  0.234643  0.00822806  18  good
chromium@434915  0.234511  0.00887995  18  good
chromium@434916  0.235178  0.0105796   27  unknown
chromium@434917  0.235633  0.0106545   27  unknown
chromium@434918  0.235637  0.0123628   27  unknown
chromium@434919  0.236463  0.0102951   27  bad
chromium@434920  0.236391  0.0122314   27  bad
chromium@434923  0.236824  0.00966663  18  bad
chromium@434929  0.236459  0.00971058  18  bad
chromium@434955  0.30617   0.00559618  5   bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 670391

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=animated.gif.html thread_times.key_idle_power_cases
Test Metric: thread_total_all_cpu_time_per_second/animated-gif.html
Relative Change: 30.46%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/4394
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8994450483567086784


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5259541669740544

| 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 Tests>AutoBisect.  Thank you!
Cc: robert...@chromium.org dtu@chromium.org
This bisect doesn't seem like it went the way you'd expect. At chromium@434929, this gets marked as BAD but the data is way way closer to the good value than the bad. The compare data is:


Comparison vs Good
{
  "result": {
    "U": 1, 
    "p": 0.0067665000350596305, 
    "significance": "REJECT"
  }, 
  "sampleA": [
    0.23864875337000382, 
    0.23588109715440517, 
    0.2397168113315397, 
    0.2370708916008479, 
    0.2366153342733322
  ], 
  "sampleB": [
    0.2356097441879304, 
    0.23381916180783394, 
    0.23551152998467123, 
    0.231928238251592, 
    0.23284523840075097, 
    0.23660063598123102, 
    0.2356152545836532, 
    0.23548959217386697
  ]
}

Comparison vs Bad:
{
  "result": {
    "U": 0, 
    "p": 0.012185780355344789, 
    "significance": "NEED_MORE_DATA"
  }, 
  "sampleA": [
    0.23864875337000382, 
    0.23588109715440517, 
    0.2397168113315397, 
    0.2370708916008479, 
    0.2366153342733322
  ], 
  "sampleB": [
    0.3045929957541189, 
    0.30318833300339465, 
    0.310615011131698, 
    0.306522739171334, 
    0.30593039723967996
  ]
}



+dtu, robertocn for ideas on what went wrong here
Again, raising min reps should fix this case.
Blockedon: 664760
Kicked a bisect on the 5x which had a very clean step regression: https://chromeperf.appspot.com/buildbucket_job_status/8993377217001104800
Project Member

Comment 9 by 42576172...@developer.gserviceaccount.com, Dec 13 2016

Mergedinto: 670352
Status: Duplicate (was: Untriaged)

===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : binding: Makes non-cross-origin-accessible attrs be accessor props.
Author  : yukishiino
Commit description:
  
Makes non-cross-origin accessible attributes be accessor
properties.

[CachedAccessor] already made window.document an accessor
property, and this CL doesn't affect it.

Cross-origin accessible attributes remain being data
properties.

The previous attempt was http://crrev.com/1380503002

BUG=475556

Review-Url: https://codereview.chromium.org/2506393004
Cr-Commit-Position: refs/heads/master@{#434942}
Commit  : 5b2adfa4d4e2fb6b74d177625605dcece20a40b7
Date    : Tue Nov 29 09:53:57 2016


===== TESTED REVISIONS =====
Revision         Mean      Std Dev     N  Good?
chromium@434895  0.26358   0.0059097   6  good
chromium@434925  0.264424  0.00548704  6  good
chromium@434940  0.262049  0.00500986  6  good
chromium@434941  0.262975  0.00840568  6  good
chromium@434942  0.334129  0.00737958  6  bad    <--
chromium@434944  0.334014  0.00636412  6  bad
chromium@434948  0.337905  0.00533505  6  bad
chromium@434955  0.337451  0.00802515  6  bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 670391

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=blank.html thread_times.key_idle_power_cases
Test Metric: thread_total_all_cpu_time_per_second/blank.html
Relative Change: 28.03%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/4453
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993377217001104800


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4992723638026240

| 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 Tests>AutoBisect.  Thank you!

Sign in to add a comment