New issue
Advanced search Search tips

Issue 664102 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Nov 2016
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 0
Type: Bug-Regression

Blocked on:
issue 664765
issue 616932
issue 664512



Sign in to add a comment

25.1%-25.3% regression in blink_perf.parser at 430486:430557

Project Member Reported by perezju@chromium.org, Nov 10 2016

Issue description

(Also testing new bisect recipe)
 
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Nov 11 2016


===== BISECT JOB RESULTS =====
Status: failed


===== TESTED REVISIONS =====
Revision         Mean     Std Dev   N   Good?
chromium@430485  16.2181  9.03602   25  good
chromium@430521  14.978   7.42545   25  good
chromium@430539  14.9303  7.44399   25  good
chromium@430548  15.9192  7.97257   25  good
chromium@430553  12.1459  1.24524   25  bad
chromium@430557  10.7326  0.901072  25  bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 664102

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.parser
Test Metric: query-selector-all-attribute/query-selector-all-attribute
Relative Change: 33.82%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/4331
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8996382686693440816


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5903368231321600

| 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!
Blocking: 616932
Cc: sullivan@chromium.org
Labels: -Pri-2 Pri-0
Owner: robert...@chromium.org
Status: Assigned (was: Untriaged)
Bisect failed.
Blockedon: 664512
Project Member

Comment 7 by 42576172...@developer.gserviceaccount.com, Nov 12 2016


===== BISECT JOB RESULTS =====
Status: failed


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N   Good?
chromium@430485  15.67    9.89456  25  good
chromium@430521  14.9908  7.65282  25  good
chromium@430539  15.219   6.60582  25  good
chromium@430548  15.4198  7.63327  25  good
chromium@430553  12.2283  1.39446  25  bad
chromium@430557  10.7464  0.78597  25  bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 664102

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.parser
Test Metric: query-selector-all-attribute/query-selector-all-attribute
Relative Change: 31.42%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/4334
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8996215233985005552


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5229426390859776

| 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!
Blockedon: 616932
Blocking: -616932
Project Member

Comment 11 by 42576172...@developer.gserviceaccount.com, Nov 14 2016


===== BISECT JOB RESULTS =====
Status: failed


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N   Good?
chromium@430485  16.1704  8.41239  25  good
chromium@430521  14.7542  7.57002  25  good
chromium@430539  15.4631  7.13546  25  good
chromium@430548  15.2793  7.53196  25  good
chromium@430553  12.1027  1.09366  25  bad
chromium@430557  10.7133  1.02192  25  bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 664102

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.parser
Test Metric: query-selector-all-attribute-complex/query-selector-all-attribute-complex
Relative Change: 33.75%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/4335
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8996046410849454448


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5841171333513216

| 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!
Blockedon: 664765
Project Member

Comment 13 by 42576172...@developer.gserviceaccount.com, Nov 14 2016

Cc: vogelheim@chromium.org
Owner: vogelheim@chromium.org

=== Auto-CCing suspected CL author vogelheim@chromium.org ===

Hi vogelheim@chromium.org, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


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


===== SUSPECTED CL(s) =====
Subject : [CachedAccessor] for window.document.
Author  : vogelheim
Commit description:
  
- Implemend [CachedAccessor] attribute to enable cached accessors.
- Implement a cached accessor for window.document.

This is based on crrev.com/2335203006 by peterssen@google.com, with
only a minor fix added.

Intent to implement & ship:
https://groups.google.com/a/chromium.org/forum/#!topic/blink-dev/OhIFnre7ytQ

Previous (reviewed) CL:
https://codereview.chromium.org/2335203006/

Depends on V8 CL:
https://codereview.chromium.org/2405213002/

BUG=chromium:634276

Review-Url: https://codereview.chromium.org/2429343004
Cr-Commit-Position: refs/heads/master@{#430551}
Commit  : 460ec9e9068eb8bf57ce3fae68bb277ca4106f9a
Date    : Tue Nov 08 09:18:55 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N   Good?
chromium@430548  16.1762  7.77477  25  good
chromium@430550  16.9649  9.16329  25  good
chromium@430551  12.1172  1.04847  25  bad    <--
chromium@430553  12.1317  1.0854   25  bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 664102

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.parser
Test Metric: query-selector-all-attribute/query-selector-all-attribute
Relative Change: 25.00%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/4336
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8996013812381474944


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5839146575200256

| 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!
I've look at this, but am very unsure what to make it of it.

- The graphs are rather convincing, so there's something up with my CL.
- The CL itself is generally fine, so wouldn't want to revert it as-is.
- The observed performance drop puzzles me:
  - It's only on Nexus 5. It's *not* on Nexus 10 or other Androids.
  - The affected tests indeed use the document property, but only peripherally.
  - There are a number of other tests is the query-selector-*/* range that use
    the document property in exactly the same way, but aren't affected, even
    on Nexus 5.

I'm at a loss what would be specific to this setup.

I'll look into this a bit more, but am tempted to 'WontFix'. Either way, I think priority 0/Emergency is quite exaggerated.


@perezju/sullivan: Please let me know if you disagree on priority.
Status: Fixed (was: Assigned)
I'm so sorry about this, Daniel! The P0 was for the bisect failures, not for the actual performance test regression. I think we can actually mark "Fixed" because the bisect works correctly now.

I agree the actual regression is minor.
Project Member

Comment 17 by 42576172...@developer.gserviceaccount.com, Nov 22 2016


===== BISECT JOB RESULTS =====
Status: failed


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N   Good?
chromium@430502  76.8899  5.24566  25  good
chromium@430530  73.432   3.34807  25  good
chromium@430544  77.6142  3.81502  25  good
chromium@430548  77.277   8.53103  25  good
chromium@430551  66.2582  3.36291  25  bad
chromium@430558  65.9379  4.85307  25  bad

Bisect job ran on: android_nexus7_perf_bisect
Bug ID: 664102

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.parser
Test Metric: query-selector-last/query-selector-last
Relative Change: 14.24%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus7_perf_bisect/builds/3509
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8995299122867384592


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5320735902400512

| 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!
Project Member

Comment 19 by 42576172...@developer.gserviceaccount.com, Apr 11 2017


=== BISECT JOB RESULTS ===
Perf regression found but unable to narrow commit range

Build failures prevented the bisect from narrowing the range further.


Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : blink_perf.parser
  Metric       : query-selector-all-attribute/query-selector-all-attribute

Suspected Commit Range
  1 commits in range
  https://chromium.googlesource.com/chromium/src/+log/8a65ad68ecff0e06b76c412c8ad539053b21dc81..e62f1a5d8886fc1086a180f4da94e0a3d1ca03dc


Revision             Result                  N
chromium@430485      15.67 +- 9.89456        25      good
chromium@430521      14.9908 +- 7.65282      25      good
chromium@430539      15.219 +- 6.60582       25      good
chromium@430548      15.4198 +- 7.63327      25      good
chromium@430553      12.2283 +- 1.39446      25      bad
chromium@430557      10.7464 +- 0.78597      25      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.parser

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8982651612202801680

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5229426390859776


| 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!
Project Member

Comment 21 by 42576172...@developer.gserviceaccount.com, Apr 11 2017


=== BISECT JOB RESULTS ===
Perf regression found but unable to narrow commit range

Build failures prevented the bisect from narrowing the range further.


Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : blink_perf.parser
  Metric       : query-selector-last/query-selector-last

Suspected Commit Range
  1 commits in range
  https://chromium.googlesource.com/chromium/src/+log/8a65ad68ecff0e06b76c412c8ad539053b21dc81..460ec9e9068eb8bf57ce3fae68bb277ca4106f9a


Revision             Result                  N
chromium@430502      76.8899 +- 5.24566      25      good
chromium@430530      73.432 +- 3.34807       25      good
chromium@430544      77.6142 +- 3.81502      25      good
chromium@430548      77.277 +- 8.53103       25      good
chromium@430551      66.2582 +- 3.36291      25      bad
chromium@430558      65.9379 +- 4.85307      25      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.parser

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8982620147188801008

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5320735902400512


| 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