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

Issue 713114 link

Starred by 1 user

Issue metadata

Status: WontFix
Owner:
Closed: Aug 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: Mac
Pri: 1
Type: Bug-Regression

Blocked on:
issue 713153



Sign in to add a comment

2.6%-44.9% regression in system_health.common_desktop at 463520:463939

Project Member Reported by alexclarke@chromium.org, Apr 19 2017

Issue description

See the link to graphs below.
 
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Apr 19 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: mac_pro_perf_bisect
  Benchmark    : system_health.common_desktop
  Metric       : timeToFirstMeaningfulPaint_avg/load_search/load_search_amazon


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.search.amazon system_health.common_desktop

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981888053706168768

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5798384163618816


| 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!
Blockedon: 713153
Project Member

Comment 6 by 42576172...@developer.gserviceaccount.com, Apr 19 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: mac_pro_perf_bisect
  Benchmark    : system_health.common_desktop
  Metric       : timeToFirstMeaningfulPaint_avg/load_search/load_search_amazon


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.search.amazon system_health.common_desktop

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981884365727554400

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5798384163618816


| 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 8 by 42576172...@developer.gserviceaccount.com, Apr 20 2017

Cc: samans@chromium.org
Owner: samans@chromium.org

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

Hi samans@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 : samans
  Commit : 2040988b1345e6ce738bd03ab7ccd5f4b73a956b
  Date   : Tue Apr 11 23:58:35 2017
  Subject: Use MojoCompositorFrameSink in RendererCompositorFrameSink

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : system_health.common_desktop
  Metric       : timeToFirstMeaningfulPaint_avg/load_search/load_search_amazon
  Change       : 29.79% | 284.604857144 -> 369.381777777

Revision             Result                  N
chromium@463726      284.605 +- 204.853      14      good
chromium@463833      287.265 +- 215.799      14      good
chromium@463840      288.258 +- 287.867      21      good
chromium@463844      268.282 +- 185.723      14      good
chromium@463845      362.558 +- 49.8569      6       bad       <--
chromium@463846      357.905 +- 72.2337      14      bad
chromium@463847      370.153 +- 114.55       14      bad
chromium@463860      364.71 +- 53.7142       6       bad
chromium@463886      364.041 +- 67.4524      9       bad
chromium@463939      369.382 +- 70.8859      9       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 --story-filter=load.search.amazon system_health.common_desktop

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981843579626372752

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6164589449838592


| 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 10 by 42576172...@developer.gserviceaccount.com, Apr 20 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: mac_retina_perf_bisect
  Benchmark    : system_health.common_desktop
  Metric       : after_load:power_avg/load_social/load_social_vk


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.social.vk system_health.common_desktop

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981785652682379504

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5798384163618816


| 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!
Labels: -Pri-2 OS-Mac Pri-1
Status: Assigned (was: Untriaged)
samans - please either justify, fix your regression or revert your CL.
I'm trying to understand what's going on. Not sure if reverting is a good option. There are metrics that improved as a result of my CL. For example, on the same bot (chromium-rel-mac11) and same test (system_health.common_desktop) but a different subtest (load_media vs load_social), the same metric (timeToFirstMeaningfulPaint_avg) improved 57% (See the link below). So basically the revert can cause regression. Not sure if the metric is broken or things legitimately improved/regressed. I'll keep you posted.
https://bugs.chromium.org/p/chromium/issues/detail?id=713716
We should not trade big regressions in some cases for big speedups in others, at least not without a good reason.
Project Member

Comment 15 by 42576172...@developer.gserviceaccount.com, Apr 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found

Bisect Details
  Configuration: mac_retina_perf_bisect
  Benchmark    : system_health.common_desktop
  Metric       : after_load:power_avg/load_social/load_social_vk

Revision             Result                  N
chromium@463519      14.3637 +- 3.57519      21      good
chromium@463909      14.4562 +- 5.65395      21      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 --story-filter=load.social.vk system_health.common_desktop

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981142590678512320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5798384163618816


| 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!
We had some discussion on another bug that might also apply here. We suspect that some small variation in delivery of BeginFrames can sometimes cause big changes in scheduling tasks in the renderer and some metrics can randomly get better or worse. Unless things get worse on average, I don't believe we should be worried. load_search_amazon is worse, but some other metrics are doing better.

10.10% improvement in load_social_twitter:  http://crbug.com/715767 
52.3% improvement in startup.warm.blank_page:  http://crbug.com/715831 
38.60% improvement in load_media_dailymotion:  http://crbug.com/713716 
29.71% improvement in load_tools_stackoverflow: http://crbug.com/711856
68.1% improvement in startup.cold.blank_page:  http://crbug.com/715766 
Status: WontFix (was: Assigned)
As I explained before this is just noise so marking as WontFix.

Sign in to add a comment