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

Issue 673404 link

Starred by 1 user

Issue metadata

Status: WontFix
Owner:
Last visit > 30 days ago
Closed: Jun 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression

Blocked on:
issue 669608



Sign in to add a comment

9.2% regression in media.tough_video_cases at 437363:437418

Project Member Reported by hubbe@chromium.org, Dec 12 2016

Issue description

See the link to graphs below.
 

Comment 1 by hubbe@chromium.org, Dec 12 2016

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=673404

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgt7CrvAsM


Bot(s) for this bug's original alert(s):

chromium-rel-mac11
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Dec 12 2016


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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: Bisect failed to reproduce the regression with enough confidence.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N    Good?
chromium@437362  671.8    8911.98  156  good
chromium@437418  743.611  9535.98  156  bad

Bisect job ran on: mac_10_11_perf_bisect
Bug ID: 673404

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests media.tough_video_cases
Test Metric: cputime/cputime
Relative Change: 10.69%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_10_11_perf_bisect/builds/1114
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993457878542857136


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5800954207666176

| 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!

Comment 4 by hubbe@chromium.org, Dec 13 2016

Owner: hubbe@chromium.org
Project Member

Comment 6 by 42576172...@developer.gserviceaccount.com, Dec 13 2016


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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: Bisect failed to reproduce the regression with enough confidence.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N    Good?
chromium@437362  677.517  8984.26  156  good
chromium@437418  740.549  9344.46  156  bad

Bisect job ran on: mac_10_11_perf_bisect
Bug ID: 673404

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests media.tough_video_cases
Test Metric: cputime/cputime
Relative Change: 9.30%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_10_11_perf_bisect/builds/1120
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993435353447943952


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5820690253479936

| 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 8 by 42576172...@developer.gserviceaccount.com, Dec 14 2016


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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: Bisect failed to reproduce the regression with enough confidence.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N    Good?
chromium@437362  688.681  9016.64  156  good
chromium@437418  742.494  9375.68  156  bad

Bisect job ran on: mac_10_11_perf_bisect
Bug ID: 673404

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests media.tough_video_cases
Test Metric: cputime/cputime
Relative Change: 7.81%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_10_11_perf_bisect/builds/1126
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993275310238875088


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5884225486585856

| 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: simonhatch@chromium.org crouleau@chromium.org
Simon,

Is there anything we can do at this point to figure out what caused the regression?
Blockedon: 669608
Looks like the bisect is actually reproducing this quite reliably, but we're failing the MWU test. I have a bug out to try averaging the values, in the same way the dashboard does, just haven't had time to get to it. I'll block on that for now.
For anyone reading this later: MWU = https://en.wikipedia.org/wiki/Mann%E2%80%93Whitney_U_test

Simon,

Is the loop here https://cs.chromium.org/chromium/src/tools/auto_bisect/bisect_perf_regression.py?l=2365 succeeding in finding a possible culprit but then later on a test of confidence in that culprit is failing?

Would it be possible to print the suspected culprit to logs that way we could look it up and see what we think?

Also, why does the TESTED REVISIONS list above (comment #8) only list two tested revisions? It seems like it should have tested a full bisect's worth of revisions. I might be helpful to share that data so that developers could analyze it and understand what was happening.
re: #c11

It does print the suspected culprit, or even a range, if it has one to print. In this case, the bisect tested the first/last revisions, but it wasn't confident there was enough of a difference in the values from those revisions and aborted. Kinda wanna try the fix suggested in  crbug.com/669608  to see if it would help, might have some time this week or next depending on what takes priority. I'd be happy to chat offline if you still have any questions.
Project Member

Comment 14 by 42576172...@developer.gserviceaccount.com, Jan 10 2017


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

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : media.tough_video_cases
  Metric       : cputime/cputime

Revision             Result                  N
chromium@437362      681.524 +- 9141.99      156      good
chromium@437418      740.906 +- 9372.33      156      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 media.tough_video_cases

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8990895199863938368

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5227323898462208


| 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 17 by 42576172...@developer.gserviceaccount.com, Jan 13 2017


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

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : media.tough_video_cases
  Metric       : cputime/cputime

Suspected Commit Range
  16 commits in range
  https://chromium.googlesource.com/chromium/src/+log/e433914989869a14fe15f6369fd3149a1b3b7eb8..5c49d183f4713e682965e831387a1f4bdac32034


Revision             Result                  N
chromium@437362      682.049 +- 17.1962      6        good
chromium@437379      684.781 +- 9.18838      6        good
---                  ---                     ---      too many build failures to list
chromium@437394      ---                     ---      build failure
chromium@437395      738.345 +- 9.97353      6        bad
chromium@437418      741.742 +- 10.4483      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 media.tough_video_cases

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8990641799651861520

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5548471890739200


| 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, Jan 14 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: mac_10_11_perf_bisect
  Benchmark    : media.tough_video_cases
  Metric       : cputime/cputime
  Change       : 9.85% | 674.161241282 -> 740.561521603

Suspected Commit Range
  15 commits in range
  https://chromium.googlesource.com/chromium/src/+log/f32ade75b611e31c6f43914a93c3f7067bcc8cce..5c49d183f4713e682965e831387a1f4bdac32034


Revision             Result                  N
chromium@437362      674.161 +- 20.813       6        good
chromium@437379      685.498 +- 16.329       6        good
chromium@437380      684.21 +- 20.0923       6        good
chromium@437381      ---                     ---      build failure
---                  ---                     ---      too many build failures to list
chromium@437394      ---                     ---      build failure
chromium@437395      740.754 +- 12.8966      6        bad
chromium@437418      740.562 +- 9.64975      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 media.tough_video_cases

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8990533899224542752

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6394481932238848


| 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 24 by 42576172...@developer.gserviceaccount.com, Apr 12 2017


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

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : media.tough_video_cases
  Metric       : cputime/cputime

Revision             Result                  N
chromium@437362      681.524 +- 9141.99      156      good
chromium@437418      740.906 +- 9372.33      156      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 media.tough_video_cases

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8982651630306108320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5227323898462208


| 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!

Comment 25 by hubbe@chromium.org, Jun 23 2017

Status: WontFix (was: Untriaged)

Sign in to add a comment