New issue
Advanced search Search tips

Issue 733232 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner:
Closed: Jun 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression


Show other hotlists

Hotlists containing this issue:
Hotlist-1


Sign in to add a comment

2.9%-41% regression in v8.runtimestats.browsing_desktop at 475300:475361

Project Member Reported by tebbi@chromium.org, Jun 14 2017

Issue description

See the link to graphs below.
 
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Jun 14 2017

Cc: mythria@chromium.org
Owner: mythria@chromium.org

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

Hi mythria@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 : Mythri
  Commit : c4b187baa608b68be91c193c4dc166d65b5e4943
  Date   : Tue May 23 09:25:46 2017
  Subject: [Turbofan] Increase max_inlined_nodes to 230.

Bisect Details
  Configuration: mac_retina_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : Optimize:duration_avg/browse_search/browse_search_google
  Change       : 36.31% | 8.1795 -> 11.1498333333

Revision                           Result                   N
chromium@475301                    8.1795 +- 1.06042        6      good
chromium@475320                    7.95833 +- 0.737707      6      good
chromium@475330                    7.7735 +- 0.920424       6      good
chromium@475333                    7.598 +- 0.426987        6      good
chromium@475333,v8@9b1d22fc2c      7.993 +- 0.953803        6      good
chromium@475333,v8@6a9f54e875      7.982 +- 1.04259         6      good
chromium@475333,v8@e1b4b405b1      8.062 +- 1.57734         6      good
chromium@475333,v8@c4b187baa6      11.4773 +- 1.90566       6      bad       <--
chromium@475333,v8@ea48d83d37      10.7707 +- 0.519037      6      bad
chromium@475333,v8@611ec69d85      11.2622 +- 1.31552       6      bad
chromium@475334                    11.4245 +- 0.596275      6      bad
chromium@475335                    10.5698 +- 0.9689        6      bad
chromium@475339                    11.1498 +- 1.29632       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 --story-filter=browse.search.google v8.runtimestats.browsing_desktop

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8976807618163442320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6163626345365504


| 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!
Owner: tebbi@chromium.org
Status: Assigned (was: Untriaged)
Assigning to performance sheriff.
Apart from the regressions on twitter Parse the others are in the optimize bucket. The absolute delta is not very high and this not regress the Total:V8 metric. So I guess, the regressions on the optimize bucket can be ignored.

The parse regression on twitter seems more like noise to me since there was an improvement and then a regression back to the original state and no V8 changes in between. I am just triaging them into a different bug, just incase I missed something.
There are also regressions on the classic (crankshaft) benchmarks where the absolute delta is much higher, but we can ignore classic benchmarks, since this configuration is no longer tested.
Project Member

Comment 7 by 42576172...@developer.gserviceaccount.com, Jun 18 2017

 Issue 733234  has been merged into this issue.

Comment 9 by tebbi@chromium.org, Jun 26 2017

Status: WontFix (was: Assigned)
Won't fix based on mythria@'s comment.
Project Member

Comment 10 by 42576172...@developer.gserviceaccount.com, Jun 26 2017


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Mythri
  Commit : c4b187baa608b68be91c193c4dc166d65b5e4943
  Date   : Tue May 23 09:25:46 2017
  Subject: [Turbofan] Increase max_inlined_nodes to 230.

Bisect Details
  Configuration: mac_10_11_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : Optimize:duration_avg/browse_search/browse_search_google
  Change       : 36.82% | 8.59016666667 -> 11.7531666667

Revision                           Result                   N
chromium@475303                    8.59017 +- 0.78023       6      good
chromium@475322                    8.55317 +- 0.539967      6      good
chromium@475332                    8.21433 +- 0.576166      6      good
chromium@475333                    8.1735 +- 0.632992       6      good
chromium@475333,v8@9b1d22fc2c      8.19467 +- 0.397471      6      good
chromium@475333,v8@6a9f54e875      8.438 +- 0.844227        6      good
chromium@475333,v8@e1b4b405b1      8.374 +- 1.29006         6      good
chromium@475333,v8@c4b187baa6      11.5277 +- 0.708947      6      bad       <--
chromium@475333,v8@ea48d83d37      11.77 +- 1.52741         6      bad
chromium@475333,v8@611ec69d85      11.896 +- 1.74959        6      bad
chromium@475334                    11.7918 +- 2.0932        6      bad
chromium@475335                    11.5182 +- 0.939006      6      bad
chromium@475337                    11.5527 +- 1.37239       6      bad
chromium@475341                    11.7532 +- 1.56295       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 --story-filter=browse.search.google v8.runtimestats.browsing_desktop

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8975735058956562032

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6426803670876160


| 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