All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=775849

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=4fdf2dd7148598e4523a06250eb28866cecfa72bcc1d6b66214a5a43a8ba7acf


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

chromium-rel-mac-retina
chromium-rel-mac11
chromium-rel-mac11-pro
win-high-dpi

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8965406865297457072

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

Hi toyoshim@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 : Takashi Toyoshima
  Commit : 36160f0427ad5657e299fa932c920d1f490bc1e1
  Date   : Mon Oct 16 10:23:14 2017
  Subject: ResourceLoadScheduler: update fieldtrial_testing_config.json

Bisect Details
  Configuration: winx64_high_dpi_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : V8 C++:duration_avg/browse_tools/browse_tools_earth
  Change       : 185.96% | 3061.7395 -> 8755.326

Revision             Result                  N
chromium@508991      3061.74 +- 581.476      6      good
chromium@509001      3098.98 +- 655.071      6      good
chromium@509006      3106.16 +- 425.133      6      good
chromium@509008      2968.63 +- 793.661      6      good
chromium@509009      8385.03 +- 2259.14      6      bad       <--
chromium@509010      8873.22 +- 2878.13      6      bad
chromium@509029      8755.33 +- 2318.12      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=browse.tools.earth v8.runtimestats.browsing_desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8965406865297457072


For feedback, file a bug with component Speed>Bisection

cc: ulan@ (from go/chrome-benchmarks)
Could you help to understand what these tests actually measure, and if this regression is expected and reasonable?

The modified perf test params control ResourceLoadScheduler that throttles background tab resource loading. So, some sort of regression is expected, but I'd double check it if this is really expected regression, and the test is still doing something meaningful before/after my change.

 Issue 776401  has been merged into this issue.

toyoshim@, this metric measures time spent in C++ code in V8 (and maybe Blink callbacks in V8). cbruni@, please correct me if I am wrong.

Metric       : V8 C++:duration_avg/browse_tools/browse_tools_earth
Change       : 185.96% | 3061.7395 -> 8755.326

Regression of 185% is huge, and looks like your CL is affecting the foreground tab because the benchmark runs only one tab at a time.

Thank you for inputs, ulan.

This config change enabled memory reporting code that would cost on Windows.

Since the problem happens only on Windows, probably that was a root cause.
Merged bug of #6 is also Windows specific.
The group report contains one mac bot case, but it looks a false alert. The result went back to the original level.

I will land a change to remove the reporting flag.

supplements: I suspect memory walks affects mem$, TLB and so on in a bad manner.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a3d458c7e9c8ebf21833bbaa48ee4e4159a516dc

commit a3d458c7e9c8ebf21833bbaa48ee4e4159a516dc
Author: Takashi Toyoshima <toyoshim@chromium.org>
Date: Fri Mar 09 23:22:35 2018

ResourceLoadScheduler: remove stale configurations

Remove the entry "Enabled_bg_limit_8", that was for the first field trial.
Also, remove "ReportRendererPeakMemoryStats" feature, that is not
planned to be used in a launching configuration, but it was just for
gathering memory metrics until Beta.

Bug:  775849 
Change-Id: Ib496db8d69b6891915cfab453ad4abc13d66e522
Reviewed-on: https://chromium-review.googlesource.com/956945
Reviewed-by: Ilya Sherman <isherman@chromium.org>
Commit-Queue: Ilya Sherman <isherman@chromium.org>
Cr-Commit-Position: refs/heads/master@{#542275}
[modify] https://crrev.com/a3d458c7e9c8ebf21833bbaa48ee4e4159a516dc/testing/variations/fieldtrial_testing_config.json

https://chromeperf.appspot.com/report?sid=89478bc47a37cda2f281ad594a7121de98b47b512ed3920f8999bf4da3979873

Since the test suite was renamed, the original graph could not show recent data in the same picture, but as I can see in the recent graph linked above, the problem looks fixed by the CL of #11.

In the revision range 542150 - 542329, the result in ms (lower is better) was changed from 10272 to 3135.

 Issue 782225  has been merged into this issue.