All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=714902

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4vC0ngoM


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

chromium-rel-win7-gpu-ati

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8981385236410481008

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

Hi sunnyps@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 : Sunny Sachanandani
  Commit : 43bd95af39ce6dc2c2ac11d70a1fbff816ef0794
  Date   : Thu Apr 20 03:32:47 2017
  Subject: gpu: Fix reduced frame rate when animating multiple windows on Linux.

Bisect Details
  Configuration: winx64ati_perf_bisect
  Benchmark    : scheduler.tough_scheduling_cases
  Metric       : mean_input_event_latency/mean_input_event_latency
  Change       : 49.92% | 2.49715394089 -> 3.74361904762

Revision             Result                   N
chromium@465818      2.49715 +- 0.908766      6      good
chromium@465879      2.32522 +- 0.261494      6      good
chromium@465887      2.66814 +- 1.11951       6      good
chromium@465888      2.38533 +- 0.769596      6      good
chromium@465889      4.00135 +- 0.336566      6      bad       <--
chromium@465891      3.87663 +- 0.942153      6      bad
chromium@465894      3.78311 +- 0.630279      6      bad
chromium@465909      3.95221 +- 0.788768      6      bad
chromium@465939      3.74362 +- 0.662567      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 scheduler.tough_scheduling_cases

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981385236410481008

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6429648222683136


| 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!

sunnyps, can you take a look?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3e58482d43b23ea6445a56adc731f1e5b0ce0b0f

commit 3e58482d43b23ea6445a56adc731f1e5b0ce0b0f
Author: Sunny Sachanandani <sunnyps@chromium.org>
Date: Fri Apr 28 03:43:04 2017

gpu: Multi window force swap interval zero tweak.

When I moved the multi window swap interval logic, I changed a condition
by mistake. With that change the swap interval is reset to 1 for the
first swap in a new swap generation. I suspect that caused the input
latency regression although I can't reproduce it in local testing.

R=kbr
BUG= 714902 

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Change-Id: If7f7ae4f824b39dade4cd9d81e68ce47cfc8f38d
Reviewed-on: https://chromium-review.googlesource.com/487875
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Sunny Sachanandani <sunnyps@chromium.org>
Cr-Commit-Position: refs/heads/master@{#467877}
[modify] https://crrev.com/3e58482d43b23ea6445a56adc731f1e5b0ce0b0f/gpu/ipc/service/pass_through_image_transport_surface.cc

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8980763021153695856

Started a bisect to confirm that the regression is fixed by the above CL.

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

Hi sunnyps@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 : Sunny Sachanandani
  Commit : 3e58482d43b23ea6445a56adc731f1e5b0ce0b0f
  Date   : Fri Apr 28 03:43:04 2017
  Subject: gpu: Multi window force swap interval zero tweak.

Bisect Details
  Configuration: winx64ati_perf_bisect
  Benchmark    : scheduler.tough_scheduling_cases
  Metric       : mean_input_event_latency/mean_input_event_latency
  Change       : 36.36% | 3.77502019704 -> 2.40226262999

Revision             Result                   N
chromium@467871      3.77502 +- 0.512261      6      good
chromium@467876      3.6883 +- 0.584601       6      good
chromium@467877      2.55375 +- 0.856299      6      bad       <--
chromium@467878      2.35521 +- 0.642338      6      bad
chromium@467880      2.48691 +- 0.823633      6      bad
chromium@467892      2.57534 +- 0.814992      6      bad
chromium@467913      2.36388 +- 0.395168      6      bad
chromium@467954      2.40226 +- 0.912314      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 scheduler.tough_scheduling_cases

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8980763021153695856

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5901372412657664


| 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!

Ok, so the regression is fixed.

The multi window disable vsync logic isn't supposed to trigger for only one window. When porting that to Linux I accidentally fixed a bug which kept vsync disabled for the first 60 swaps. I'm going to fix this bug (not by accident this time :) so this regression will be WontFix. skyostil@ tdresser@ WDYT?

SGTM, thanks for digging in here.

Marking WontFix per #10 and #11.