All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=909795

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=e2e404ed1f2aa7e20c1dc9c6b3bea5f65a2594db088299802c50f2bc13a06365


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

mac-10_12_laptop_low_end-perf

rendering.desktop - Benchmark documentation link:
  https://bit.ly/rendering-benchmarks

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11349da4140000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/11349da4140000

Add JankTracking to field trial testing config. by skobes@chromium.org
https://chromium.googlesource.com/chromium/src/+/8984b234c4f2dc97bacb79d7139437ed044b7806
thread_total_all_cpu_time_per_frame: 50.56 → 61.51 (+10.94)

Reland "GWP-ASan: Add crash handler hook." by vtsyrklevich@chromium.org
https://chromium.googlesource.com/chromium/src/+/9caa221f953cd6e0abf770c452645c8e741137b9
thread_total_all_cpu_time_per_frame: 60.98 → 62.14 (+1.157)

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks

The JankTracking change is related to the largest increase, I think the GWP-ASan increase is a fluke (the code landed in the change doesn't get wired up to be used until a later change.)

Hmm I thought I tested JankTracker on balls_css_key_frame_animations and balls_javascript_css in  http://crbug.com/847252#c34  but maybe they aren't fixed after all.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/13dd6c44140000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/13ff956be40000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/13dd6c44140000

All of the runs failed. The most common error (20/20 runs) was:
SwarmingTestError: The test failed. No Python exception was found in the log.

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/13ff956be40000

All of the runs failed. The most common error (19/20 runs) was:
SwarmingTestError: The test failed. No Python exception was found in the log.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11fec213e40000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/11fec213e40000

All of the runs failed. The most common error (19/20 runs) was:
SwarmingTestError: The test failed. No Python exception was found in the log.

I am unable to repro locally on Linux.  Possibly Mac-specific?

It looks like these graphs have regressed again for some other reason ( issue 911118 ), which may hamper efforts to investigate this.

Update: the regression I saw in the graph is between 614130 - 614248, but it was tagged to  issue 911118  where I guess Pinpoint implicated r612776 despite it being outside that range.

https://chromeperf.appspot.com/report?sid=61f4987c3481aa11bb3889119d6edb398976e16758ece7c3e41ae1a7010ad997

 Issue 911118  is fixed but the regression persists, so I filed a new bug for it (issue 914857).

Update: I can workaround issue 912180 by passing a story filter such as "balls.javascript.css".

I can reproduce the impact of JankTracker on thread_total_all_cpu_time_per_frame in a perf try job: https://pinpoint-dot-chromeperf.appspot.com/job/136f788e140000

Skipping the calls to GeometryMapper doesn't eliminate the regression:
https://pinpoint-dot-chromeperf.appspot.com/job/16a434e9140000

Therefore, the proposal in  issue 842282  wouldn't help.

It might make sense to accept this regression.  The story is a somewhat artificial "stress test" (large number of animated DOM elements).  We don't see real-world regressions from JankTracker in UMA.

It looks like JankTrackingSweepLine might fix this. :-O

https://pinpoint-dot-chromeperf.appspot.com/job/1483b36a940000

I'd like to enable JankTrackingSweepLine in the field trial testing config to see how the graphs respond.  But for ease of verifying I'll hold off on doing that until the regression from r614152 is addressed.

The regression from r614152 recovered at r622120.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa

commit aa9c7c217a846c43dced445cf2e9c86e14f8a8fa
Author: Steve Kobes <skobes@chromium.org>
Date: Mon Jan 21 18:24:13 2019

Enable sweep line algorithm in JankTracker field trial testing config.

We want to see how the perf bots respond when this lands.  Some
improvement is expected in balls.javascript.css (see bug), but there's
a chance other benchmarks could regress.

If the sweep line algorithm is an overall improvement, we'll roll it out
via Finch (after the JankTracker rollout is finished).

Bug: 909795
Change-Id: I91d2b3b3ecfca600787432c841da8e116dbf0d99
Reviewed-on: https://chromium-review.googlesource.com/c/1417191
Reviewed-by: Timothy Dresser <tdresser@chromium.org>
Reviewed-by: Avi Drissman <avi@chromium.org>
Reviewed-by: Chris Harrelson <chrishtr@chromium.org>
Reviewed-by: Robert Kaplow <rkaplow@chromium.org>
Commit-Queue: Steve Kobes <skobes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#624632}
[modify] https://crrev.com/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa/content/child/runtime_features.cc
[modify] https://crrev.com/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa/testing/variations/fieldtrial_testing_config.json
[modify] https://crrev.com/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa/third_party/blink/common/features.cc
[modify] https://crrev.com/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa/third_party/blink/public/common/features.h
[modify] https://crrev.com/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa/third_party/blink/public/platform/web_runtime_features.h
[modify] https://crrev.com/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa/third_party/blink/renderer/platform/exported/web_runtime_features.cc
[modify] https://crrev.com/aa9c7c217a846c43dced445cf2e9c86e14f8a8fa/third_party/blink/renderer/platform/runtime_enabled_features.json5