All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=803267

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=31857fddded1aab397437c333855999590263fc158a5a2b8167c67a30454f60c


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

android-nexus5
android-nexus6
android-nexus7v2
android-webview-nexus5X
chromium-rel-win7-gpu-ati
chromium-rel-win7-gpu-intel
chromium-rel-win7-gpu-nvidia
linux-release

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8957107909105286768

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/169dfcc8840000

📍 Found significant differences after each of 21 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/169dfcc8840000

v8binding: Fixes StorageQuota/StorageUsage's callbacks
By yukishiino@chromium.org · Mon Jan 15 06:51:44 2018
chromium @ aa90b2313dbe917661487bceb4535c527ee78157

Automated Commit: Committing new LKGM version 10309.0.0 for chromeos.
By chromeos-commit-bot@chromium.org · Mon Jan 15 07:13:04 2018
chromium @ e736f6f27ebe595d847f3631c64a7189700de94f

Remove Memory.RendererAll.Committed
By bashi@chromium.org · Mon Jan 15 07:23:47 2018
chromium @ 33455ae9ab04dc4d624c1929a02155fab91d8d32

[turbofan] Fix type confusion in NodeProperties::InferReceiverMaps.
By bmeurer@chromium.org · Mon Jan 15 06:56:47 2018
v8 @ e272a2f722422651cf5bbbe0168702ee5d38cfe8

Deduplicate IsScaleAndIntegerTranslate
By bratell@opera.com · Mon Jan 15 11:02:02 2018
chromium @ bd0a8738015bb53b26791e12e1e54669baa58228

Add flag/glue for mojo audio input stream factories.
By maxmorin@chromium.org · Mon Jan 15 11:17:00 2018
chromium @ 3d464a42a785b72f46d8e61b418dfdc21e916702

Split OmniboxFocuser protocol into two protocols.
By stkhapugin@chromium.org · Mon Jan 15 14:24:57 2018
chromium @ 2f95cbe701e7978e7f2cb18d009fd65397f0a50e

webui: Replace base::MakeUnique with std::make_unique
By jinho.bang@samsung.com · Mon Jan 15 14:36:07 2018
chromium @ 7ab7a4e7644eebdae66084611e1ea1a6f9ee0cf2

Reenable WebRTC tests affected by harmony
By hta@chromium.org · Mon Jan 15 14:46:05 2018
chromium @ 536395c2955c6f117a0070132d57e3a0c9adab5c

Virtual time needs to pause while there are PendingScripts waiting to load
By alexclarke@chromium.org · Mon Jan 15 14:46:55 2018
chromium @ a3d7d63f8c05d5ae6f8eaadea61c1cc682a270f3

[Minor] Rename variable in VibrationController::SanitizeVibrationPattern
By awdf@chromium.org · Mon Jan 15 14:53:29 2018
chromium @ efdd6741042772d8d24884f13c86bdb855b2cc8b

Traces for audio playback/capture timestamps and delays.
By olka@chromium.org · Mon Jan 15 14:57:27 2018
chromium @ acbf1d475adfd72964c449e7db273f92e0b7b3fe

Fix offset-anchor alignment for offset-path
By fs@opera.com · Mon Jan 15 15:03:59 2018
chromium @ 5bc544493e96018f011162c09186d317a558674f

[NTP::Push] Do not use auto for the handler in the tests.
By vitaliii@chromium.org · Mon Jan 15 16:31:27 2018
chromium @ 86f0b4a812bf5d05688d5aa9b228cdcc2da57c31

Move export button under the passwords list
By ioanap@chromium.org · Mon Jan 15 16:31:38 2018
chromium @ 1c590501405b4a931e05784935a4dfa39ececd02

vr: Replace base::MakeUnique with std::make_unique
By jinho.bang@samsung.com · Mon Jan 15 18:04:42 2018
chromium @ a6bfaaa7154655078e4adf098a854f3e31de5bd8

Android: Fail build on lint unexpected
By wnwen@chromium.org · Mon Jan 15 18:30:34 2018
chromium @ fda404f68e955fc94c490d74d1efef7c5a42625e

Revert "Remove a bunch of prerender lookups in PLM and metrics code"
By csharrison@chromium.org · Mon Jan 15 19:36:57 2018
chromium @ efed514b1d911aa33b4f2b9b9abf3b11a92a74ad

Write WebAPK integration test which uses WebappLauncherActivity
By ranj@chromium.org · Mon Jan 15 20:46:42 2018
chromium @ ddbc1a049317170cd1e2477e0dd002ba5b6f00cb

Replace webgl_buffer_ in canvas and Remove ImageBuffer(Surface) etc
By xlai@chromium.org · Mon Jan 15 20:58:51 2018
chromium @ d02bb8346f13117d4a81d1124a3aeeaaed83ae99

Add ChromeRenderFrame interface for all frames.
By nverne@chromium.org · Mon Jan 15 21:42:20 2018
chromium @ 72577024b7a402caa9ab137f0f19a2549408bd90

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

I think xlai's change is a more likely candidate for this regression. Assigning.

junov@: my speculation is right--it is the eager creation of CanvasResourceProvider in
canvas that is making the regression. And after the partial revert CL, I can see that
the performance has returned back to normal as seen below

https://chromeperf.appspot.com/report?sid=8e7d732ba4cf615a8083bef4e0811c9963ff4935fed3eabee97ea6fb862afda4
(see image on screenshot)

Performance regressed at 529323 - 529339 and returned back to normal at 533804 - 533905 .

I will mark this bug as fixed by CL https://chromium-review.googlesource.com/c/chromium/src/+/896084

But I am wondering if we should add TRACE_EVENT0 to the creation of CanvasResourceProvider. That will be a different discussion topic though.

Deleted: Screen Shot 2018-02-02 at 3.03.23 PM.png 243 KB

	Screen Shot 2018-02-02 at 3.03.23 PM.png 243 KB View Download

 Issue 809552  has been merged into this issue.