New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 801134 link

Starred by 1 user

Issue metadata

Status: WontFix
Owner: ----
Closed: Jan 2018
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

2% regression in blink_perf.image_decoder at 526726:526820

Project Member Reported by hjd@chromium.org, Jan 11 2018

Issue description

See the link to graphs below.
 
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Jan 11 2018

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=801134

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=fa4bd2205bded059a187f5a5a4f356ce579483ff1a3f23dae92a6542804526b7


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

linux-release
Project Member

Comment 2 by 42576172...@developer.gserviceaccount.com, Jan 11 2018

馃搷 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/169a2b17040000
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Jan 11 2018

Cc: zhaobin@chromium.org pmarko@chromium.org sahel@chromium.org anthonyvd@chromium.org mdjones@chromium.org thakis@chromium.org wangxianzhu@chromium.org imch...@chromium.org dgozman@chromium.org rouslan@chromium.org shend@chromium.org thildebr@chromium.org meade@chromium.org r...@chromium.org rkaplow@chromium.org dtapu...@chromium.org rnk@google.com mfo...@chromium.org mge...@chromium.org h...@chromium.org rsesek@chromium.org nzolghadr@chromium.org xiy...@chromium.org mustaq@chromium.org tdres...@chromium.org mmenke@chromium.org
Owner: shend@chromium.org
Status: Assigned (was: Untriaged)
馃搷 Found significant differences after each of 12 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/169a2b17040000

[cast_channel] Check if socket is closed before sending channel message
By zhaobin@chromium.org 路 Wed Jan 03 18:07:37 2018
chromium @ 6bec5ff5e5f95084c6179f61dc77696a700c3f26

[Payments] Block interactions when spinning.
By rouslan@chromium.org 路 Wed Jan 03 18:15:45 2018
chromium @ ab867f3ed12010a7e96cc319a9d7b745a361060b

Skip fast/forms/select/listbox-in-multi-column.html
By wangxianzhu@chromium.org 路 Wed Jan 03 18:16:06 2018
chromium @ 862023f8adc0a91832aa24c460ea7799964ffc8d

Don't crash browser on invalid input received from the injection channels.
By dtapuska@chromium.org 路 Wed Jan 03 18:17:49 2018
chromium @ a79ff1d1787e61497996f9a6beed3557f26da1fd

Fix bottom navigation crash bug on global layout.
By thildebr@chromium.org 路 Wed Jan 03 18:26:07 2018
chromium @ d496e41172a70ac7f920a58db2fc362435b2b56c

Roll clang 318667:321529.
By rnk@google.com 路 Wed Jan 03 18:32:36 2018
chromium @ 0d72bb1494c94523ae7bd944476a495e8b281bcc

Properly propagate proxy auth details on sign-in screen
By pmarko@chromium.org 路 Wed Jan 03 20:07:54 2018
chromium @ 14a0def082f8568c1423b9c14bee21070b86fb5a

DevTools: only inject extensions API once.
By dgozman@chromium.org 路 Wed Jan 03 20:17:51 2018
chromium @ 81504ef63932f7ad2af235f44f59e37165b7430d

touchpad_browserside_fling_flaky_tests_fixed.
By sahel@chromium.org 路 Wed Jan 03 20:18:02 2018
chromium @ c00c1189b6358b400bc17e8c62c49e6275f5bb0c

Remove virtual from DnsConfig's destructor
By mmenke@chromium.org 路 Wed Jan 03 20:26:32 2018
chromium @ 04dee1f1b44ed7bac935a9fa4ccbe37f2a7dc87e

Add TimeToHandled & IsMainThread input ukm metrics
By nzolghadr@chromium.org 路 Wed Jan 03 20:39:25 2018
chromium @ 8b08db3af3baa38986d4bfa1e3bc1b4060130d6f

[css-typed-om] Implement transform value serialization.
By shend@chromium.org 路 Wed Jan 03 20:50:51 2018
chromium @ f4649f3a002823b9e3b9cdac5ec9ba7286fa81f1

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

Comment 4 by shend@chromium.org, Jan 11 2018

Owner: ----
Status: Available (was: Assigned)
I don't think this is caused by my patch. It changes some code in a feature that's not even shipped yet.

Comment 5 by mmenke@chromium.org, Jan 11 2018

I removed "virtual" from a destructor of a class with no subclasses (And this is the second perf regression email I've received about that one CL).  :(
Cc: -wangxianzhu@chromium.org -sahel@chromium.org
Cc: sahel@chromium.org
rouslan@ is OoO but it's basically impossible for his patch in Payment
Request UI to be responsible for a regression in blink image decoding
(unless those tests call into Payment Request for some reason!).
Not it :-) I changed a CHECK to a LOG(WARNING)

Comment 10 by hjd@chromium.org, Jan 12 2018

Cc: dtu@chromium.org
+cc:dtu@ - does this look right to you?

Meanwhile I'll re-kick pinpoint to see if this reproduces.

Comment 12 by r...@chromium.org, Jan 12 2018

Cc: -rnk@google.com inglorion@chromium.org
The first pinpoint job seems to clearly show that it's the clang roll. The compiler now generates code that's slightly slower on this benchmark.

It's unlikely that we'll have the bandwidth to run this down, though. It's probably lower priority than the size regression issue ( http://crbug.com/799646 ).
Seems not related to my patch. Just added a null check to an extension function.

Comment 14 by r...@chromium.org, Jan 12 2018

Cc: -mfo...@chromium.org -xiy...@chromium.org -rsesek@chromium.org -nzolghadr@chromium.org -rouslan@chromium.org -meade@chromium.org -sahel@chromium.org -imch...@chromium.org -dtu@chromium.org -pmarko@chromium.org -dtapu...@chromium.org -mdjones@chromium.org -mustaq@chromium.org -mge...@chromium.org -anthonyvd@chromium.org -tdres...@chromium.org -dgozman@chromium.org -mmenke@chromium.org -shend@chromium.org -zhaobin@chromium.org -hjd@chromium.org -rkaplow@chromium.org -thildebr@chromium.org
I'm going to un-cc everyone else. Feel free to re-add if you want to follow along.

Comment 15 by hjd@chromium.org, Jan 12 2018

Cc: hjd@chromium.org
Project Member

Comment 16 by 42576172...@developer.gserviceaccount.com, Jan 12 2018

馃樋 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/14c2e0f7040000

Comment 17 by dtu@chromium.org, Jan 12 2018

Yes, the Clang roll.
Status: WontFix (was: Available)
WontFix-ing per #12

Sign in to add a comment