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

Issue 596823 link

Starred by 0 users

Issue metadata

Status: Fixed
Owner:
Last visit > 30 days ago
Closed: Apr 2016
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

6.8% regression in thread_times.tough_scrolling_cases at 382336:382360

Project Member Reported by alexclarke@chromium.org, Mar 22 2016

Issue description

See the link to graphs below.
 
All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=596823

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgkNW9rgoM


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

android-nexus5
Cc: dtapu...@chromium.org tdres...@chromium.org
This is an anticipated regression.

See change; https://codereview.chromium.org/1776843003/

We've reproduced this exact scenario on the trybot when I was trying this change.
This regression in wheel latency shouldn't make wheel any worse than touch.

As far as I know, we don't have any telemetry tests which are run with both wheel and touch on the same device for comparison, which is a bit of a shame.

I think we're fine calling this WontFix though.
Project Member

Comment 4 by 42576172...@developer.gserviceaccount.com, Mar 22 2016

Owner: dtapu...@chromium.org

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

Hi dtapuska@chromium.org, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Re-land enabling wheel gestures on all platforms.
Author  : dtapuska
Commit description:
  
Changes since last attempt in https://codereview.chromium.org/1724783002/

MacOSX implementation has landed in
https://codereview.chromium.org/1749343004/

The Android perf regression has been eliminated because the default
wheel transaction is set to 0. The perf regression was because the
wheel transaction time was added to the total time of the test.

Attention perf sherrifs; a minor perf regression is expected in
this change. 8-10% when locally on my Nexus 5; this minor regression is
anticipated because of the new code paths.

BUG= 568183 

Review URL: https://codereview.chromium.org/1776843003

Cr-Commit-Position: refs/heads/master@{#382345}
Commit  : ca4feba9a31e10d1baab5813650c192e06cfd643
Date    : Mon Mar 21 19:00:47 2016


===== TESTED REVISIONS =====
Revision                Mean Value  Std. Dev.   Num Values  Good?
chromium@382335         34.319136   0.096045    5           good
chromium@382342         34.288702   0.047703    5           good
chromium@382344         34.318446   0.041487    5           good
chromium@382345         36.582866   0.184803    5           bad
chromium@382348         36.515909   0.110469    5           bad
chromium@382360         36.494983   0.102015    5           bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 596823

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --also-run-disabled-tests thread_times.tough_scrolling_cases
Test Metric: thread_total_all_cpu_time_per_frame/thread_total_all_cpu_time_per_frame
Relative Change: 6.34%
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/3515
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9017496542426103152


| 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 label Cr-Tests-AutoBisect.  Thank you!
Status: WontFix (was: Assigned)
Anticipated regression. Owner of metric agrees in #3 this is ok.
Project Member

Comment 6 by 42576172...@developer.gserviceaccount.com, Mar 22 2016

Cc: rdsmith@chromium.org
Owner: rdsmith@chromium.org

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

Hi rdsmith@chromium.org, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Fix bug in net::RequestPriority -> HTTP/2 dependency conversion.
Author  : rdsmith
Commit description:
  
This CL also defaults this behavior on (subject to finch override), refactors the finch control to be in line with network stack defaults (== based in io_thread.cc),  and removes the non-default performance bot configuration.

BUG= 590225 
R=bnc@chromium.org
R=isherman@chromium.org

Review URL: https://codereview.chromium.org/1779733003

Cr-Commit-Position: refs/heads/master@{#382360}
Commit  : 2e54d1f83e6ae57cb404658f361e469dd9ed2a92
Date    : Mon Mar 21 19:50:41 2016


===== TESTED REVISIONS =====
Revision                Mean Value  Std. Dev.   Num Values  Good?
chromium@382335         32.363611   1.99554     8           good
chromium@382348         31.744833   3.166965    8           good
chromium@382354         31.031188   2.402719    5           good
chromium@382357         31.232222   1.819694    5           good
chromium@382359         31.372123   1.909671    5           good
chromium@382360         35.900939   0.207964    5           bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 596823

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --also-run-disabled-tests thread_times.tough_scrolling_cases
Test Metric: thread_total_all_cpu_time_per_frame/thread_total_all_cpu_time_per_frame
Relative Change: 6.64%
Score: 99.5

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/3514
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9017496558845783696


| 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 label Cr-Tests-AutoBisect.  Thank you!
I'm confused.  I've been assigned as the owner, and my change fingered by bisect, but several people have indicated that this is an anticipated regression and the bug is currently marked WontFix (not by me).  Can anyone explain to me what the actual status of this bug is and whether I should pay attention to it?

Owner: ----
rdsmith: apologies for the confusion. The bisect bot seems to have encountered unusually high noise somehow when doing a second run (std dev values way higher in #6 than #4), which I think is why it pinpointed your CL.

We also have an outstanding bug to not let the bot update closed issues.

Removing you as owner, and reporting this at go/badbisects
Project Member

Comment 9 by 42576172...@developer.gserviceaccount.com, Mar 22 2016


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Re-land enabling wheel gestures on all platforms.
Author  : dtapuska
Commit description:
  
Changes since last attempt in https://codereview.chromium.org/1724783002/

MacOSX implementation has landed in
https://codereview.chromium.org/1749343004/

The Android perf regression has been eliminated because the default
wheel transaction is set to 0. The perf regression was because the
wheel transaction time was added to the total time of the test.

Attention perf sherrifs; a minor perf regression is expected in
this change. 8-10% when locally on my Nexus 5; this minor regression is
anticipated because of the new code paths.

BUG= 568183 

Review URL: https://codereview.chromium.org/1776843003

Cr-Commit-Position: refs/heads/master@{#382345}
Commit  : ca4feba9a31e10d1baab5813650c192e06cfd643
Date    : Mon Mar 21 19:00:47 2016


===== TESTED REVISIONS =====
Revision                Mean Value  Std. Dev.   Num Values  Good?
chromium@382342         26.697165   0.166469    5           good
chromium@382344         26.705234   0.198604    5           good
chromium@382345         27.900835   0.060334    5           bad
chromium@382347         27.818438   0.068391    5           bad
chromium@382352         27.974375   0.099292    5           bad
chromium@382362         27.784961   0.088241    5           bad

Bisect job ran on: android_nexus6_perf_bisect
Bug ID: 596823

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --also-run-disabled-tests thread_times.tough_scrolling_cases
Test Metric: thread_total_all_cpu_time_per_frame/thread_total_all_cpu_time_per_frame
Relative Change: 4.07%
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus6_perf_bisect/builds/2043
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9017496371945590176


| 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 label Cr-Tests-AutoBisect.  Thank you!
Project Member

Comment 10 by 42576172...@developer.gserviceaccount.com, Mar 22 2016


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Re-land enabling wheel gestures on all platforms.
Author  : dtapuska
Commit description:
  
Changes since last attempt in https://codereview.chromium.org/1724783002/

MacOSX implementation has landed in
https://codereview.chromium.org/1749343004/

The Android perf regression has been eliminated because the default
wheel transaction is set to 0. The perf regression was because the
wheel transaction time was added to the total time of the test.

Attention perf sherrifs; a minor perf regression is expected in
this change. 8-10% when locally on my Nexus 5; this minor regression is
anticipated because of the new code paths.

BUG= 568183 

Review URL: https://codereview.chromium.org/1776843003

Cr-Commit-Position: refs/heads/master@{#382345}
Commit  : ca4feba9a31e10d1baab5813650c192e06cfd643
Date    : Mon Mar 21 19:00:47 2016


===== TESTED REVISIONS =====
Revision                Mean Value  Std. Dev.   Num Values  Good?
chromium@382320         37.22964    0.167509    5           good
chromium@382341         37.049772   0.255215    5           good
chromium@382344         37.036818   0.165561    5           good
chromium@382345         39.11133    0.297739    5           bad
chromium@382346         39.192681   0.192095    5           bad
chromium@382347         39.107328   0.137678    5           bad
chromium@382352         39.051074   0.255769    4           bad
chromium@382362         39.179862   0.181319    5           bad

Bisect job ran on: android_nexus7_perf_bisect
Bug ID: 596823

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --also-run-disabled-tests thread_times.tough_scrolling_cases
Test Metric: thread_total_all_cpu_time_per_frame/thread_total_all_cpu_time_per_frame
Relative Change: 5.24%
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus7_perf_bisect/builds/2855
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9017496283469536144


| 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 label Cr-Tests-AutoBisect.  Thank you!
Project Member

Comment 11 by 42576172...@developer.gserviceaccount.com, Mar 31 2016


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Re-land enabling wheel gestures on all platforms.
Author  : dtapuska
Commit description:
  
Changes since last attempt in https://codereview.chromium.org/1724783002/

MacOSX implementation has landed in
https://codereview.chromium.org/1749343004/

The Android perf regression has been eliminated because the default
wheel transaction is set to 0. The perf regression was because the
wheel transaction time was added to the total time of the test.

Attention perf sherrifs; a minor perf regression is expected in
this change. 8-10% when locally on my Nexus 5; this minor regression is
anticipated because of the new code paths.

BUG= 568183 

Review URL: https://codereview.chromium.org/1776843003

Cr-Commit-Position: refs/heads/master@{#382345}
Commit  : ca4feba9a31e10d1baab5813650c192e06cfd643
Date    : Mon Mar 21 19:00:47 2016


===== TESTED REVISIONS =====
Revision                Mean Value  Std. Dev.   Num Values  Good?
chromium@382100         40.722275   0.342667    5           good
chromium@382328         41.2182     0.505202    5           good
chromium@382342         41.284833   0.989065    5           good
chromium@382344         41.823108   0.961187    5           good
chromium@382345         108.871588  26.370787   5           bad         <-
chromium@382346         107.609368  17.876584   5           bad
chromium@382349         121.769553  38.399338   5           bad
chromium@382355         106.55348   39.684257   5           bad
chromium@382386         106.955013  22.4081     5           bad
chromium@382440         102.5606    16.499414   5           bad
chromium@382549         88.469724   28.611929   5           bad
chromium@382999         81.043807   28.352971   5           bad

Bisect job ran on: mac_hdd_perf_bisect
Bug ID: 596823

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --also-run-disabled-tests smoothness.top_25_smooth
Test Metric: mean_input_event_latency/mean_input_event_latency
Relative Change: 99.02%
Score: 99.8

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_hdd_perf_bisect/builds/461
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9016761715768843680


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=596823

| 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 Tests>AutoBisect.  Thank you!
Cc: -dtapu...@chromium.org
Owner: dtapu...@chromium.org
Status: Assigned (was: WontFix)
This looks like a catastrophe on Mac, and I'm not sure why.

dtapuska@, any thoughts? In some cases this doubled the latency.
This appears to be the fact that the perf bot on mac has stoppped running a number of tests and the mean is now higher.

see https://chromeperf.appspot.com/report?sid=8b6c663caff03c063b278338ddfb8cef2ab9a44b5d2a63dbdd4e99988427dec4&start_rev=382024&end_rev=384019
Cc: -rdsmith@chromium.org
Pulling myself off the bug because I don't think my change is at all related to the issues being discussed; feel free to toss me back on if you disagree.


Project Member

Comment 15 by bugdroid1@chromium.org, Apr 4 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/581656043e63a1e26fb6b038823b2d0bb2a28710

commit 581656043e63a1e26fb6b038823b2d0bb2a28710
Author: dtapuska <dtapuska@chromium.org>
Date: Mon Apr 04 17:59:04 2016

Fix Event Latency Info on generate gesture events.

Generated gesture events weren't marking the begin component
correctly which caused the tag not to be set correctly.
In order to achieve this proper marking the events need
to propagate up from the InputRouterImpl into the
RenderWidgetHostImpl which adds the latency information.

BUG= 596823 

Review URL: https://codereview.chromium.org/1847823005

Cr-Commit-Position: refs/heads/master@{#384951}

[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/input_router_client.h
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/input_router_impl.cc
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/input_router_impl.h
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/input_router_impl_perftest.cc
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/mock_input_router_client.cc
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/mock_input_router_client.h
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/mouse_wheel_event_queue.cc
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/mouse_wheel_event_queue.h
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/input/mouse_wheel_event_queue_unittest.cc
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/render_widget_host_impl.cc
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/render_widget_host_impl.h
[modify] https://crrev.com/581656043e63a1e26fb6b038823b2d0bb2a28710/content/browser/renderer_host/render_widget_host_view_aura_unittest.cc

Status: Fixed (was: Assigned)
mac build bots has recovered; no regression on their metrics shown.

Project Member

Comment 17 by 42576172...@developer.gserviceaccount.com, Apr 30 2016

Cc: jaydasika@chromium.org
Owner: jaydasika@chromium.org

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

Hi jaydasika@chromium.org, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : cc : Make tree synchronization independent of layer tree hierarchy (2)
Author  : jaydasika
Commit description:
  
This CL :
  * Stores layers that need to push properties in LayerTreeHost(layer_set)
  * Deletes bools needs_push_properties and dependants_needs_push_properties
    from Layer.
  * Iterate the layer_set tp push properties during commit.
  * Changes what layers are serialized (Both dirty layers and their ancestors
    are serialized currently. With this CL, only dirty layers serialize).

BUG= 568874 
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Review URL: https://codereview.chromium.org/1808373002

Cr-Commit-Position: refs/heads/master@{#382379}
Commit  : 9234e405c7089c47ca9d30b34846f54b1fd9b8fd
Date    : Mon Mar 21 20:45:22 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev     N  Good?
chromium@382349  3.30927  0.00923424  8  good
chromium@382369  3.34161  0.0201525   8  good
chromium@382374  3.33607  0.0202837   8  good
chromium@382377  3.36448  0.0211073   8  good
chromium@382378  3.3676   0.00669336  5  good
chromium@382379  3.41358  0.0178216   5  bad    <--
chromium@382389  3.40325  0.0231589   5  bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 596823

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --also-run-disabled-tests thread_times.key_silk_cases
Test Metric: thread_renderer_compositor_cpu_time_per_frame/thread_renderer_compositor_cpu_time_per_frame
Relative Change: 2.84%
Score: 99.8

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/3642
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9014015675788188096


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6440505802489856

| 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 Tests>AutoBisect.  Thank you!

Sign in to add a comment