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

Issue 713570 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
OOO until 2019-02-10
Closed: Apr 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression

Blocking:
issue 700835



Sign in to add a comment

91.6% regression in blink_perf.bindings at 465634:465663

Project Member Reported by alexclarke@chromium.org, Apr 20 2017

Issue description

See the link to graphs below.
 
All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=713570

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgoqP8pwoM


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

android-webview-nexus6
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Apr 21 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array


To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981811234378648288

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4948239746334720


| 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!
Project Member

Comment 7 by 42576172...@developer.gserviceaccount.com, Apr 21 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_nexus6_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array


To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981696430349730272

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5885040631742464


| 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!
Project Member

Comment 9 by 42576172...@developer.gserviceaccount.com, Apr 22 2017

Cc: petermarshall@chromium.org
Owner: petermarshall@chromium.org

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

Hi petermarshall@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 : Peter Marshall
  Commit : 356e9246b2f9653f1ef3bf47b3e801391b1824c3
  Date   : Wed Apr 19 12:45:20 2017
  Subject: [builtins] Use the ElementsAccessor to copy TypedArrays.

Bisect Details
  Configuration: android_nexus6_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array
  Change       : 92.47% | 143.984241531 -> 10.8414768815

Revision                           Result                   N
chromium@465602                    143.984 +- 0.990245      6      good
chromium@465625                    143.89 +- 0.599245       6      good
chromium@465637                    144.223 +- 0.680597      6      good
chromium@465643                    142.039 +- 7.2555        6      good
chromium@465645                    141.238 +- 9.36369       6      good
chromium@465645,v8@3e721358aa      142.377 +- 6.77253       6      good
chromium@465645,v8@356e9246b2      10.7144 +- 0.286192      6      bad       <--
chromium@465645,v8@b4c8d38387      10.6304 +- 0.475932      6      bad
chromium@465646                    10.8612 +- 0.183776      6      bad
chromium@465648                    10.8415 +- 0.224238      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981681666075599136

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5885040631742464


| 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!
Project Member

Comment 10 by 42576172...@developer.gserviceaccount.com, Apr 23 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array


To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981696421417605392

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5886959676817408


| 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!
Project Member

Comment 11 by 42576172...@developer.gserviceaccount.com, Apr 23 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array


To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981693490559017136

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4948239746334720


| 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!
Blocking: 700835
Project Member

Comment 15 by 42576172...@developer.gserviceaccount.com, Apr 25 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array


To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981529892804680960

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5886959676817408


| 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!
Project Member

Comment 16 by 42576172...@developer.gserviceaccount.com, Apr 25 2017


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Peter Marshall
  Commit : 356e9246b2f9653f1ef3bf47b3e801391b1824c3
  Date   : Wed Apr 19 12:45:20 2017
  Subject: [builtins] Use the ElementsAccessor to copy TypedArrays.

Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array
  Change       : 86.26% | 47.4070638608 -> 6.51282875611

Revision                           Result                   N
chromium@465625                    47.4071 +- 0.497975      6      good
chromium@465642                    47.7142 +- 0.884738      6      good
chromium@465645                    47.3599 +- 0.433257      6      good
chromium@465645,v8@3e721358aa      47.294 +- 0.743992       6      good
chromium@465645,v8@356e9246b2      6.51604 +- 0.423459      6      bad       <--
chromium@465645,v8@b4c8d38387      6.51685 +- 0.644365      6      bad
chromium@465646                    6.66437 +- 0.332264      6      bad
chromium@465647                    6.44769 +- 0.662902      6      bad
chromium@465651                    6.47862 +- 0.439561      6      bad
chromium@465659                    6.6049 +- 0.297006       6      bad
chromium@465693                    6.51283 +- 0.455176      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981441012273277760

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4948239746334720


| 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!
Status: Started (was: Untriaged)
We need a fast case for holey arrays in the elements accessor TryCopyElementsHandleFastNumber
Project Member

Comment 19 by bugdroid1@chromium.org, Apr 26 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/v8/v8.git/+/e855e514d111261a185f9de98add42c224d7bcbe

commit e855e514d111261a185f9de98add42c224d7bcbe
Author: Peter Marshall <petermarshall@chromium.org>
Date: Wed Apr 26 15:36:36 2017

[builtins] Add a fast path to construct TypedArrays from holey arrays.

For holey Smi and double source arrays, we would go to the general
case, which is much slower than before. We already check that there
are no prototype chain changes in IterableToListCanBeElided, and
there is no JS-code run between that check and the copying of the
elements, so we can safely check for the hole and convert it to
undefined, which is then converted to 0/NaN appropriately for the
given TypedArray.

Bug:  chromium:713570 , chromium:711275 
Change-Id: I5b21c915907d71eebb73b7b1eea8eb58b4a5436d
Reviewed-on: https://chromium-review.googlesource.com/485520
Commit-Queue: Peter Marshall <petermarshall@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#44899}
[modify] https://crrev.com/e855e514d111261a185f9de98add42c224d7bcbe/src/elements.cc
[add] https://crrev.com/e855e514d111261a185f9de98add42c224d7bcbe/test/mjsunit/es6/typedarray-construct-by-array-like-prototype-element-added.js
[modify] https://crrev.com/e855e514d111261a185f9de98add42c224d7bcbe/test/mjsunit/es6/typedarray-construct-by-array-like.js

Status: Fixed (was: Started)
Project Member

Comment 21 by 42576172...@developer.gserviceaccount.com, Apr 27 2017


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Peter Marshall
  Commit : 356e9246b2f9653f1ef3bf47b3e801391b1824c3
  Date   : Wed Apr 19 12:45:20 2017
  Subject: [builtins] Use the ElementsAccessor to copy TypedArrays.

Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : blink_perf.bindings
  Metric       : typed-array-construct-from-array/typed-array-construct-from-array
  Change       : 86.43% | 47.5231770451 -> 6.44745295024

Revision                           Result                   N
chromium@465625                    47.5232 +- 0.546905      6      good
chromium@465642                    47.4364 +- 0.771881      6      good
chromium@465645                    47.7426 +- 0.592837      6      good
chromium@465645,v8@3e721358aa      47.3615 +- 0.843976      6      good
chromium@465645,v8@356e9246b2      6.31133 +- 0.658841      6      bad       <--
chromium@465645,v8@b4c8d38387      6.48424 +- 0.337161      6      bad
chromium@465646                    6.48706 +- 0.464317      6      bad
chromium@465647                    6.54433 +- 0.40899       6      bad
chromium@465651                    6.5753 +- 0.39055        6      bad
chromium@465659                    6.50553 +- 0.462067      6      bad
chromium@465693                    6.44745 +- 0.528274      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.bindings

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981269404980024496

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5886959676817408


| 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!

Sign in to add a comment