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

Issue 700829 link

Starred by 5 users

Issue metadata

Status: Duplicate
Merged: issue 700364
Owner: ----
Closed: Mar 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression

Blocking:
issue 698746


Participants' hotlists:
I-TF-Launch


Sign in to add a comment

21.4% regression in octane (Chromeperf) at 455717:455748

Project Member Reported by hablich@chromium.org, Mar 13 2017

Issue description

See the link to graphs below.
 
All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=700829

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg1N_p7QsM


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

android-nexus5
Cc: rmcilroy@chromium.org
Status: Available (was: Untriaged)
Cc: mythria@chromium.org
Blocking: 698746
Some of this regression is due to the 40% regression in the splay benchmark due to the timing of the GC. In Octane, we have a warmup phase that is not measured and an actual run that is measured. With Ignition + TF, the GC happens in the actual run as opposed to the warmup run in FCG + Crankshaft case. With Ignition, we actually allocate lesser memory and hence the GC happens later.

Some of it is also due to the slower compilation times. When we disable concurrent compilation, the scores actually improve slightly. 
Project Member

Comment 7 by 42576172...@developer.gserviceaccount.com, Mar 13 2017

Mergedinto: 700364
Status: Duplicate (was: Available)

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

Suspected Commit
  Author : Michael Hablich
  Commit : 7c3936c67a6361371d9a00d83e2c03ecc603ce76
  Date   : Thu Mar 09 08:05:42 2017
  Subject: Version 5.9.34.1 (Turn on I+TF)

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : octane
  Metric       : Total/Score
  Change       : 20.97% | 2916.83333333 -> 2305.16666667

Revision                           Result                  N
chromium@455716                    2916.83 +- 122.453      6      good
chromium@455718                    2917.5 +- 198.039       6      good
chromium@455719                    2912.33 +- 156.248      6      good
chromium@455719,v8@7c3936c67a      2389.0 +- 195.786       6      bad       <--
chromium@455719,v8@fbffc377e3      2336.67 +- 152.883      6      bad
chromium@455720                    2358.17 +- 95.1884      6      bad
chromium@455724                    2364.0 +- 84.0476       6      bad
chromium@455732                    2321.67 +- 118.234      6      bad
chromium@455748                    2305.17 +- 129.286      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 octane

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8985244821790569984

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5462137184452608


| 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!
Cc: hjd@chromium.org
 Issue 700904  has been merged into this issue.
Cc: mvstan...@chromium.org chromeos-performance-bugs@google.com bccheng@chromium.org cros-perf-detectives@google.com deanliao@chromium.org
 Issue 703018  has been merged into this issue.

Sign in to add a comment