Project: chromium Issues People Development process History Sign in
New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.
Issue 572525 3.7%-35.6% regression in rasterize_and_record_micro.top_25_smooth at 366902:366904
Starred by 0 users Reported by rschoen@google.com, Dec 27 2015 Back to list
Status: Fixed
Owner:
Closed: Jan 2016
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression

Blocking:
issue 559247



Sign in to add a comment
See the link to graphs below.
 
Owner: tha...@chromium.org
thakis, pretty sure this is e13537fe418eff11d3cab9077f6a647d7c74f103. Was this expected?
Project Member Comment 3 by 42576172...@developer.gserviceaccount.com, Dec 27 2015
Cc: tha...@chromium.org

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

Hi thakis@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 status: Completed
Bisect job ran on: linux_perf_bisect



===== BISECT JOB RESULTS =====
Status: Positive: A suspected commit was found.

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --also-run-disabled-tests rasterize_and_record_micro.key_silk_cases
Test Metric: record_time/record_time
Relative Change: 37.31%
Score: 99.9
Retested CL with revert: Not Implemented.


===== SUSPECTED CL(s) =====
Subject : tcmalloc: Use C++11 atomics where appropriate.
Author  : thakis, thakis@chromium.org
Commit description:

Ports these CLs to tcmalloc:
https://codereview.chromium.org/636783002/
https://codereview.chromium.org/1466833002/ (except mac)

No intended behavior change, but it should remove
the static initializer in atomicops_internals_x86_gcc.h
on Linux.  It's also less code.

BUG=94925,559247

Review URL: https://codereview.chromium.org/1549913002

Cr-Commit-Position: refs/heads/master@{#366904}
Commit  : e13537fe418eff11d3cab9077f6a647d7c74f103
Date    : Sat Dec 26 18:19:59 2015


===== TESTED REVISIONS =====
Depot    Revision Mean Value Std. Dev. Num Values Good?
chromium r366903  0.049198   0.000834  4          good

chromium r366904  0.067552   0.000193  4          bad   <-




| 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

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/5188
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9025260779136510320

Comment 4 by tha...@chromium.org, Dec 27 2015
Cc: jfb@chromium.org
No. Will revert Tuesday. We should probably not ist the _portable atomics until this is understood and fixed then.
Project Member Comment 5 by 42576172...@developer.gserviceaccount.com, Dec 28 2015
Issue 572712 has been merged into this issue.
Project Member Comment 6 by 42576172...@developer.gserviceaccount.com, Dec 29 2015
Issue 572789 has been merged into this issue.
Comment 7 by tha...@chromium.org, Dec 29 2015
Blocking: chromium:559247
Status: Fixed
The recovery after that revert is visible in the graphs: https://chromeperf.appspot.com/group_report?bug_id=572525

Thanks thakis@ :-)
I wonder if it was caused by issue 592903 (additional mfences as was discovered in issue 593344). If so, then this issue only affects Linux?
That's the working theory (and all the graphs above are on linux), but we never investigated in depth.
Sign in to add a comment