New issue
Advanced search Search tips

Issue 649109 link

Starred by 1 user
Status: WontFix
Owner:
ossu@chromium.org
Closed: Sep 2016
Cc:
kwiberg@chromium.org
hlundin@chromium.org
terelius@chromium.org
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

5.8% regression in webrtc_perf_tests at 14300:14300

Project Member Reported by hlundin@chromium.org, Sep 21 2016 
See the link to graphs below.

Comment 1 by hlundin@chromium.org, Sep 21 2016 

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=649109

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg-YT15ggM


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

webrtc-android-tests-nexus5

Comment 2 by hlundin@chromium.org, Sep 21 2016

Cc: terelius@chromium.org hlundin@chromium.org kwiberg@chromium.org
Owner: ossu@chromium.org
ossu: looks like your change "Added a ParsePayload method to AudioDecoder" took a toll on performance. https://chromium.googlesource.com/external/webrtc/+/61a208b1b8e88716747971fe4ba1da8ddf521bb1

Comment 3 by ossu@chromium.org, Sep 22 2016 

Looking at the plot, it seems to have settled down at about 2% higher than before that CL range. Since we're doing slightly more after that change I think it's reasonable. If I were to hazard a guess, it would be the increased number of allocations that causes this. We're looking at ways of improving that in future revisions. I also have some ideas on how to reduce the number of allocations in other ways. Suggest we leave this for now.

Comment 4 by hlundin@chromium.org, Sep 22 2016

Status: WontFix (was: Assigned)
Fair enough.
►

Sign in to add a comment