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

Issue 830778 link

Starred by 1 user
Status: Assigned
Owner:
g...@chromium.org
Cc:
█ afdo-chr...@skia-buildbots.google.com.iam.gserviceaccount.com
primiano@chromium.org
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

1.9% regression in blink_perf.image_decoder at 549070:549078

Project Member Reported by primiano@chromium.org, Apr 9 2018 
See the link to graphs below.
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Apr 9 2018 

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=830778

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=cc71d03afa86da8e55770e5bcf88101e34872108261cf180d71bb0ad7f262129


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

android-nexus6
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Apr 9 2018

Cc: afdo-chr...@skia-buildbots.google.com.iam.gserviceaccount.com
Status: Assigned (was: Untriaged)
📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/15a5c3ecc40000

Roll AFDO from 67.0.3379.0_rc-r1 to 67.0.3390.0_rc-r1 by afdo-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/c1e3fa4c733bf46e284e257377b5443bdb549910

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Comment 4 by sullivan@google.com, May 3 2018

Owner: g...@chromium.org
gbiv: As sheriff for the roll, can you take a look?

Comment 5 by g...@chromium.org, May 7 2018 

Yup; thanks!

FWIW, it looks like this *may* already be resolved by a later AFDO roll. In particular, the `550108 - 550165` datapoint -- which includes an AFDO roll -- looks close to the pre-regression values, but there's way too much noise after it at the moment for me to say for sure.

I'll keep an eye on it, in any case.
►

Sign in to add a comment