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

Issue 884309 link

Starred by 1 user

Issue metadata

Status: Untriaged
Owner: ----
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

2.5%-39.3% regression in rendering.mobile at 591056:591271

Project Member Reported by npm@chromium.org, Sep 14

Issue description

See the link to graphs below.
 
All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=884309

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=173ba4e70c3aa06289b9c043a9e01c1c263d430f1f7fefef45f007af6d366b85


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

Android Nexus6 WebView Perf

rendering.mobile - Benchmark documentation link:
  https://bit.ly/rendering-benchmarks
Cc: chromium...@skia-public.iam.gserviceaccount.com shend@chromium.org
Owner: shend@chromium.org
Status: Assigned (was: Untriaged)
📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/17b16ac3640000

Roll src/third_party/catapult 994bb1389b1d..5f6da8a57db1 (3 commits) by chromium-autoroll@skia-public.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/7b2b2dca23cca0862f674758c9a3933e685c27d5
0.6664 → No values

Revert "Roll src/third_party/catapult 3e071665b9f9..c4d326feb2ff (5 commits)" by shend@chromium.org
https://chromium.googlesource.com/chromium/src/+/393391985b970fe0f33952d108f2470c8d1fdb83
No values → 0.8954

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

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks
Cc: nednguyen@chromium.org
Hi Ned, I got a assigned a bunch of these because I reverted a catapult roll? Do you know what's going on here?
📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/1705af59640000

Roll src/third_party/catapult 994bb1389b1d..5f6da8a57db1 (3 commits) by chromium-autoroll@skia-public.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/7b2b2dca23cca0862f674758c9a3933e685c27d5
22.08 → No values

Revert "Roll src/third_party/catapult 3e071665b9f9..c4d326feb2ff (5 commits)" by shend@chromium.org
https://chromium.googlesource.com/chromium/src/+/393391985b970fe0f33952d108f2470c8d1fdb83
No values → 22.86

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

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks
Status: WontFix (was: Assigned)
Oh I'm guessing what happened is that "No values" meant catapult was broken. Marking as WontFix then?
There's many graphs here and they haven't recovered hmmm... I'm running another bisect to ensure this was just caused by a catapult breakage.
Owner: ----
Status: Untriaged (was: WontFix)
Owner: shend@chromium.org
Status: Assigned (was: Untriaged)
📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/14837483640000

Roll src/third_party/catapult 994bb1389b1d..5f6da8a57db1 (3 commits) by chromium-autoroll@skia-public.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/7b2b2dca23cca0862f674758c9a3933e685c27d5
29.63 → No values

Revert "Roll src/third_party/catapult 3e071665b9f9..c4d326feb2ff (5 commits)" by shend@chromium.org
https://chromium.googlesource.com/chromium/src/+/393391985b970fe0f33952d108f2470c8d1fdb83
No values → 32.49

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

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks
Seems like it?
Cc: dtu@chromium.org
Owner: ----
Status: Untriaged (was: Assigned)
My only concern is that the initial values are different from the last values (after roll plus revert). So some commit in between probably caused a regression but we'd need to manually bisect (including the fever so that the values are actually computed).
Ah fair enough, makes sense.
Looking at before/after traces, I see a lot of new tasks in the browser IO thread. You can see the spike in tasks_per_frame_IO, too, which is not recovered later, e.g. https://chromeperf.appspot.com/report?sid=7ce3ff7570c8c3866bddc7456fd1d772115df5483f973bfe26cad29ee073de75&start_rev=589502&end_rev=595340

I'll start a pinpoint job for tasks_per_frame_IO.
Cc: chiniforooshan@chromium.org
Owner: shend@chromium.org
Status: Assigned (was: Untriaged)
📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/157adeb4e40000

Roll src/third_party/catapult 994bb1389b1d..5f6da8a57db1 (3 commits) by chromium-autoroll@skia-public.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/7b2b2dca23cca0862f674758c9a3933e685c27d5
tasks_per_frame_IO: 2.493 → No values

Revert "Roll src/third_party/catapult 3e071665b9f9..c4d326feb2ff (5 commits)" by shend@chromium.org
https://chromium.googlesource.com/chromium/src/+/393391985b970fe0f33952d108f2470c8d1fdb83
tasks_per_frame_IO: No values → 3.226

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

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks
Owner: ----
Unassigning...see issue 884309.
Status: Untriaged (was: Assigned)
Assigned, but no owner or component? Please find a component and/or owner

Sign in to add a comment