New issue
Advanced search Search tips

Issue 914365 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner:
Closed: Dec 14
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

6.5%-36.5% regression in frame_times,percentage_smooth,tasks_per_frame_total_all at 615562:615676

Project Member Reported by 42576172...@developer.gserviceaccount.com, Dec 12

Issue description

See the link to graphs below.
 
All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=914365

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=b95cb7620ac5a71fd41ae6f7dbd9b81aa114a23d68ff9ca5fe69972d22cfeef0


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

linux-perf
mac-10_12_laptop_low_end-perf

rendering.desktop - Benchmark documentation link:
  https://bit.ly/rendering-benchmarks
Cc: pbos@chromium.org eseckler@google.com
Owner: eseckler@google.com
Status: Assigned (was: Untriaged)
📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/14c49d82140000

Reland "Disable the new tab-loading animation" by pbos@chromium.org
https://chromium.googlesource.com/chromium/src/+/ad12d44124812f22484eb4d123e4a3b4aaac93b9
tasks_per_frame_total_all: 50.86 → 49.17 (-1.695)

trace writer: Send patches after completing fragmented packet by eseckler@google.com
https://android.googlesource.com/platform/external/perfetto/+/15697965801191b657fa3213139ff8532b2e6cc7
tasks_per_frame_total_all: 48.95 → 53.25 (+4.308)

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

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks
 Issue 914368  has been merged into this issue.
Cc: primiano@chromium.org skyos...@chromium.org oysteine@chromium.org
Components: Speed>Tracing
Labels: Perfetto
These changes in metrics look reasonable. The perfetto CL may cause us to see some  trace events that were previously dropped by the tracing backend. (That's because we probably never received patches for some chunks before the CL and thus couldn't read the committed data. The CL sends these patches earlier, so makes it more likely that we can emit the events.)

From looking at the before/after traces for some of these metrics changes, I could see that some of the previously missing threads (e.g. some CompositorTileWorkers) now started appearing.

We may still continue to be missing some events though, and issues  907069  and  907067  track the work to fix that.
Status: WontFix (was: Assigned)
Cc: lalitm@google.com sullivan@google.com
 Issue 914540  has been merged into this issue.
Cc: kouhei@chromium.org eseckler@chromium.org
 Issue 917267  has been merged into this issue.
 Issue 917225  has been merged into this issue.

Sign in to add a comment