All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=814705

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=fd0cb15ab26521c482aef7d44ef7daea4fb7a58f3c706e90b74e65ade220efd3


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

android-webview-nexus6

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/148bfdef840000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/126ecc01440000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/126ecc01440000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12c3d247440000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/12c3d247440000

[heap-profiler] Merge embedder nodes and V8 wrapper nodes. by ulan@chromium.org
https://chromium.googlesource.com/v8/v8/+/252e8a1c2d629ef5300a3b4610e626b71e75973b

Avoid creating AXObject at unsafe times by dmazzoni@chromium.org
https://chromium.googlesource.com/chromium/src/+/eebef6ebc57245087743bcdc5b99dfa106843109

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

It looks like pinpoint didn't actually correctly narrow it down, it was unable to do a build.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14ce83c9c40000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/14ce83c9c40000

The swarming task 3d174887ce6e2f10 expired. The bots are likely overloaded or dead, or may be misconfigured.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14e0a0fbc40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/14e0a0fbc40000

WebView: enable cc::switches::kCheckDamageEarly by jamwalla@chromium.org
https://chromium.googlesource.com/chromium/src/+/05381aa3a9570439a16774f120bc3c0d20c9de81

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

not sure what this is measuring. Each BeginMainFrame takes 200+ms in that test, but whatever is being measured regressed from 100 to 130ms

and I don't see anything interesting in the trace

+senorblanco who owns the test

I think this is another test that takes an average over the number of frames, and this CL reduces the number of frames that occur. Tentatively closing as WontFix.

See the benchmark https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/web_perf/metrics/smoothness.py which reports mean_frame_time, and  issue 814697 , which we closed for the same reason

frame_times is essentially milliseconds per frame. Reducing the denominator (number of frames rendered) should also reduce the numerator, leaving the average roughly the same. I think we should dig a little deeper.

BTW you can run the test yourself by visiting https://testdrive-archive.azurewebsites.net/performance/chalkboard/Default.html (on Android, you may have to force on gpu rasterization in about:flags).

Is the regression only on Nexus 6, or does it repro on other devices?

Is it possible that we're no longer getting GPU raster in this case? Or is it possible that we're not getting MSAA where we were getting before? (--show-fps-counter should show this: either you'll see "gpu raster" or "MSAA"). Those would explain a large frame_times delta.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12af2fe6640000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/12af2fe6640000

All of the runs failed. The most common error (20/20 runs) was:
BuildError: Build failed: BUILD_FAILURE

It mostly recovered.  I don;t hold out much hope of fixing this now after so much time.