All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=909628

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=064103660dab74eae0641e79a4cceba71ea23840b27f5324bb1ad9319cb1e3e1


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

Android Nexus5X WebView Perf
Android Nexus6 WebView Perf

rendering.mobile - Benchmark documentation link:
  https://bit.ly/rendering-benchmarks

system_health.memory_mobile - Benchmark documentation link:
  https://bit.ly/system-health-benchmarks

speedometer - Benchmark documentation link:
  None

loading.mobile - Benchmark documentation link:
  https://bit.ly/loading-benchmarks

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/179e899fe40000

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/179e899fe40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/145f16b4140000

Loading triage:

Kick another bisect for the test that showed the largest regression. But actually several tests were regressed, and the others were improved in that revision range. If we knew the root cause CL, it would help to consider if this regression could be acceptable or not.

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/145f16b4140000

Deadline exceeded while waiting for HTTP response from URL: https://chromium.googlesource.com/chromium/src/+log/0377e72dd246e3b00308e1c92f97a016928a0135..b261389bd87f3fe77d9a6ea1a321866bf46f7bed?format=JSON

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14be9418140000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/14be9418140000

The swarming task expired. The bots are likely overloaded, dead, or misconfigured.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/111176f4140000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/111176f4140000

Use a smaller tolerance when chopping conics to quads by brianosman@google.com
https://skia.googlesource.com/skia/+/e3deee131979b5e2c6ee07cad59a06c247a40d91
frame_times: 28.67 → 30.34 (+1.665)

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

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks

I believe that my CL would trigger a regression on that test page, but I also think it's justified. That page draws a large number of bouncing circles. Prior to my CL, none of them actually looked like circles, because we weren't subdividing the geometry enough. So they're slower now, but at least they look correct. (You can actually see the improvement by comparing http://themaninblue.com/experiment/AnimationBenchmark/canvas/ with stable vs. Canary). Also, see  crbug.com/888453