All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=758276

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=b3778272839b14a829fc0201da96d0875dc02ee7d0e002ea15bfaadfa5145d41


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

android-webview-nexus6

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970449101609667280

=== Auto-CCing suspected CL author fmalita@chromium.org ===

Hi fmalita@chromium.org, the bisect results pointed to your CL, please take a look at the
results.


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Florin Malita
  Commit : b9e02f8f670683edaac8b9e6c7a6ce23a5a71194
  Date   : Fri Aug 18 16:57:23 2017
  Subject: Disable Skia raster pipeline for image tiling

Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : thread_times.key_mobile_sites_smooth
  Metric       : thread_other_cpu_time_per_frame/http___cuteoverload.com
  Change       : 78.51% | 2.90236356249 -> 5.18102317909

Revision             Result                   N
chromium@495551      2.90236 +- 0.572048      6      good
chromium@495579      2.9575 +- 0.908143       6      good
chromium@495581      2.96438 +- 0.799232      6      good
chromium@495582      2.9194 +- 0.671915       6      good
chromium@495583      5.22518 +- 0.446084      6      bad       <--
chromium@495586      5.14076 +- 0.252653      6      bad
chromium@495593      5.1867 +- 0.319815       6      bad
chromium@495607      5.26707 +- 0.160473      6      bad
chromium@495662      5.18102 +- 0.259285      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=http...cuteoverload.com thread_times.key_mobile_sites_smooth

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970449101609667280


For feedback, file a bug with component Speed>Bisection

This is kinda weird: https://chromium-review.googlesource.com/c/chromium/src/+/620528 is a revert, landed to address two perf bugs -  issue 753905  and  issue 744674 .

753905 in particular, was a serious raster time regression on Android/N6, same platform as the graph linked here.  So I'm kinda skeptical that the change can produce opposing effects on the same HW.

The metric regressed here is thread_other_cpu_time_per_frame, and it's unclear to me what exactly is included there.

I'm actually starting to suspect this is one of those weird perf inversions, where because we're rasterizing faster we have more CPU time to dedicate to other per/frame tasks - so unrelated metrics appear to regress.  I've actually seen another case of this recently, also for a smoothness metric.

This is the raster time improvement linked to https://chromium-review.googlesource.com/c/chromium/src/+/620528: 
52f7558b3e39560d90b6f9d742ed2d1bc47fb37a2abdd5605ed75c3db5cacf40

See also https://bugs.chromium.org/p/chromium/issues/detail?id=752457#c5

Based on the above, I'll go out on a limb and call this a pseudo-regression in an un-normalized/frame-rate-dependent metric, unrelated to rasterization (thread_other_cpu_time_per_frame).

 Issue 759590  has been merged into this issue.