All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=758593

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=550760716b0d3bdc0f059c09005689f43af569a6ee7c2dbbf84cae2383762309


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

android-one

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970368705472231456

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

Hi ericrk@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 : Eric Karl
  Commit : 2fc4409b9b179323f14888b74968b88a7432ee07
  Date   : Fri Aug 18 18:48:39 2017
  Subject: Lower GPU SharedMemoryLimits on Low-End and Consolidate Logic

Bisect Details
  Configuration: android_one_perf_bisect
  Benchmark    : smoothness.gpu_rasterization.tough_path_rendering_cases
  Metric       : frame_times/MotionMark Canvas Stroke Shapes
  Change       : 33.34% | 393.036789117 -> 524.064488091

Revision             Result                  N
chromium@495474      393.037 +- 52.0979      6      good
chromium@495625      414.113 +- 76.7499      6      good
chromium@495630      392.142 +- 32.5321      6      good
chromium@495631      519.999 +- 107.136      6      bad       <--
chromium@495632      544.733 +- 89.6366      6      bad
chromium@495633      571.452 +- 80.1627      6      bad
chromium@495635      540.552 +- 72.7506      6      bad
chromium@495644      552.016 +- 78.0194      6      bad
chromium@495663      534.929 +- 137.402      6      bad
chromium@495700      520.638 +- 58.0027      6      bad
chromium@495775      524.064 +- 109.837      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=MotionMark.Canvas.Stroke.Shapes smoothness.gpu_rasterization.tough_path_rendering_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970368705472231456


For feedback, file a bug with component Speed>Bisection

 Issue 759681  has been merged into this issue.

 Issue 760157  has been merged into this issue.

ericrk: any update on this bug?

It's not too surprising that a throughput based metric like motionmark's Canvas metric would regress here. 

Note that this is a low-end-only change. Overall, this was a concious decision to reduce memory limits on low-end android, with hopefully minimal impact on real world performance. I ran smoothness.top_25_smooth locally, and saw a small but acceptable impact. This impact is greater, but is on a synthetic benchmark. I haven't seen other alerts from the bots related to this change, so I'm tempted to ignore the impact on this synthetic benchmark.