All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=755239

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=c7ceed84ebfb41ca10b2a0c3738cb26989563e8b1b008645f605268f4e46b8e7


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

android-one

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8971260559732741920

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

Hi xidachen@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 : Xida Chen
  Commit : ea20f077cb3981dce10270048decfab2a33e9fcd
  Date   : Thu Aug 10 22:21:06 2017
  Subject: Enable Waterfall testing for reduced soft tile memory experiment

Bisect Details
  Configuration: android_one_perf_bisect
  Benchmark    : smoothness.tough_scrolling_cases
  Metric       : mean_pixels_approximated/text_05000_pixels_per_second
  Change       : 1125.76% | 1.362 -> 16.6948333333

Revision             Result                    N
chromium@493555      1.362 +- 2.46235          6      good
chromium@493557      0.743833 +- 0.739833      6      good
chromium@493558      14.5895 +- 9.13054        6      bad       <--
chromium@493559      13.4085 +- 5.2526         6      bad
chromium@493563      14.9305 +- 3.31819        6      bad
chromium@493571      14.2077 +- 4.16666        6      bad
chromium@493586      12.886 +- 6.256           6      bad
chromium@493616      14.2227 +- 4.43505        6      bad
chromium@493676      16.6948 +- 6.99315        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=text.05000.pixels.per.second smoothness.tough_scrolling_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8971260559732741920


For feedback, file a bug with component Speed>Bisection

fmeawad@: thank you for the report. The CL is an experiment that we are conducting to evaluate whether this is valuable or not for low-end android devices. Once we have more data from UMA, we will make a decision to revert this or not. Thank you.

ericrk@, flackr@: FYI

 Issue 755305  has been merged into this issue.

 Issue 755324  has been merged into this issue.

Issue 755342 has been merged into this issue.

weiliangc@, ajuma@: FYI, this change also caused mean_pixels_checkerboarded to regress by 1000%, see the graph here:
https://chromeperf.appspot.com/report?sid=cd67e0045dc02c8801ec01310d2783479d1a42fa34670fdee9fa73faea32bdb1

I'll be curious to see how this compares when we just look at the stats for just GPU memory constrained devices.

flackr@: the metric is from android-one only, which has 512MB.

 Issue 755613  has been merged into this issue.

 Issue 755763  has been merged into this issue.

 Issue 756110  has been merged into this issue.

 Issue 759685  has been merged into this issue.

Issue 762039 has been merged into this issue.

I have terminated the experiment to reduce soft tile memory to half, and I can see that the graph are recovering.