All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=714899

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg4unclggM


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

android-nexus5

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8981385286154955264

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

Hi senorblanco@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 : senorblanco
  Commit : 8e75e9c587155933fe3e775605e18111a6cc2b9c
  Date   : Tue Apr 18 17:30:39 2017
  Subject: cc: remove the veto-to-software for complex content.

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_os:gpu_memory:proportional_resident_size_avg/load_social/load_social_tumblr
  Change       : 15.35% | 74383360.0 -> 85798912.0

Revision             Result                   N
chromium@465195      74383360 +- 92681.9      6      good
chromium@465245      74394283 +- 59825.9      6      good
chromium@465270      74383360 +- 0.0          6      good
chromium@465277      74372437 +- 59825.9      6      good
chromium@465278      74394283 +- 59825.9      6      good
chromium@465279      85777067 +- 75674.5      6      bad       <--
chromium@465280      85787989 +- 59825.9      6      bad
chromium@465283      85787989 +- 59825.9      6      bad
chromium@465295      85798912 +- 0.0          6      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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=load.social.tumblr system_health.memory_mobile

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981385286154955264

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5890678682288128


| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Speed>Bisection.  Thank you!

This seems to be limited to tumblr; will dig deeper.

ericrk@: are these comparable to what has been seen on other platforms when enabling GPU raster? (I'm guessing that we didn't have these metrics when Ganesh was first enabled on Android, so it might be hard to compare).

Also strange that the veto-to-software was happening on Nexus 5; we should be vetoing to MSAA on that device, IIRC.

No corresponding regression on Nexus7v2, which has the same GPU IIRC.

FWWIW, I probably wouldn't worry too much about this - we expect to see GPU memory go up and system memory go down. Looking at some other metrics: https://chromeperf.appspot.com/report?sid=f1bbc845a9cc0c732d13abc15001d68c93e3944d5cd89e26978d95675de60b82&rev=465295

10MB increase in GPU memory
*but*
10MB decrease in resident size
10MB decrease in peak resident size
6MB decrease in private dirty size
no change in proportional resident size

Also note that, depending on the driver, GPU memory may partially count in both GPU and resident size.

OK, I'll WontFix it. Feel free to reopen if you disagree.

Also, thanks for your investigation, Eric!

Issue 712455 has been merged into this issue.