All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=705373

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgrPXbsQkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgrL_y9gsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgzMiY3QoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgrLHktwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgrPiGrAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgrNST4QsM


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

android-nexus5
android-nexus5X

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8983990269345936608

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

Hi rmcilroy@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 : Ross McIlroy
  Commit : f9c4085904ec8579549df3ada44b0cc9bfe47acf
  Date   : Tue Mar 21 08:15:10 2017
  Subject: [TurboFan] Use temporary zone for effect linearization schedule.

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : v8.mobile_infinite_scroll_tbmv2
  Metric       : memory:chrome:all_processes:reported_by_chrome:v8:effective_size_max/pinterest
  Change       : 143.79% | 67974828.8571 -> 165713364.889

Revision                           Result                      N
chromium@458410                    67974829 +- 193573354       14      good
chromium@458430                    48910673 +- 11558357        6       good
chromium@458440                    46380854 +- 8940595         6       good
chromium@458441                    47031289 +- 10035449        6       good
chromium@458441,v8@25f86ffea5      47409361 +- 8921310         6       good
chromium@458441,v8@0a7d313800      47231280 +- 10169485        6       good
chromium@458441,v8@f9c4085904      189607985 +- 9138033        6       bad       <--
chromium@458441,v8@48e08319ca      183555758 +- 24775344       6       bad
chromium@458441,v8@6ec45d7495      191429137 +- 12164370       6       bad
chromium@458442                    192023449 +- 10125946       6       bad
chromium@458443                    189790065 +- 8516033        6       bad
chromium@458445                    193383955 +- 9716453        6       bad
chromium@458449                    165713365 +- 109724828      9       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=pinterest v8.mobile_infinite_scroll_tbmv2

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8983990269345936608

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6692757973762048


| 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!

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8983974701041777632

Hmm, this seems very surprising. That CL should have improve zone memory usage, and shouldn't have had any impact on the heap. Looking at the graph [1] it looks like an increase in the size of the old space. It also doesn't appear on any other Android device (only the Nexus 5X bot) and the graph appears somewhat bimodal (there have been previous regressions and recoveries to the same level.

My guess is that this is a GC timing issue. Hannes, does that sound reasonable?

I've kicked off another bisect to make sure. 

[1] https://chromeperf.appspot.com/report?sid=9625ae87e77e31e93e43c5f0cda1e073477041fa1142e1dee2b9f16504fabdcc&rev=458483

=== BISECT JOB RESULTS ===
NO Perf regression found

Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : v8.mobile_infinite_scroll_tbmv2
  Metric       : memory:chrome:renderer_processes:reported_by_chrome:v8:heap:old_space:effective_size_max/facebook

Revision             Result                    N
chromium@458401      56729885 +- 31637660      17      good
chromium@458483      56634039 +- 24762774      14      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=facebook v8.mobile_infinite_scroll_tbmv2

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8983974701041777632

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5835869719625728


| 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!

Hmm, if it is old space then it might be GC timing.