All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=716394

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg0rHGqwoM


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

android-webview-nexus6

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8981077978884949536

=== Auto-CCing suspected CL author nednguyen@google.com ===

Hi nednguyen@google.com, 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 : nednguyen
  Commit : c5cc2a4f4e1ae4dfd1459cba00e7d7a217bc1f70
  Date   : Thu Apr 27 11:25:47 2017
  Subject: Support tracing metrics for measureTime & measureFrameTime methods in blink_perf (Reland)

Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : blink_perf.layout
  Metric       : character_fallback_aat/character_fallback_aat
  Change       : 1900.85% | 0.0247777777778 -> 0.441111111111

Revision             Result                      N
chromium@467627      0.0247778 +- 0.0320867      9       good
chromium@467633      0.0872857 +- 0.604054       14      good
chromium@467636      0.0203333 +- 0.0023094      6       good
chromium@467637      0.368 +- 0.464711           6       bad       <--
chromium@467638      0.441111 +- 0.726242        9       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 blink_perf.layout

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8981077978884949536

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5251703579869184


| 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!

 Issue 718035  has been merged into this issue.

I wasn't expected that CL to cause huge regression on blink_perf.layout. I will look at this next week.

Ok, turn out that this is because these tests are too fast that just adding a few js method for emitting trace event around the test run would regress it significantly. I will tweak the trace event emitting code.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5f01ea694ea9f77fe1424d8fefac44a28e3d297e

commit 5f01ea694ea9f77fe1424d8fefac44a28e3d297e
Author: nednguyen <nednguyen@google.com>
Date: Thu May 04 14:44:50 2017

Move trace event emitting calls to outside of test run time measurement

This removes the overhead of emitting trace events from run time
measurement.
On tests like Layout/character_fallback_aat.html, this reduces the
test's duration 3x (0.015ms --> 0.005ms)

BUG= 716394 ,  701059 
TBR=wangxianzhu@chromium.org

Review-Url: https://codereview.chromium.org/2857403002
Cr-Commit-Position: refs/heads/master@{#469329}

[modify] https://crrev.com/5f01ea694ea9f77fe1424d8fefac44a28e3d297e/third_party/WebKit/PerformanceTests/resources/runner.js

The graph is recovered: https://chromeperf.appspot.com/report?sid=57d1f4e36b8b74f6a9c05f95dbb002d912df026f5f77c7825bbd8288de685d3d&start_rev=465987&end_rev=469615 

I filed another separate bug ( issue 718836 ) for regressions to blink_perf.paint & page_cycler_v2.