Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970870628890700800

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970869883139824304

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

Hi chrishtr@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 : Chris Harrelson
  Commit : 602f4cdf9e0944a68fe91d12ead5b2ece186b8e3
  Date   : Sat Jun 24 01:00:35 2017
  Subject: Account for perspective of the scroller when determining a transformed

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : thread_times.tough_compositor_cases
  Metric       : thread_GPU_cpu_time_per_frame/http___jsbin.com_giqafofe_1_quiet?JS_POSTER_CIRCLE
  Change       : 39.03% | 9.96985073561 -> 13.8613815044

Revision             Result                   N
chromium@482013      9.96985 +- 0.252875      6      good
chromium@482064      10.1929 +- 0.348248      6      good
chromium@482090      10.024 +- 0.281492       6      good
chromium@482097      10.1218 +- 0.322062      6      good
chromium@482100      10.0897 +- 0.252418      6      good
chromium@482101      13.5371 +- 0.31756       6      bad       <--
chromium@482102      13.4474 +- 0.189993      6      bad
chromium@482103      13.4543 +- 0.421731      6      bad
chromium@482115      13.8614 +- 0.461133      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=http...jsbin.com.giqafofe.1.quiet.JS.POSTER.CIRCLE thread_times.tough_compositor_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970870628890700800


For feedback, file a bug with component Speed>Bisection

=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Chris Harrelson
  Commit : 602f4cdf9e0944a68fe91d12ead5b2ece186b8e3
  Date   : Sat Jun 24 01:00:35 2017
  Subject: Account for perspective of the scroller when determining a transformed

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : thread_times.tough_compositor_cases
  Metric       : thread_raster_cpu_time_per_frame/http___jsbin.com_giqafofe_1_quiet?JS_POSTER_CIRCLE
  Change       : 243.16% | 0.404467317207 -> 1.38797801112

Revision             Result                     N
chromium@482013      0.404467 +- 0.0177418      6      good
chromium@482064      0.410391 +- 0.0248079      6      good
chromium@482090      0.402025 +- 0.019443       6      good
chromium@482097      0.4007 +- 0.0205677        6      good
chromium@482100      0.402955 +- 0.0241359      6      good
chromium@482101      1.43686 +- 0.101429        6      bad       <--
chromium@482102      1.39923 +- 0.0653857       6      bad
chromium@482103      1.43863 +- 0.160925        6      bad
chromium@482115      1.38798 +- 0.135012        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=http...jsbin.com.giqafofe.1.quiet.JS.POSTER.CIRCLE thread_times.tough_compositor_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970869883139824304


For feedback, file a bug with component Speed>Bisection

My CL fixed a bug in which the size of the scroller was not big enough to contain
the 3D content. Now it is correct. WontFix.

Thanks Chris!

Annie, is there any way to retrospectively add alert marks to benchmark timelines?  I would like to tag regressions like this so we know they've been looked at / are OK.

This would be useful for benchmarks that aren't monitored, or for which we don't monitor sub-tests.