All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=776765

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=992bd8529ad1acffdf59971ba6351233e741210e7da3ac0587e0efc96aa5bb9a


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

chromium-rel-win7-gpu-ati
win-high-dpi

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8965206196166672496

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

Hi bsalomon@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 : Brian Salomon
  Commit : 43f8bf0f784f4182ed0fca9053ecf570caf7ad70
  Date   : Wed Oct 18 12:55:06 2017
  Subject: Move clear-as-draw workaround to GrGLGpu and expose via GrContextOptions.

Bisect Details
  Configuration: winx64_high_dpi_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstContentfulPaint_avg/cold/Kakaku
  Change       : 10.16% | 1318.52222222 -> 1452.51033333

Revision                             Result                  N
chromium@509839                      1318.52 +- 155.415      9      good
chromium@509889                      1306.66 +- 123.193      6      good
chromium@509890                      1289.65 +- 101.583      6      good
chromium@509890,skia@57caa660c0      1312.87 +- 102.189      6      good
chromium@509890,skia@43f8bf0f78      1489.3 +- 138.285       6      bad       <--
chromium@509890,skia@43938b8533      1453.04 +- 142.004      6      bad
chromium@509891                      1469.55 +- 116.128      6      bad
chromium@509893                      1464.38 +- 100.266      6      bad
chromium@509896                      1460.76 +- 111.141      6      bad
chromium@509902                      1465.2 +- 132.198       6      bad
chromium@509914                      1477.65 +- 139.682      6      bad
chromium@509938                      1452.51 +- 118.263      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=Kakaku loading.desktop

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8965206196166672496


For feedback, file a bug with component Speed>Bisection

I believe this is showing the extra shader compile that occurs for the new way the workaround is implemented. We got some fairly big gains out of this on canvas perf tests.

 Issue 777794  has been merged into this issue.