All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=768554

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=75aa285e485e347f48a7822d08cf4ca6c053fc775862b13484b07edfe56b0051


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

chromium-rel-win7-gpu-ati

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8967444281562102624

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

Hi mtklein@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 : Mike Klein
  Commit : 65e706379657d64371012c2e95ccf1354d741145
  Date   : Thu Sep 21 18:33:23 2017
  Subject: Don't call GrBuffer::map() in createPatternedIndexBuffer.

Bisect Details
  Configuration: winx64ati_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstMeaningfulPaint_avg/cold/uol.com.br
  Change       : 25.89% | 432.124571426 -> 544.004222227

Revision                             Result                  N
chromium@503499                      432.125 +- 487.604      14      good
chromium@503549                      395.124 +- 44.1167      6       good
chromium@503574                      394.62 +- 46.6765       9       good
chromium@503587                      408.279 +- 259.476      14      good
chromium@503590                      391.359 +- 20.3353      6       good
chromium@503590,skia@ebf648e57a      399.064 +- 46.6661      9       good
chromium@503590,skia@65e7063796      433.478 +- 102.607      14      bad       <--
chromium@503590,skia@780f5bfdfe      463.103 +- 184.375      9       bad
chromium@503591                      427.004 +- 96.5354      9       bad
chromium@503592                      456.027 +- 83.2455      6       bad
chromium@503593                      420.261 +- 87.0789      14      bad
chromium@503599                      544.004 +- 661.622      9       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=uol.com.br loading.desktop

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8967444281562102624


For feedback, file a bug with component Speed>Bisection

I intend to revert this in the next day or so.  (An immediate revert would muck with our Mac TSAN bots.)

A fix for this is in Skia but not quite yet rolled into Chromium.

Charts are back down.