All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=762205

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=71bc7625424b5fd027b9178890e09a6385c602178348c05e4924d3b8fee165a2


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

chromium-rel-win7-gpu-ati

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8969256474061861200

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

Hi mattm@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 : Matt Mueller
  Commit : ad563205f554412bd207e0dcd9337abc8d13ed9d
  Date   : Fri Aug 18 03:37:24 2017
  Subject: Make CertVerifyProcNSS use_byte_certs-friendly.

Bisect Details
  Configuration: winx64ati_perf_bisect
  Benchmark    : blink_perf.shadow_dom
  Metric       : v0-changing-classname-without-shadow-dom/v0-changing-classname-without-shadow-dom
  Change       : 6.69% | 84.9535 -> 90.6348333333

Revision             Result                   N
chromium@494175      84.9535 +- 1.8717        6       good
chromium@494832      86.915 +- 2.20508        6       good
chromium@495160      86.2447 +- 1.2593        6       good
chromium@495324      87.7555 +- 0.906503      6       good
chromium@495406      88.4587 +- 1.67723       6       good
chromium@495447      88.7719 +- 1.45426       9       good
chromium@495450      87.5181 +- 3.08079       14      good
chromium@495451      88.4932 +- 2.33559       14      bad       <--
chromium@495453      88.788 +- 4.5683         14      bad
chromium@495458      89.6252 +- 1.80132       9       bad
chromium@495468      90.016 +- 1.25468        6       bad
chromium@495488      90.6348 +- 2.26934       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 blink_perf.shadow_dom

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8969256474061861200


For feedback, file a bug with component Speed>Bisection

I suspect the bisect is incorrect.

Annie: There's a bump in the ref, but it doesn't coincide exactly with the regression. Is it possible this isn't a real regression?

Also, the bisect decision looks a bit funny considering some of the "good" results are worse than the culprit. I suppose it is within the confidence interval, so I will kick off another bisect with more iterations.

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8969178463223632048

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

Hi stevenjb@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 : Steven Bennetts
  Commit : 2026190e15abe0c26a2bb97c028eba1ff98ac43d
  Date   : Mon Aug 21 23:59:45 2017
  Subject: Settings: Move settings strings to settings_strings.grdp

Bisect Details
  Configuration: winx64ati_perf_bisect
  Benchmark    : blink_perf.shadow_dom
  Metric       : v0-changing-select-without-shadow-dom/v0-changing-select-without-shadow-dom
  Change       : 8.95% | 252.170833333 -> 274.749833333

Revision             Result                  N
chromium@494003      252.171 +- 5.23414      6      good
chromium@495480      258.755 +- 2.65586      6      good
chromium@495849      268.944 +- 6.59064      6      good
chromium@496034      270.625 +- 5.38729      6      good
chromium@496126      267.835 +- 3.80558      6      good
chromium@496129      271.359 +- 4.7532       6      good
chromium@496131      270.747 +- 3.12207      6      good
chromium@496132      277.368 +- 3.34831      6      bad       <--
chromium@496138      284.742 +- 4.00944      6      bad
chromium@496149      282.312 +- 4.21189      6      bad
chromium@496172      274.709 +- 4.06405      6      bad
chromium@496218      279.721 +- 5.68442      6      bad
chromium@496956      274.75 +- 2.35106       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 blink_perf.shadow_dom

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8969178463223632048


For feedback, file a bug with component Speed>Bisection

Moving strings from one .grd to another also doesn't seem like a likely culprit :)

SInce this only happened on one config, and bisect isn't reproing well, it's likely a hardware issue.