All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=768335

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=4ab1dbc7915504dff4dff55f476369311df5452a6e95a03a59f0ca0621362970


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

chromium-rel-mac-retina

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8967482512088277568

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

Hi miu@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 : Yuri Wiitala
  Commit : a757f6ec69794ab39a3d7f3e18d718bc861252f9
  Date   : Thu Sep 21 23:22:44 2017
  Subject: GLHelper: Reusable fixed-ratio scaling and YUV-readback pipelines.

Bisect Details
  Configuration: mac_retina_perf_bisect
  Benchmark    : performance_browser_tests
  Metric       : CastV2Performance_gpu_30fps_fast/total_latency
  Change       : 1.78% | 26.6598451667 -> 27.1334518333

Revision             Result                   N
chromium@503520      26.6598 +- 0.383975      6      good
chromium@503569      26.7155 +- 0.298375      6      good
chromium@503593      26.7363 +- 0.553344      9      good
chromium@503605      26.5791 +- 0.106534      6      good
chromium@503607      26.7184 +- 0.226695      6      good
chromium@503608      27.1629 +- 0.290749      6      bad       <--
chromium@503611      27.1568 +- 0.398074      6      bad
chromium@503617      27.1335 +- 0.261116      6      bad

To Run This Test
  ./src/out/Release/performance_browser_tests --test-launcher-print-test-stdio=always --enable-gpu

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8967482512088277568


For feedback, file a bug with component Speed>Bisection

Hmm...Interesting that it only regressed on Mac. FWIW, a lot of things are moving around under the umbrella of a rather large project. It's likely the new "arrangement," at project completion, will result in overall improved performance.

Given the the regression is only 1ms of latency, and it's unlikely this will translate into any kind of user-perceivable regression, I'd like to wait on this and revisit later to see if it remains an issue. Marking this bug as blocked on  bug 754872 .

Later work improved performance. Closing/Obsolete.