All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=775299

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=ca2edff3a082a43231eed8cb0163757ed6831dd4798a32d5805bd0221a63ca66


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

chromium-rel-mac-retina

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8965532693360390288

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

Hi dalecurtis@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 : Dale Curtis
  Commit : 5341599c0b2b5a74f62a7734d592d7faae9983c8
  Date   : Sat Oct 14 01:01:26 2017
  Subject: Add support for discarding video frames post-decode.

Bisect Details
  Configuration: mac_retina_perf_bisect
  Benchmark    : media.desktop
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/video.html?src_tulip2.vp9.webm_seek
  Change       : 27.68% | 30800877.3333 -> 39325677.3333

Revision             Result                    N
chromium@508851      30800877 +- 3832625       6      good
chromium@508888      32562157 +- 11202991      9      good
chromium@508893      30677997 +- 5923973       6      good
chromium@508894      32261784 +- 4564054       6      good
chromium@508895      40144877 +- 3601997       6      bad       <--
chromium@508896      39052611 +- 3888684       6      bad
chromium@508898      37741891 +- 4978508       6      bad
chromium@508907      40035651 +- 3018097       6      bad
chromium@508925      39325677 +- 8861508       6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=video.html.src.tulip2.vp9.webm.seek media.desktop

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8965532693360390288


For feedback, file a bug with component Speed>Bisection

 Issue 775298  has been merged into this issue.

Ah, hmm, I guess it's possible we're now eliding a trampoline that previously allowed a vended by useless frame to be reclaimed before the next decode cycle came in. I.e. before it was:

OutputReady -> PostTask(Is frame good?) -> PostTask(Return Frame)
OutputReady -> Is frame good? -> PostTask(ReturnFrame)

I'll play with it locally but I suspect this is just a change in timing so probably WontFix.