All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=759680

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=cfe73e6bb8e7caa8f1904ff7135bb6fedeab9e69754d2f2640df15236499c90e


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

chromium-rel-win7-gpu-nvidia

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970000166600954032

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

Hi jmadill@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 : Jamie Madill
  Commit : abf66fb35d23734ac969eea4665f755d9f7e45d2
  Date   : Wed Aug 23 18:17:54 2017
  Subject: D3D11: Use dirty bits for applyVertexBuffers.

Bisect Details
  Configuration: winx64nvidia_perf_bisect
  Benchmark    : blink_perf.canvas
  Metric       : putImageData/putImageData
  Change       : 16.74% | 458.636991342 -> 381.847240151

Revision                              Result                  N
chromium@497214                       458.637 +- 25.9457      6      good
chromium@497218                       459.591 +- 9.55308      6      good
chromium@497219                       427.868 +- 20.3804      6      good
chromium@497219,angle@e3d8628d1d      451.384 +- 13.8987      6      good
chromium@497219,angle@abf66fb35d      354.396 +- 20.7476      6      bad       <--
chromium@497219,angle@e080387e43      363.855 +- 26.3851      6      bad
chromium@497220                       358.698 +- 23.7818      6      bad
chromium@497221                       364.31 +- 28.324        6      bad
chromium@497228                       349.152 +- 27.7371      6      bad
chromium@497241                       365.954 +- 15.6961      6      bad
chromium@497267                       369.224 +- 26.6735      6      bad
chromium@497319                       381.847 +- 26.5031      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.canvas

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970000166600954032


For feedback, file a bug with component Speed>Bisection

Apologies I haven't had time to look into this directly. Not sure what's happening here, my CL is supposed to be a speed improvement. It was also disabled in Canary the re-enabled yesterday.

jmadill@ have you had time to look into this? Did the graph move with the disabling / reenabling you mention in #4?

Sorry, haven't had time, I haven't been into the office since I've been sick.

The graph did not appear to move when my CL was disabled and re-enabled. It's possible it is something to do with a related change in the CL other than enabling the optimization. We can always re-run the bisect.

I'd always welcome help if someone wants to try and repro (build Chrome, set ANGLE to the bad revision, test, set angle to the revision before that, test again).

Maybe someone else could help here?

Adding reviewer geofflang: can you help out?

jmadill: any update on this bug?

No updates. Maybe one of the owners of the benchmark could help?

Unfortunately won't be able to help with this in the short term.

This metric has suffered from two regressions, this seems to be the minor one (but most recent). CCing metric owner.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/118dd4cb440000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/118dd4cb440000

WontFix-ing since this is over 6 months old and no progress.