All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=780510

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=e5c9a40b72c91db3ee9ad91eb8a1bec393936551740990a6f54af9735990ed17


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

chromium-rel-mac12
chromium-rel-mac12-mini-8gb

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8964109333072242720

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

Hi sunnyps@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 : Sunny Sachanandani
  Commit : d2e9756d46f6a324a1e98d3553eb045556a5f4af
  Date   : Mon Oct 30 20:45:44 2017
  Subject: cc: Use CHROMIUM_texture_storage_image for GMB backed texture resources.

Bisect Details
  Configuration: mac_10_12_mini_8gb_perf_bisect
  Benchmark    : media.desktop
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/video.html?src_tulip2.wav_type_audio_seek
  Change       : 25.11% | 14095640.0 -> 17634584.0

Revision             Result                   N
chromium@512508      14095640 +- 1605298      6      good
chromium@512572      14189263 +- 1716135      7      good
chromium@512604      13767960 +- 1310720      8      good
chromium@512612      13658733 +- 1196518      6      good
chromium@512614      13986413 +- 1440799      6      good
chromium@512615      16586008 +- 0.0          7      bad       <--
chromium@512616      17241368 +- 2452133      6      bad
chromium@512620      17241368 +- 1605298      6      bad
chromium@512635      17634584 +- 1172344      5      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.wav.type.audio.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/8964109333072242720


For feedback, file a bug with component Speed>Bisection

 Issue 781250  has been merged into this issue.

 Issue 781250  has been merged into this issue.

 Issue 781250  has been merged into this issue.

Not a real regression, just a memory tracking bug in our instrumentation. See above issue for details.

 Issue 781546  has been merged into this issue.

 Issue 781956  has been merged into this issue.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/105aa2b3a40000

Update: this is still blocked on  issue 781490 , but I might have a fix for that. I suspect issue 863121 is also a manifestation of the same issue i.e. we're using anonymous IOSurfaces for the browser after mac views but the memory infra is double counting that memory.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/17e73843a40000

📍 Job complete. See results below.
https://pinpoint-dot-chromeperf.appspot.com/job/105aa2b3a40000

📍 Job complete. See results below.
https://pinpoint-dot-chromeperf.appspot.com/job/17e73843a40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/13838009a40000

📍 Job complete. See results below.
https://pinpoint-dot-chromeperf.appspot.com/job/13838009a40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11ef9b95a40000

📍 Job complete. See results below.
https://pinpoint-dot-chromeperf.appspot.com/job/11ef9b95a40000

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/4da52fed413909a85bb4e52fc7c82e3a62bec3df

commit 4da52fed413909a85bb4e52fc7c82e3a62bec3df
Author: Sunny Sachanandani <sunnyps@chromium.org>
Date: Wed Jul 25 19:48:45 2018

Fix anonymous GLImageIOSurface memory double counting

Anonymous GLImageIOSurfaces are textures from the client's perspective
so the client adds a shared ownership edge with the GPU process texture
instead of considering it as a GpuMemoryBuffer. However, the service
also adds another shared ownership edge with the GpuMemoryBuffer using
the GMB id at the face_X/level_Y nested level. This adds edges at
different nesting levels which causes double counting by memory-infra.

This change removes the shared ownership edge when the GLImageIOSurface
is anonymous, and replaces that with a nested "anonymous-iosurface"
level. This gets rid of the double counting as the memory is charged to
cc's tile or resource memory due to its higher importance.

R=ericrk,ccameron
BUG= 781490 , 780510 ,863121

Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: Id5b69e38b525d5bfdbb997fec2479b644cfae656
Reviewed-on: https://chromium-review.googlesource.com/1147674
Reviewed-by: ccameron <ccameron@chromium.org>
Reviewed-by: Antoine Labour <piman@chromium.org>
Reviewed-by: Eric Karl <ericrk@chromium.org>
Commit-Queue: Sunny Sachanandani <sunnyps@chromium.org>
Cr-Commit-Position: refs/heads/master@{#578021}
[modify] https://crrev.com/4da52fed413909a85bb4e52fc7c82e3a62bec3df/gpu/command_buffer/service/texture_manager.cc
[modify] https://crrev.com/4da52fed413909a85bb4e52fc7c82e3a62bec3df/gpu/ipc/service/gpu_memory_buffer_factory_io_surface.cc
[modify] https://crrev.com/4da52fed413909a85bb4e52fc7c82e3a62bec3df/gpu/ipc/service/gpu_memory_buffer_factory_io_surface.h
[modify] https://crrev.com/4da52fed413909a85bb4e52fc7c82e3a62bec3df/ui/gl/gl_image_io_surface.mm

The graphs have recovered after the above CL.

Example: https://chromeperf.appspot.com/report?sid=6abefa9838134306447ebb8ced73d526eec0018674346b84681376054eca519d&start_rev=550649&end_rev=578303

Note: The bots have been renamed. The closest bot to chromium-rel-mac12 is probably mac-10_12_laptop_low_end.

Full report: https://chromeperf.appspot.com/group_report?rev=578021