All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=630461

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgqoePzQsM


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

chromium-rel-mac-retina

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/9006484861874608336

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

Hi amistry@chromium.org, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Use ChannelMojo for GpuChannels.
Author  : amistry
Commit description:
  
BUG= 604282 
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2127693002
Cr-Commit-Position: refs/heads/master@{#406719}
Commit  : cef5cfa8e1e017d0818979d820ea77023d50fd11
Date    : Thu Jul 21 00:13:59 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N  Good?
chromium@406716  5606.26  382.76   8  good
chromium@406718  5767.26  75.8227  5  good
chromium@406719  4669.57  82.0282  8  bad    <--
chromium@406722  4675.3   48.8086  5  bad
chromium@406728  4645.79  36.8742  5  bad
chromium@406739  4660.49  62.4341  5  bad

Bisect job ran on: mac_retina_perf_bisect
Bug ID: 630461

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --also-run-disabled-tests blink_perf.canvas
Test Metric: upload-canvas-2d-to-texture/upload-canvas-2d-to-texture
Relative Change: 16.08%
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_retina_perf_bisect/builds/1432
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9006484861874608336


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5906417874108416

| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Tests>AutoBisect.  Thank you!

Re-assigning to Ken, who will decide how to proceed. This and 630453 are regressions, almost certainly caused by switching to ChannelMojo for GPU channels.

Also merged in  issue 630460 .

Not sure why we're not getting the "merged" messages on this bug, so I filed https://bugs.chromium.org/p/monorail/issues/detail?id=1585

Perf sheriff ping: reminder to follow up on possible performance issues

Ping rockot@, please take a look at this CL.

Ping!

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8998976243790541248

Bisect failed: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus7_perf_bisect/builds/3391
Failure reason: the build has failed.
Additional errors:
The metric was not found in the test output.
Either of the initial "good" or "bad" revisions failed to be tested or built.

If this is almost certainly caused by switching to ChannelMojo for GPU channels, can this be justified initially? Is there a plan to address the performance regressions introduced by making the switch?

Some of the regressions (like +CPU time spent on IO thread) are not
entirely avoidable. We continue to make iterative improvements on them, but
we do expect (and accept) that Mojo will be slightly more expensive than
vanilla IPC.

The one benchmark which stands out here is upload-webgl-to-texture, which
indicates there's some real room for improvement...somewhere. The plan is
loosely to use it as a good profiling case and let that guide future
optimizations. In this case it's just a matter of priority: we have limited
eng resources, and it's not clear to me that this particular regression (a
WebGL texture upload microbenchmark) is an urgent one to address. Of course
if I'm wrong about that, we can shuffle priorities a bit and spend some
more cycles on aggressive optimization.

Merging into a similar performance bug. The most interesting regression here was upload-webgl-to-texture, but that benchmark appears to have disappeared a few months ago.

If there is a more current benchmark which might be reasonably representative of that one, please let me know and I'll add it to the set of things to monitor. The overall tracking bug for Mojo performance optimizations is bug 726489.

The upload-webgl-to-texture benchmark still exists: https://cs.chromium.org/chromium/src/third_party/WebKit/PerformanceTests/Canvas/upload-webgl-to-texture.html

Hmm. When I try to find data for the benchmark on the perf dashboard, it doesn't seem to exist past January 2017. Maybe only some bots are still running it?

It's possible a change in GPU drivers could make a bot stop running it. I definitely see it running on mac retina: https://chromeperf.appspot.com/report?sid=7feb1aa2778c8a0810e058397cdd73887a8b087ec7c4956c86c974fb0e7b1e2c