Disable CA compositor if there are too many quads.

Disable CA compositor if there are too many RenderPassDrawQuads

If there are too many RenderPassDrawQuads, we shouldn't use Core Animation to
present them as individual layers, since that potentially doubles the amount of
work needed to present them. cc has to blit them into an IOSurface, and then
Core Animation has to blit them to the final surface.

BUG=636884
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/e4e1cdc9897a759cf154dad8e81c6dd88ac4b22d
Cr-Commit-Position: refs/heads/master@{#420188}

[erikchen, 2016-09-20 20:17:58]

Description was changed from

==========
Disable CA compositor if there are too many quads.

If there are too many quads, we shouldn't use Core Animation to present them as
individual layers, since both passing them to Core Animation, and the
subsequent blits from the Window Server are too expensive.

BUG=636884
==========

to

==========
Disable CA compositor if there are too many quads.

If there are too many quads, we shouldn't use Core Animation to present them as
individual layers, since both passing them to Core Animation, and the
subsequent blits from the Window Server are too expensive.

BUG=636884
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[erikchen, 2016-09-20 20:24:20]

The CQ bit was checked by erikchen@chromium.org to run a CQ dry run

[erikchen, 2016-09-20 20:24:28]

erikchen@chromium.org changed reviewers:
+ ccameron@chromium.org

[erikchen, 2016-09-20 20:24:29]

ccameron: Please review.

[commit-bot: I haz the power, 2016-09-20 20:25:02]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[ccameron, 2016-09-20 20:27:36]

lgtm (guessing that we're going to see this be hit for <0.01% of frames, which
means we're doing this "just for the benchmark" ... but not everyone has done
time at NV to learn how that game ends ...)

[erikchen, 2016-09-20 20:31:36]

On 2016/09/20 20:27:36, ccameron wrote:
> lgtm (guessing that we're going to see this be hit for <0.01% of frames, which
> means we're doing this "just for the benchmark" ... but not everyone has done
> time at NV to learn how that game ends ...)

actually, I changed my mind. The additional cost for most IOSurfaces is minimal
(either we do the blit, or CA does the blit). The cost is particularly high for
RPDQs, since we blit to an IOSurface, so that CA can blit to the final surface.
One sec...

[erikchen, 2016-09-20 20:46:00]

Description was changed from

==========
Disable CA compositor if there are too many quads.

If there are too many quads, we shouldn't use Core Animation to present them as
individual layers, since both passing them to Core Animation, and the
subsequent blits from the Window Server are too expensive.

BUG=636884
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Disable CA compositor if there are too many RenderPassDrawQuads

If there are too many RenderPassDrawQuads, we shouldn't use Core Animation to
present them as individual layers, since that potentially doubles the amount of
work needed to present them. cc has to blit them into an IOSurface, and then
Core Animation has to blit them to the final surface.

BUG=636884
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[erikchen, 2016-09-20 20:46:27]

The CQ bit was checked by erikchen@chromium.org to run a CQ dry run

[erikchen, 2016-09-20 20:46:37]

On 2016/09/20 20:31:36, erikchen wrote:
> On 2016/09/20 20:27:36, ccameron wrote:
> > lgtm (guessing that we're going to see this be hit for <0.01% of frames,
which
> > means we're doing this "just for the benchmark" ... but not everyone has
done
> > time at NV to learn how that game ends ...)
> 
> actually, I changed my mind. The additional cost for most IOSurfaces is
minimal
> (either we do the blit, or CA does the blit). The cost is particularly high
for
> RPDQs, since we blit to an IOSurface, so that CA can blit to the final
surface.
> One sec...

ccameron: PTAL

[commit-bot: I haz the power, 2016-09-20 20:46:58]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-20 22:12:11]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-09-20 22:12:12]

Dry run: This issue passed the CQ dry run.

[erikchen, 2016-09-20 22:25:18]

erikchen@chromium.org changed reviewers:
+ asvitkine@chromium.org

[erikchen, 2016-09-20 22:25:18]

asvitkine: Please review tools/metric/histograms.

[ccameron, 2016-09-20 22:35:55]

still lgtm

[erikchen, 2016-09-21 20:06:04]

erikchen@chromium.org changed reviewers:
+ isherman@chromium.org

[erikchen, 2016-09-21 20:06:05]

isherman: Please review tools/metrics

[Ilya Sherman, 2016-09-21 21:12:09]

histograms.xml lgtm

[erikchen, 2016-09-21 21:56:08]

The CQ bit was checked by erikchen@chromium.org

[commit-bot: I haz the power, 2016-09-21 21:56:39]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-21 22:59:45]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-09-21 23:04:08]

Description was changed from

==========
Disable CA compositor if there are too many RenderPassDrawQuads

If there are too many RenderPassDrawQuads, we shouldn't use Core Animation to
present them as individual layers, since that potentially doubles the amount of
work needed to present them. cc has to blit them into an IOSurface, and then
Core Animation has to blit them to the final surface.

BUG=636884
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Disable CA compositor if there are too many RenderPassDrawQuads

If there are too many RenderPassDrawQuads, we shouldn't use Core Animation to
present them as individual layers, since that potentially doubles the amount of
work needed to present them. cc has to blit them into an IOSurface, and then
Core Animation has to blit them to the final surface.

BUG=636884
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/e4e1cdc9897a759cf154dad8e81c6dd88ac4b22d
Cr-Commit-Position: refs/heads/master@{#420188}
==========

[commit-bot: I haz the power, 2016-09-21 23:04:09]

Patchset 3 (id:??) landed as
https://crrev.com/e4e1cdc9897a759cf154dad8e81c6dd88ac4b22d
Cr-Commit-Position: refs/heads/master@{#420188}